Web: http: // www. pearl - hifi . com 86008, 2106 33 Ave.

SW, Calgary, AB; CAN T2T 1Z6

E-mail: custserv @ pearl - hifi . com Ph: + .1.403.244.4434 Fx: + .1.403.245.4456
Inc.
Perkins Electro-Acoustic Research Lab, Inc. ❦ Engineering and Intuition Serving the Soul of Music

Please note that the links in the PEARL logotype above are “live”
and can be used to direct your web browser to our site or to
open an e-mail message window addressed to ourselves.

To view our item listings on eBay, click here.

To see the feedback we have left for our customers, click here.

This document has been prepared as a public service . Any and all trademarks
and logotypes used herein are the property of their owners.

It is our intent to provide this document in accordance with the stipulations with
respect to “fair use” as delineated in Copyrights - Chapter 1: Subject Matter and
Scope of Copyright; Sec. 107. Limitations on exclusive rights: Fair Use.

Public access to copy of this document is provided on the website of Cornell Law School
at http://www4.law.cornell.edu/uscode/17/107.html and is here reproduced below:

Sec. 107. - Limitations on exclusive rights: Fair Use

Notwithstanding the provisions of sections 106 and 106A, the fair use of a copyrighted work, includ-
ing such use by reproduction in copies or phono records or by any other means specified by that section,
for purposes such as criticism, comment, news reporting, teaching (including multiple copies for class-
room use), scholarship, or research, is not an infringement of copyright. In determining whether the use
made of a work in any particular case is a fair use the factors to be considered shall include:

1 - the purpose and character of the use, including whether such use is of a
commercial nature or is for nonprofit educational purposes;

2 - the nature of the copyrighted work;

3 - the amount and substantiality of the portion used in relation to the copy-
righted work as a whole; and

4 - the effect of the use upon the potential market for or value of the copy-
righted work.

The fact that a work is unpublished shall not itself bar a finding of fair use if such finding is made
upon consideration of all the above factors

♦ PDF Cover Page ♦

♦ Verso Filler Page ♦
 PATENT SPECD'ICATION
- - I
Application Date: June 5, 1937. No. 15620/37.
Complete Specification Left: March 4. 1938.
Complete Specificat{on Ar;c;epted: Dec;, S, 1938. 496.883
PROVISION AL SPECIFICATION
Improvements in or relatnng te Thermionic Valve Amplifying
Circuits
I, ALAN DOWER BLUMLEIN, of 37, The former, or alternatively, the screening
Ridings, Ealing, London, "V.5, a British grid may be connected to a source of
subject, do hereby declare the nature of potential through a resistance and con- 55
this invention to be as follows:- nected through a decoupling condenser to
Ii This invention relates to thermionic a tapping point on the primary winding
valve amplifying circuits, and is pa.rticu- of the output transformer. Other alter-
larly concerned with power amplifying l'.ative connections for causing the poten-
circuits. tial of the screening grid to vary in pro- 60
It is usual in such eircuits to employ portion to the potential of the anode may
10 as the last valve of the stage a pentode be employed. For example, the screening
valve or a power tetrode. These valves grid may be connected to a tapping point
are provided with screening grids which in a choke connecting the anode of the
are usually maintained at a constant posi- valve to the source of anode current. 65
tive potential with respect to the cathode. The effect of causing the screen to vary
15 Valves of this. kind give a greater output in potential proportional to the variations
for a given anode potential than triodes, in anode potential is to cause the valve to
but they suffer from the disadvantage that function somewhat after the manner of a
their characteristic impedance to the triode. It is known that in a triode the 70
anode is very much greater than the ratio anode impedance is less than the ratio of
20 of the 'Steady D.C. voltage to the steady the steady D.C. anode voltage to the
D.C. current. Consequently, in order to steady D.C. anode current. Hence, by
obtain a good power output the imped- choosin~ a suitable tapping point ror the
ance of the load must be very much less screening grid the t!haracteristic imped- 70
than the characteristic impedance of the ance of the anode and screen effectively
25 valve. Furthermore, particularly in the combined can be given any dE'sired value
case of either a pentode or a power tetrode between the impedance of a pentode or
valve, which is coupled to its load through tetrode and the impedance of a triode. A
a transformer, it is known that if the load suitable tapping point, in practice, may 80
is removed then for relatively moderate be found to be such that the voltage swing
80 variations in potential of the 'input grid, on the screen is between about a quarter
excessively high output voltages will be and a half the voltage swing on the anode.
produced which may be sufficient to cause For the purpose of describing the in-
disruption of the valve or of the output vention more in detail reference will be 85
transformer. made to the accompanying diagrammatic
85 It is the chief object of the present in- drawings in which:
vention to provide an improved circuit in Fig. 1 illustrates the application of the
which the characteristic impedance of invention to a circuit embodying a pen-
valves of the kind referred to can be more tode valve, and 90
nearly matched to their optimum loads Fig. 2 illustrates the application of the
40 and in which the danger of an excessive invention to a tetrode valve.
increase of volta~'e on removal of the load As shown in Fig. 1, signals to be am-
may be substantlally avoided. plified are applied between the control
According to the invention a thermionic grid 3 and cathode 4 of a pentode valve 95
valve amplifying circuit is provided em- thro'ugh a coupling condenser 5 and leak
45 ploying a pentrode or tetrode valve in resistance 6, the cathode of the valve
which the screening grid of the valve is being biassed by a dropper resistance 7
so connected that the potential thereof shunted by a by-pass condenser 8. The
varies in proportion to the A.C. potential suppressor grid 9 is connetlted to the 100
variations of the anode or the valve. This cathode in the usual manner. The anode
50 result may be atlhievec1 by connecting the 10 is connected through the primary wind-
screening grjd to a tapping point on the ing 11 of an output transformer to the
primary winding of an output trans- positive terminal of a suitable source of
-
C)
496,883

high tension current the secondary wind- the primary winding 22, it may be con-
ing 12 of the tra;nsformer being con- nected to a suitable tapping point on the
nected to a load not shown. The screen- choke 20. ]orany other alternative con·
ing grid 13 is connected as shown to a nections are possible. 40
5 tapping point on the primary winding' 11 The circuits described also llave the
so that the screening grid 13 varies in advantage that if tIle loads are remowd
potential in proportion to the potential the output voltage obtained for relatiwly
variations on the anode 10. As stated 8mall potential variations of the control
above, the tapping point may be arranged grids is not as great as with the arrange- 45
10 so that the voltage swing on the screen ments usuall~' employed, since the screen-
grid 13 is between about a quarter and a ing grid which is partially effective in
haH the voltage swing on the anode 10. controlling the cathode current will swing'
The arrangement shown in Fig. 2 of in opposite phase to the grid and so pre-
the drawings illustrates the invention as vent or reduce an excessiw output voltage. 50
15 applied to a tetrode valve and, in this In some cases the decoupling condensers
case, signals to be amplified are applied 8 and 18 of Figs. 1 and 2 respectivel~- mn~­
betweell. the control grid 14 and cathode 15 not be employed, in which cases negath-e
through an input transformer 16 the feedback will occur tending to reduce the
cathode 15 being biassed by a dropper second harmonic output. The effed of 55
20 resistance 17 shunted by a by-pass con- such feedback will, however.. be to in-
denser 18. The anode 19 is connected to crease the impedance of the valves so that
the positive terminal of a source of anode in cases where it is desired to reduce the
current through the choke 20 and is con- second harmonic output it should be
nected through a; by-pass condenser 21 to arranged that the potential variations of 60
25 the primary winding 22 of an output the screen should be greater compared
transformer the secondary winding 23 of with the cases in which the decoupling
which is connected to a load, not shown. condensers are employed in order to main-
In this example of the invention the tain the anode impedance at the required
screening grid 24 is connected through a value. In each example described it will 65
30 resistance 25 to the source of high tension also be appreciated that the current which
current and is also connected in the man- flows in the screening grid also contrilmtes
ner shown through a decoupling condenser to the useful output of the circuit.
~6 to a tapping point on the primary wind-
mg 22 of the output transformer. If Dated this 4th da~- of June, 1937.
35 desired, in Fig. 2, instead of connecting F. 'V. CACKETT,
the screen grid 24 to a tapping point on Chartered Patent Agent.

COMPLETE SPECIFICATION

Improvements in or relating to Thermionic Valve Amplifying

Circuits
I, ALAN DOWER BLUMLEIN, of 37, rrhe than the ratio of the steadv D.C. voltag'e
70 Ridings, Ealing, London, W.5, a British to the steady D.C. current. Consequentl,\-,
L

Subject, do hereby declare the nature of ~n order to obtain 0. good power output the
this invention, and in what manner the Impedance of the load must he very much
same is to be performed, to be particularly less than the internal impedar.ce of the 95
described and ascertained in and by the valve. Furthermore, particularlY in the
75 following statement:- . case of either a. pentode or a. power tetrode
This invention relates to thermionic valve, which is coupled to its load through
valve amplifying circuits, and is p~rti~u­ a transformer, it is known that if the load
larly concerned with power amphfymg is removed then for relatively moderate 100
circuits. variations in potential of the input grid,
.80 It is usual in such circuits to employ excessively high output voltag'es will he
as the last valve of the stage a pentode produced which may be sufficient to cause
valve or a power tetrode. These disruption of the valve or of the output
valves are provided with screening elec- transformer. 105
trodes or grids which are usual1y main- It is the object' of the present invention
8f1 tained at a constant. positive potential to provide an improwd circuit in which
with respect to the cathode. Valves of the internal impedance of valves of the
this kind give a greater output for a given kind referred to can be more nearly
anode potential than triodes, but they
suffer from the disadvantage that. their
matched to their optimum loads and
which the danger of an excessive increase
in 110
90 internal impedance is very IDl1Ch greater of voltage on removal of the load may be
 496,888 3

sUbstantially avoided. that in a triode the anode impedance" is

According to one feature of the inven- less than the ratio oJ tbie steady D.O.
tion a thermionic valve amplifying tir- anode voltage to the steady D.O. anode
cuit is provided comprising a valve having current. Hence, by choosing a suitable
I; at least a cathode, control electrode, an tapping point for the screening electrode 70
anode and ru screening electrode adapted the internal impedance of the anode and
to form an electrostatic screen between the screening electrode effectively combined
control electrode and anode, and a rela- can be given any desired value between
tively low impedance load connected to the impedance of a pentode or a tetrode
1() said valve, wherein instead of maintain- and the impedanCe of a triode. A"suitable 76
ing said screening electrode at the usual tapping point, in practice, may be found
constant positive potential less than the to be such that the voltage swing on the
anode potential which would cause said screening electrode is between about a
valve to have a high impedance, it is so quarter and a half the voltage swing on
15 connec.ted to the anode circuit of said: ihe anode. 80
valve that in operation the potential of In order that the invention may be
said screening electrode will vary in clearly understood and readily carried
phase) but at a lower ~amplitude with. into effect it will now be more fully
respect to the potential variations OT said described with reference to the drawings'
20 anode so as to reduce the impedance of accompanying the Provisional Specifica- Sf)
the valve· whereby it is more nearly tion in which:-
matched to its low impedance load. Fig. 1 illustrates the application of the
According to another feature of the invention to a circuit embodying a pen-
invention a thermionic valve amplifying tode valve, and
25 circuit is provided comprising a valve Fig. 2 illustrates the application of the 90
having at least a cathode, control elec- invention to a tetrode valve.
trode, an anode and a screening electrode As shown in Fig, 1, signals to be
adapted to form an electrostatic screen amplified are applied between the control
between the control electrode and anode, g'rid 8 and cathode 4 of a pentode valve
80 and a relatively low impedance load con- through a coupling condenser 5 and leak 95
nected to said valve, wherein instead of resistance 6, the cathode of the valve
maintaining said screening electrode at being biassed by a dropper resistance 7
the usual constant positive potential less shunted by a decoupling condenser 8.
than the anode potential which would ':Phe suppressor grid. 9 is connected to the
35 ca"use said valve to have a high imped- cathode in the usual ma.nner. The anode ] 00
ance, it is so connected with the output 10 is connected through the primary
circuit of said valve that in operation the "'-inding 11. of an output transformer to
potential of said screening electrode will the positive terminal of a suitable source
vary in phase) but at a lower amplitude of high tension (jurrent the secondary
40 with respect to the potential variations of winding 12 of the transformer being con- 105
said anode so as to reduce the impedance nected to a load not shown. rrhe SCl'een-
of the valve whereby it is more nearly ing grid 18 is connected a'l shown to a
matched to its low impedance load. The tapping point on the primary winding 11
desired result may be achieved by con- so that the screening grid 13 varies in
45 necting the screening electrode to a tap- potential in direct propOl,tion to or in 110
ping point On the primary winding of an phase with rthe potential 'Variations on
output transformer, or alternatively, the the anode 10. As stated !1bove, the tap-
screening electrode may be connected to a ping point may be arranged so that the
source of potential through a resistance voltage swin!!' on the screen grid 13 is
50 and connected through a by-pass con- between about a quarter and a haH the 115
denser to a tapping point on the primary voltage swing on the anode 10.
winding of the output transformer. Other The arrangement shown in Fig. 2 of
alternative connections for causing the the drawings illustrates the invention as
potential of the screening electrode to vary applied to a tetrode va.lve and, in this
55 in proportion to the potential of the anode case, signals to be amplified are applied l~O
may be employed. For example, the between the control grid 14 and cathode .
screening electrode may be connected to 15 through an input transformer 16 the
a tapping point in a choke connecting the cathode 15 being biassed by a dropper
anode of the valve to the source of anode resistance 17 shunted by a decoupling
60 current. condenser 18. The anode 19 is connected 125
The effect of Musing the screening elec- to the positive terminal of a source or
tf{)d e to vary in potential proportional to anode current through a choke 20 and is
the variations in anode potential is to connected through a by-pass condenser 21
cause the -valve .to .function somewhat to the primary winding 22. of an output
65 after the manner of a triode. It is known transformer the secondary winding 23 of 180
 4 ·W(l.S~3

which is connected to a load, not shown. purpose above-mentioned and the iact
In this example of the invention the that the screening properties of the screen-
screening grid 24 is connected through a ing electrode are reduced is relatively im-
resistance 25 to the source of high tension material particularly in low frequency
5 eurrent and is also connected in the man- amplifiers. 70
ner shown through a by-pass condenser 26 Having now particularly described and
to a tapping point on the primary wind- ascertained the nature of my said inven-
ing 22 of the output transformer. H tion, and in what manner the same is to
desired, in Fig. 2, instead of connecting be performed, I declare that what I claim
10 the screen grid 24 to a tapping point on is:- 75
the primary winding 22, it maj: be con- 1. A thermionic valve amplifying cir-
nected to a suitable tapping point on the cuit. comprising a valve having at least a
choke 20. Many other alternative con- cathode,· control electrode, an anode and
nections are possible. a screening electrode adapted to form an
15 The cir~uits described also have the electrostatic screen between the control 80
advantage that if the loads are removed electrode and anode.. and a relatively low
the output voltage obtained for relatively impedance load connected to said valve,
small potential variations of the control wherein instead of maintaining said
- grids is not as great as with the arrange- screening electrode at the usual constant
20 ments usually employed, since the screen- positive potential less than the anode 85
ing grid which is partially effective in potential which would cause said valY£' to
controlling the cathode current will swing have a high impedance, it is so connected
in opposite phase to the grid and so pre- to the anode circuit of said valve that in
vent or reduee an excessive output voltage. operation the potential of said sl'reening
25 In some cases the decoupling condensers eleetrode will vary in phase, but at a lower 90
8 and 18 of Figs. 1 and 2 respectively amplitude with respect to the potential
may not be employed, in which cases variations of said anode so as to reduce the
neg'ative feedback will occur tending to impedance of the valve whereby it is more
reduce the second harmonic output. The nearly matched to its low impedance load.
80 effect of such feedback will, 11Owever, be 2. A thermionic valve amplifying cir- 95
to increase the impedance of the valves euit comprising a valve having at least a
so that in cases where it is desired to cathode, control electrode, an anode and
reduce the second harmonic output it a. screening electrode adapted to form an
should be arranged that the potential electrostatic screen between the control
:15 variations of the screen should be greater electrode and anode, and a relativel~' low 100
compared with the cases in which t.he; impedance load connected to said valve,
decoupling condensers are employed in wherein instead of maintaining said
order to maintain the anode impedance at screening electrode at the usual constant
the required value. In each example positive potential less than the anode
40 described it will also be appreciated that potential which would cause said valve to 105
the current which flows in the screening have a high impedance, it is so connected
grid also contributes to the useful output with the output circuit of said valve that
of the circuit. in operation the potential of said screen-
With the arrangements described the ing electrode will vary in phase, but at n
45 internal impedance o-f the valves is reduced lower amplitude with respect to the 110
thus enabling the valves to be matched potential variations of said anode so as
more nearly to their low impedance loaas to reduce the impedance of the valve
such as are normally employed with power whereby it is more nearly matched to its
output valves. The effect of causing the low impedance load. .
50 screening electrode 13 (Fig. 1) or 24 (Fig. 3. A thermionic valve amplifying cir- 115
2) to vary in phase with the potential cuit according to claim 1 or 2, wherein
variations of the anodes of their respective the screening electrode is connected to a
valves sacrifices however some of the tapping point on an impedance connected
screening properties of the electrodes 13 between the anode of the valve and the
65 and 24 compared with the case where source of anode current. 120
these electrodes are maintained at con- ~. A the.rmionic :alve amplifying eir-
stant potentials. Sbreening action does CUlt accordIng to clalm 1 or 2, wherein the
however to some extent occur since the screening electrode is connected through a
potential differences between the screen- resistance to a source of potential and
60 ing electrode and the control electrode are through a b:r-pass condenser to a tapping 125
less than the potential differences between point on an impedance, across which the
the anode and control electrode. By voltage variations occurring at the anode
causing the potential of the screening of the valve are set up.
electrode to vary as above described the 5. A thermionic valve amplifying cir-
65 impedance of the valve is reduced for the cuit substantially as described herein with 130
 496,883

reference to Figure 1 or 2 or the drawings Dated this 3rd day of March, 1938.
accompanying the Provisional Speci£.ca~ F. W. CACKETT,
tion. Chartered Patent Agent.
TJeamingt{)n Spa: Printed for His Majesty's Stationery Office, by the Courier Press.-1939.
496883 PROVISIONAL SPECIFICATION i SHEET

H.T.+

Fig. I.

_1-----4~--o ItT. +

1b

:JI
Fig. z.
Malby &- Sons, Photo -lIth,.
 UNITED STATES PATENT OFFICE
2,218,902
THERMIONIC VALVE AMPLIFYING ClRCUITS
Alan Dower Blumlein, London, England, assignor
to Electric & Musical Industries Limited, Hayes,
Middlesex, England, a corporation of Great
2,218,902 Britain
Patented Oct. 22, 1940 Application Ma.y 18, 1938, Seria.l No. 208,619
In Grea.t Britain June 5, 193'1
4 Claims. (CI.l79-171)
This invention relates to thermionic valve am- that in a triode the anode impedance is less tha·n
plifying circuits and is particularly concerned the ratio of the steady D. C. anode voltage to the
with power amplifYing circuits. steady D. C.' anode current. Hence, by choosing
It. is usual in such circUits to employ as the last a suitable tapping point for the screening grid
5 valve of the stage a pento.de valve or a power the characteristic impedance of the anode and -5
tetrode. These valves are provided with Screen- screen effectively combined can be given any de-
ing grids which are usually maintained at a con- sired value between the impedance of a pentode
stant positive potential with respect to. the cath- or tetrode and the impedance of a triode. A
ode. Valves of this kind give a greater output for sUitable tapping POint, in practice, may be found
10 a given anode potential than triodes, but they to be such that the voltage swing on the screen lO
suffer from the disadvantage that their char- is·between about a quarter-and a-half the voltage
acteristic impedance to the anode is very much swing-on the anode.
greater than the ratio. of the steady D. C. voltage In order that the invention may be clearly un-
to the steady D. C. current., 'Consequently, in derstood and readily carried into effect it will
\5: order to obtain a good power output the imped- now be more fully described with reference to l5
ance of the load must be very much less than the the accompanying drawing in which:
characteristic ,impedance of the valve. Further- Fig. 1 illustrates the application of the inyen-
more, particularly in the case of either a pentode tion to a circUit embodying a pentode valve; and,
or a power tetrode valve, which is coupled to its Fig. 2 illustrates the application of the inven-
20, load through a transformer,- it -is known that if tion to a tetrode valve. 20
the load is rem'oved, then for relatively moderate As shown in Fig. 1, signals to be amplified are
variations in potential of the input grid,exces- applied between the control grid 3 and cathode 4
sively high output voltages will be produced which of a pentode valve through a coupling condenser
may be suflicient:to cause disruption of the valve 5 and leak resistance 6, the cathode of the valve
25 or of the output transformer. being biased by a dropper resistance 1 shunted by 25
It is the chief object of the'present invention a decoupling condenser B. The suppressor grid
to provide an improved circuit in which the char- 9 is connected to the cathode in the usual manner.
acteristic impedance' of valves of, the 'kind re- The anode (0 is connected through the primary
ferred to can' he more nearly' matched to their winding II of an output transformer to the posi-
~O, optimum· loads and in which the danger of an tive terminal of a suitable source of high tension 30
excessive increase of voltage on removal of the current the secondary winding 12 of the trans-
load may be substantially avoided. ·former being. connected to a load not shown.
According to the invention a thermionic valve The screening grid 13 is connected as shown to a
amplifying circuit is provided employing a pen- tapping point on the primary winding (t - so that
:Ji) tode or tetrode valve in which the screening grid the screening grid (3 varies in potential in pro- 35
of the valve is so connected that the potential portion to the potential variations on the anode
thereof varies in proportion to the A. C. potential I D. As stated above, the tapping point may be
variations of the anode of the valve. This re- arranged so that the voltage swing on the screen
sult may be achieved by connecting the screening grid 13 is between about a quarter and a half the
.10 grid to a tapping point on the primary winding, voltage swing on the anode 10. . 40
of an output transformer, or alternatively, the The arrangement shown in Fig. 2 of the draw-
screening grid may be connected to a source of ing illustrates the- invention as applied to a tet-
potential through a resistance and connected rode valve and, in this case, signals to be ampli-
through a by-pass condenser to a tapping point on fied are applied between the control grid 14 and
'1:3 the primary winding of the output transformer. cathode (5 through an input transformer 16 the 45
Other alternative connections for causing the po- cathode (5 being biased by a dropper resistance
tential of the screening grid to vary in proportion 11 shunted by a decoupling condenser lB. ' The
to the potential of the anode may be employed. anode 19 is connected to the positive terminal of
For example, the screening grid may be connected a source of anode current through a choke 20 and
:it! to a tapping point in a choke connecting the an- is connected through a by-pass condenser 21 to 50
ode of the valve to the source of anode current. the primary winding 22 of an output transformer
The effect of causing the screen to vary in po- the secondary winding 23 of which is connected
tential proportional to the variations in anode to a load, not shown.
potential is to cause the valve to function some- In this example of the invention the screening
5,) what after the manner of a triode. It is known grid 24 is connected through a resistance 25 to 55
 2 2,218,902
the source of high tension current and is also 2. In an amplifying circuit, a thermionic valve
connected in the manner shown through a by- having an anode, a cathode, a grid electrode and
paBS condenser 26 to a tapping point on the pri- a. screening electrode, an input circuit for said
mary winding 22 of the output transformer. If valve connected between the cathode and grid
5 desired, in Fig. 2, instead of connecting the screen electrode thereof, said input circuit including 5
grid 24 to a tapping point on the primary wind- means for connecting to a source of signal volt-
ing 22, it may be connected to a suitable tap- age, an output circuit for the valve including a
ping point on the choke 20. Many other alterna- load impedance across which is developed the
tive connections are possible. output voltage, and means for connecting the
10 The circuits described also- have the advantage screening electrode to an intermediate point of 10
that if the loads are removed the output volt- said load impedance device to thereby produce
age obtained for relatively small potential varia- a voltage swing on the screening electrode whioh
tions of the control grids is not as great as with is an amplitude reduced image of the voltage
the arrangements usually employed, since the SWing produced on the anode due to signal energy
15 screening grid which is partially effective in con- impressed upon the input circuit, the load im- 15
trolling the cathode current will swing in oppo- pedance comprising a transformer having a pri-
site phase to the grid and so prevent or reduce mary winding connected in the output circuit of
an excessive output voltage. the valve and a secondary winding adapted to
In some cases the decoupling condensers 8 and he connected to a utilizing device, said screening
20 .It of Figs. 1 and 2 respectively may not be em- electrode being connected to a point of said pri- 20
ployed, in which cases negative feedback will mary winding.
occur tending to reduce the second harmonic 3. In an amplifying circmt a thermionic tUbe
output. The effect of such feedback will, how- having an anode, a cathode, a grid electrode and
ever, be to increase the impedance of the valves an auxiliary electrode, an input circuit for said
25 so that in cases where it is desired to reduce the tube connected between the grid electrode and 25
second harmonic output it should be arranged the cathode thereof and including means for con-
that the potential variations of the screen should nection to a source of signal voltage, a source of
be greater compared with the cases in whi-ch the anode potential for said tube including a high
decoupling condensers are employed in order to tension terminal, a connection including a choke
30 maintain the anode impedance at the required coil between the anode of the tube and said high 30
value. In each example described it will also be tension terminal, an output circuit for said tube
appreciated that the current which flows in the including a load impedance connected between
screening grid also contributes to the useful out- the anode and cathode of the tube, means in-
put of the circuit. cluding an impedance device for connecting the
35 r claim: auxiliary electrode to said high tension tenninal 35
1. In an amplifying circuit, a thermionic valve and a by-pass condenser co-nnected between said
having an anode, a cathode, a grid electrode and auxiliary electrode and an intermediate point
a screening electrode, an input circuit for said of said load impedance whereby there is im-
valve connected between the cathode and grid pressed upon the auXiliary electrode an ampli-
40 electrode thereof, said input circuit including tude reduced image of the voltage swing produced 40
means for connecting to a source of signal volt- on the anode due to signal energy impressed upon
age, an output circuit for the valve including a the input circuit.
load impedance across which is developed the 4. An arrangement as described in the next
output voltage, and means for connecting the preceding claim chal'acteIized by that the out-
4;3 screening electrode to an intermediate point of put impedance comprises a transformer having 45
said load impedance device to thereby produce a a primary winding connected in the output cir-
voltage swing on the screening electrode which cuit of the tube and a secondary winding pro-
is an amplitude reduced image of the voltage vided with means for connection to a utilizing
swing produced on the anode due to signal energy device, said auxiliary electrode being connected
[i0 impressed upon the input circuit, the point of through the by-pass condenser to an intermediate 50
connection of the screening electrode to the load point of said primary winding.
impedance being so chosen that the voltage swing
on the screen is of the order of Va the voltage ALAN DOWER BLUMLEIN.
swing on the anode.
Oct. 22, 1940. A. O. BLUMLEIN 2,218,902
THERVIDHIC VALVS AMPLIFYING CIRCUITS
Filed Kay 18. 1938

H.T.+

H.T.+

23
16

j
INVENTOR
ALAN DOWER BlUMLEIN
BY~~~
·ATTORNEY t
♦ Verso Filler Page ♦
 UNITED STATES PATENT OFFICE
2,680.218
AUDIO TRANSFORMER
Herbert I. Keroes, PhDadelphia, Pa., assignor to
Aero Products Company. Philadelphia, Pa.
Application October 26, 1950, Serial No. 192,265
9 Claims. (01. 317-220)
1 2
The present invention relates generally to widely different output impedances, while super-
push-pull audio transformers, and more particu- ficially simple, is not in fact feasible in wide band,
larly to systems of relatively arranging the wind- high quality transformers, i. e., transformers
ings of audio transformers, to enable change of capable of transferring a wide band of audio fre-
·output impedance without change in per- 5 quencies without appreciable phase or amplitude
fermance. distortion, since in fact, those desiderata of a
The conventional push-pull aUdio transformer high quality transformer referred to hereinabove,
comprises three windings, two of which are con~ if present when the transformer load is con-
nected in balanced relation, and which are sub- nected to one of the taps, will not be present for
stantially identical in respect to impedance. HI any other of the taps,
These windings are mounted on a magnetic core, In accordance with the present invention, a
on which is also mounted a secondary winciL'1g, push-pull audio transformer capable of provid-
thereby coupled to the primary windings. The ing \"vide band response without appreciable dis-
primary windings are normally coupled in push- tortion, and yet capable of matching selectively
pull relation to a pair of push-pull audio amplifier 15 loads of widely varying impedance, 1s provided,
tubes, and the secondary winding may be con- by means of a novel system of secondary winding
nected either in balanced or unbalanced fashion, arrangement. More specifically, the primary
i. e., with one end grounded. windings are arranged as four concentric sections,
The high frequency response of an audio trans- two of which in series provide one of the primary
former is limited by leakage reactance between 20 halves, and the remaining two of which in series
the various windings, as well as by the dis- provide the other of the primary halves. The
tributed capacity across the primary winding. secondary winding is composed of three axially
At low power level of output, high frequency concentric sections, two of which are of equal
response is limited primarily by the leakage re- numbers of turns. The latter sections are con-
actance between primary and secondary windillgs. 25 nected in parallel, and interleaved between the
At high power levels high frequency response is primary sections taken in pairs, i. e., one between
further limited by leakage reactance between pri- the two innermost primary sections, the other be-
mary windings, because that leakage reactance tween the remaining two. The parallel connected
rele,tively decouples the audio amplifier tubes. sections provide a low impedance secondary
Audio output transformers may be employed 30 winding which is symmetrically arranged with
in amplifiers of the negative feed-back type, in respect to the primary windings, and evenly dis-
which voltages derived from the audio output tributed thereamong, and which therefore is
transformer are applied to an input circuit of electrically in balance with respect to the prL'11ary
the amplifier. In such amplifiers, the audio out- winding, i. e., possesses equal leakage reactance
put transformer may be designed to introduce :~;:; with respect to each half of the primary.
substantjally zero phase shift, which requires that Capacitive coupling from the secondary to each
leakage reactance be minimized, and that tht. primary half is approximately the same, and
transformer operate in truly balanced fashion. since voltage on one side Of the section is of
i. e., without longitudinal effects, without spurious opposite phase to that on the other side, capaci-
Signal transmission, whether through intel'-wind- '10 tive coupling is effectively balanced out.
ing capacity, unequal coupling between primary The remaining section of the secondary wind-
windings and the secondary winding, or other- ing is formed of two separate windings, parallel
wise. wound, and connected to form a "re-entrant"
A problem arises, in commercial transformers, series connection, the term "re-entrant" mean~
because these transformers may be required to Ill> ing parallel wound mutually insulated conduc-
drive speakers of various impedance, and to Pl'O- tors, each conductor connected in series with the
vide an impedance match. In particular, the succeeding winding by connecting its termination
output impedance of the transformer may be to the initial point of the succeeding winding.
required to equal 4, 8 or 16 ohms, selectively. in One of these windings is connected in series with
actual transformer designs. This requires pro- 50 the low impedance secondary to provide an in-
vision for selective variation in impedance of th:;. termediate impedance secondary. Both windings
secondary winding, without appreciable Chal1g~ taken in series, and in series with the low im-
in transformer performance at the various im- pedance winding, then provides a high impedance
pedances. The solution of providing taps on a secondary. The secondal'Y windings so formed
oonventional secondary win(Ung. to ~staQlish 55 each remains symmetrical with respect to the
Patented June 1, 1954 2,680,218
 2,680,218
3 4
primary winding or sections. Moreover, the wind- other primary half. In the embodiment illus-
ing layers of which the various secondaries are trated in Figure 3, sections Pl and P:! are con-
formed each extend over the entire length of nected in series, and Pa and P4 in series to pro-
each primary section, which reduces leakage vide the pl'!mary halves. In each case the sec-
reactance to a minimum. f) tions must be so arranged that their mutual in-
By virtue of the constl'uction above briefly ex- ductance adds to the series inductance of the
plained, leakage reactance as seen from the pri- separate sections, to establish maximum induct-
mary of the transformer is equal for all values ance in each primary half for the amount of
of secondary impedance, thereby providing copper and iron utilized.
equivalent frequency response for all values of 10 The secondary section Sl, interposed between
secondary impedance. L.ongitudinal current flow primary sections Pt and P2, is connected in paral-
is minimized, for any value of secondary im- lel with secondary section 84, interposed be-
pedance, and whether the secondary is operated tween primary sections P3 and P4. Secondary
balanced 01' unbalanced. Leakage reactance be- sections Sl and 84 are duplicates, in respect to
tween each primary ha.lf and the secondary is J:l number of turns, and are each evenly distributed
uniform, for all values of secondary impedance, with respect to the primary sections, and co-ex-
thereby equally loading both primary halves. tensive therewith. Thereby leakage reactance
The secondary copper loss is uniform, and the is reduced to a minimum, and such leakage react-
insertion loss of the transformer the same, for ance as is present is balanced with respect to
all values of secondary impedance. 20 the primary halves. The residual capacitive cou-
It is, accordingly, an object of the invention to pling from the secondary sections 81 and 84 to
provide a novel and improved wide-band high the primary sections is balanced, and since the
quality audio transformer. potential on each side of the secondary section
It is a further object of the invention to pro- is of opposite phase to that on the other side,
vide a push-pull audio transformer capable of 25 capacitive coupling is essentially balallced out.
providing selectively a l'ange of output imped- Such capacitive currents as do exist may readily
ances, while maintaining the operating char- flow to ground through the secondary windings.
acteristics of the transformer unchanged over The center section of the secondary is com-
the range. posed of two conductors or layers 82, S3, parallel
Another object of the invention resides in the 30 01' bi-filarly wound, forming a Single layer, and
provision of a novel secondary winding for a connected to form a re-entrant series connec-
push-pull audio transformer. tion, Additional taps may be provided by simi-
A further object of the invention resides in larly providing and connecting more than two
the provision of a tapped secondary in a wide- conductors, The conductor S2 is then c011-'lected
band push-pull audio transformer, which pro- 35 ill series with the sections 81 and 84, taken in
vides equal performance at each of the taps, by parallel, to provide an intel'mediate impedance
suitable distribution arrangement and intercon- secondary tap. By winding the conductor 01'
nection of the windings of the secondary. layer S2 with a reduced number of turns, as com-
The above and stilI further objects, features and pared with sections Sl or S4, the total impedance
advantages of the invention will become appar- 40 Dlay be arranged to be 8ft The conductor or
ent upon consideration of the following detailed layer S3 taken in series with &l, and in series
description of speCific embodiments thereof, es- with 81 and S4 taken in parallel, then establishes
pecially when taken in conjunction with the ac- the high impedance secondary winding.
companying drawings, wherein: Whether the secondary winding at its inter-
Figure 1 is a view in cross longitudinal section '~5 mediate Or high impedance tap be conSidered,
taken through the transformer of the invention, the secondary is evenly distributed with respect
and illustrating the physical arrangement of the to the primary, and symmetrical with respect to
sections; the primary halves.
Figure 2 is a schematic wiring diagram show- It follows, by virtue of the construction de-
ing one possible arrangement of the primary sec- 50 scribed ~"ld illustrated, that leakage reactance
tions of the transformer; I'eferred to the primary side of the transformer
Figure 3 is a schematic wiring diagram of a is the same for any secondary tap, whereby equiv-
variation of the arrangement of Figure 2; and, alent frequency response is obtained on any tap.
Figure 4 is a schematic wiring diagram of the Longitudinal transmission paths of both the
secondary v.inding of the transformer. 55 capacitive and inductive types are substantially
Referring now more especially to the drawings, eliminated, as between any tapped section of the
Figure 1 illustrates generally an audio trans- secondary and the primary, and whether the
former having four primary sections, Pl, PlI, PJ secondary be connected to a load in balanced or
and p(, capable of interconnection to provide unbalanced relation. Leakage reactan\)e between
a push-pull or balanced primary winding. In- 60 the secondary and each pl'imary half is the same,
terleaved with the primary sections are four sec- providing symmetrical loading of the Pl'im8,l'Y
ondary coils 81, S2, S3, S4, which are intercon- halves, for any secondary tap. The insertion
nected to form a secondary winding, balanced or loss of the transformer is the same for any tapped
unbalanced, and provided with taps to estab- secondary connection.
lish impedances as seen from the secondary of 65 In a specific construction embodying secondary
4, 8 and 16 ohms, respectively. It is, of course, sections each of which is composed of a single
to be understood that the speciflc values recited layer, the wire size may be the same for every
are by way of example only, and are not to be section of the secondary, and the total turns on
taken as limiting, layers S2 and S3 taken together, may be almost
The primary sections may be electrically ar- 70 the same as on layer 01" section 81 or S4. No taps
ranged in two different ways, as illustrated in need be taken from the secondary except at the
Figures 2 and 3 of the accompanying drawings. end of a layer, which provides economy of pro-
In Figure 2 sections P1 and Pi are connected in duction. Since the layers or sections 81 and 84
series to provide one primary half, and sections are connected in parallel any unbalanced leak-
Pa and P1 are connected in series to p.rQvig~ 1ihE.! 75 a~e tlu~ between the primary halves tends to
 2,680,218
5 6
set up circulating currents in the layers Sl and 84- insulated multifilarly wound conductors con-
which tend to cancel the unbalance, This is of nected re-entrantly in series, means connecting
great importance in amplifiers operating class the remainder of said secondary sections in paral~
AB-2 or B. leI, and means for interconnecting one of said
The transformers windings hereinabove dis- 5 multifilarly wound conductors, and all said multi~
closed may be applied to transformers of the filarly wound conductol'S taken in series re-en-
shell magnetic circuit type, using a single coil. trantly, in series with said remainder of said
The extension of the principle to core type units secondary sections in parallel, said remainder of
utilizing two similar coils is believed obvious, and said secondary sections having substantially equal
to come within the scope of the invention. 10 numbers of turns.
While I have described and illustrated specific 7. An audio transformer, comprising an even
forms of the invention it will be clear that varia- number of concentrically arranged longitudinally
tions thereof may be resorted to without depart- co-extensive primary sections, means connecting
ing from the true scope of the invention as £ie- said primary sections in two relatively symmetri~
fined in the appended claims. 1;) cal primary halves to provide a balanced primary
What I claim and desire to secure by Letters winding, a secondary winding comprising a plu-
Patent of the United states is: rality of concentric sections symmetrically ar-
1. An audio transformer. comprising, foul' con- ranged each with one of said primary sections,
centrically arranged primary sections, means means for connecting said last named sections
connecting said primary sections in series by 20 in parallel, said secondary winding further com-
pairs to provide a balanced primary winding, 8, priSing a plurality of similar multifilarly wound
secondary winding comprising two concentric layers, means for connecting said multifilarly
sections symmetrically arranged with respect to wound layers in re-entrant series connection with
said primary sections and connected in parallel, each other, means for connecting said multifilarly
said secondary winding further comprising two 25 wound layers in series with said parallel con-
similar multiftlarly wound layers, means con- nected first mentioned sections of said secondary
necting said two multifilarly wound layers in re- winding, said multifilarly wound layers arranged
entrant series connection with each other, and physically symmetrically with respect to said con-
in series with said parallel connected first-men- centrically arranged primary sections and said
tioned sections of said secondary winding, said 30 first mentioned sections of said secondary wind-
two multifilarly wound layers arranged physi- ing.
cally symmetrically with respect to said four 8. An audio transformer, comprising, an even
concentrically arranged primary sections and number. at least four, of concentrically arranged
said first-mentioned two sections of said sec- longitudinally co-extensive primary sections,
ondary winding. 35 means for connecting said primary sections in
2. The combination in accordance with claim 1 two relatively symmetrical primary halves to pro-
wherein physically adjacent ones of said primary vide a balanced primary winding, a secondary
sections are connected in series by pairs to pro- winding compriSing a plurality of sections sym-
Vide said balanced primary winding. metrically arranged each with one of said primary
3. The combination in accordance with claim 1 40 sections, means for connecting said last named
wherein physically separated ones of said primary sections in parallel, said secondary winding fur-
sections are connected in series by pairs to pro- ther comprising a plurality of similar closely
vide said balanced primary winding. coupled layers, means for connecting said closely
4. An audio transformer, comprising, an even coupled layers in re-entrant series connection
number of concentrically arranged longitudinally 45 with each other, means for connecting said closely
co-extensive primary sections. means connecting coupled layers in series with said first mentioned
said primary sections in two relatively symmetri- sections of said secondary winding, said closely
cal primary halves to provide a balanced primary coupled layers arranged physically symmetrically
winding, a secondary winding for said trans- with respect to said primary sections and said
former, said secondary winding comprising an 50 first mentioned sections of said secondary Wind-
odd number of concentric secondary sections, ing.
each co-extensive longitudinally with said pri- 9. An audio transformer comprising a second-
mary sections and symmetrically located with. ary winding comprising a first section, a second
respect to said primary sections, means connect-
ing the outermost and innermost of said sec- section. a third section, said third section com-
ondary sections in parallel, said last named sec- 55 prising two multifilarly wound conductors re-
tions being of substantially equal numbers of entrantly connected in series. means connecting
turns, and means connecting the remainder of said first and second sections in parallel, said
said secondary sections in series with said sec- first and second sections having substantially
ondary sections in parallel. equal numbers of turns, and means connecting
5, The combination in accordance with claim 4 60 one end of said third section to one end of said
wherein said remainder of said secondary sec- first section.
tions comprises a plurality of multifllarly wound References Cited in the file of this patent
conductors connected re-entrantly in series.
6. An audio transformer, comprising, a first 65 UNITED STATES PATENTS
primary winding section, a second primary wind- Number Name Date
ing section. a third primary winding section, a 2,477,074 McIntosh __ ~ ______ ~ July 26, 1949
fourth primary winding section, all said primary 2,535,554 Thurston __________ Dec. 26, 1950
winding sections wound as co-axial cylinders, a
plurality of secondary sections interleaved with FOREIGN PATENTS
70
said primary sections, at least one of said second- N"umber Country Date
ary sections comprising a. plurality of mutually 240,527 Switzerland _________ May 1, 1946
June 1, 1954 H. I. KEROES 2,680,218
AUDIO TRANSFORMER
Filed Oct. 26, 1950

INVENTOR
HERBERT I. KEROES

~
~ A~ORNEY
BY
 2,710,312
United States Patent Office Patented June 7, 1955

1 2
output and dissipation. We have therefore analyzed the
effects of tetrode ol,Jeration in intermediate conditions,
2,710,312 i. e. with the screen connected to part of the load as
an active tube output element. In this condition the
ULTRA LINEAR AMPLIFIERS () screen is partially energized by the output Signal due
to plate current variations, and partially by a D.-C.
David HaBer and Herbert I. Keroes, PhilalMphia, Pa., static operating potential. The distinction from full
assagnors to Aero Products Company, PbiladeJphia,
Pa., a firm triode operation of a screen grid tube is then, that the
screen is not connected directly to the anode, but across
Application May 20, 1952, Serial No. 288,908 10 part of the anode load only.
6 Claims. (CI. 179-171) Experimentation on this basis has produced remark-
able and unexpected results. It is found that for specific
ranges of percentage of screen loading taken in com-
parison with anode loading, depending on tube types
The present invention relates generally to an1plifier~, 15 employed and modes of operation, very favorable op-
and more particularly to amplifiers capable of wide erational characteristics result. In this connection a
band linear response, operable with substantially no tetrode, operated in the normal manner, is taken to have
distortion produced by harmonic response or intermodu- 0% screen loading. When operated as a full triode
lation over a wide dynamic range, and with excellent it is taken to have 100% screen loading. For inter-
efficiency. The amplifier of our invention finds par- 20 mediate values the percentage of screen loading to anode
ticular application to audio power amplifiers, but may loading is defined as the square of the voltage ratio
be applied with equal effect to a wide range of uses, between signal voltage at the screen, and signal volt-
and at supersonic and radio frequencies, as well as at age at the plate. Percentage screen loading may thus
audio frequencies. be defined as the percentage of plate circuit signal pow-
It can be demonstrated that a wide pass band and 25 er transferred to the screen. It should be noted that
low harmonic response, in an audio power amplifier, power is transferred to the screen only over part of
do not of themselves provide quality performance, when the signal cycle, i. e. when the absolute value of plate
performance is judged by the initiated listener, because potential falls below the absolute value of screen po-
speech and music are not of static nature. There have, tential. This transfer has the effect of linearizing the
accordingly, been developed criteria of amplifier per- 30 plate characteristics.
formance in terms of intermodulation distortion analysi~ It is found, for example, that with tubes of the 6L6
and square wave testing, both of which simulate dy- type, operating Class AB1, the internal impedance of
namic conditions to considerable extent. Even such the tube drops very sharply as one proceeds from zero
tests are not fuBy determinative of amplifier perform- screen loading, but levels off at a very low value be-
ance, sincc amplifiers which test well do not always 35 yond about 18% screen loading. Maximum undis-
sound well, although amplifiers which test poorly al- tor ted power Olltput falls only slightly out to about
ways sound poorly. 18% loading, and is very high, but beyond that value
On the basis of listening tests, which are conceded to drops fairly rapidly. Low level distortion decreases
provide the ultimate criteria, there are two schools of rapidly from 0% to 18% loading, and far less rapidly
thought. One adheres to the use of triode amplifiers, 40 thereafter. High level distortion is low out to 18%
on the ground that such amplifiers produce sweet or loading, but increases rapidly as loading is further in-
smooth sound. The other adheres to the use of tetrodes, creased.
usually of the beam power type, as providing crisp or It follows that with properly proportioned screen
clean sound. Each tube type obviously produces its loading, depending on tube .type, a tetrode can provide
own character of distortion, which is pleasing to its ad- 45 the high output power normally associated with tetrodes,
vocates, and displeasing to its opponents. It is found but with low internal impedance, with very little more
that these distortions are elusive, and unmeasurable, but low level distortion than is produced by a triode, but
nevertheless real to the critical listener. with much less high level distortion.
It would then appear that reconciliation of the two In effect, then, we have provided a new tube type,
schools of thought could only be accomplished by some 50 which is neither triode nor tetrode, but which possesses
new tube type, not presently available, and which might advantages of each, and which is of increased linearity.
provide a response acceptable to both. Such a tube The latter effect can perhaps be simply explained as a
should have certain desirable electrical characteristics, straightening of the plate transfer characteristics of a
in addition, which are not presently all available in triode, which are concave upward, by virtue of the plate
either the triode or the tetrode. These may be listed [15 transfer characteristics of the tetrode, which are con-
as follows: cave downward. Whatever the true explanation, the
( 1) Low internal impedance-now offered by the tri- plate transfer characteristic of the tetrode arranged in
ode but not the tetrode. accordance with our invention is remarkably linear.
(2) High power sensitivity-now offered by the It may be stressed that an approximately 18% 10ad-
tetrode but not the triode. 60 ing figure for optimum performance applies to several
(3) Lower harmonic and intermodulation distortion tube types, including the 6L6, the 5881, the 807 or
than either the triode or the tetrode, at both high and KT-66. These latter enable 24 watts output to be ob-
low levels. tained for .75 volt input, when provided with a phase
(4) High efficiency, to permit adequate power out- inverter and driver stage. For tubes of the 6V6 type
put without undue bulk. or cost. 65 a screen loading value of about 5% is preferred, and
The basic difference between the triode and the tetrode for still other tube types still a different value may be
is the screen grid in the latter. Its presence gives the optimum.
tetrode high efficiency, but its absence gives the tri- It is known that a limiting factor in operating screen
ode low plate resistance. grid tubes in the normal operation, i. e. with zero screen
When the screen grid of a tetrode is connected di- 70 loading, is screen dissipation. Tube manufacturers gen-
rectly to the anode of the tetrode the resulting tube is erally list static operating potentials and currents for
effectively a triode, which, however, has limited power maximum allowable screen dissipation. Under dynamic
 2,710,312
3 4
conditions these may exceed the allowed maxima [or a posiie response occurs, and so that tube impedance is
sufficiently long period to cause tube damage, when a low, but output high. This means that the tetrode 1
tetrode is operating as such, i. e. with screen fully un- operates like a tetrode in respect to output level, like a
loaded. triode in respect to internal impedance, and with a plate
With our novel arrangement the potential of the screen 5 transfer characteristic intermediate that of a triode and
is no longer fixed, but it follows tIle potential of the tetrode, and more "linear than either.
plate. It foHows that the screen does not as greatly The windings 16 and 14 are wound on a common
exceed the plate in potential, as in the case when the core 15, and jointly contribute output to secondary
screen is unloaded, and the tube can be safely operated winding 12.
at voltages considerably ill excess of those recommended !{I The !lame parts in Figures 1 and 2 are sim.ilarly ref-
as maxima by the manufacturer, with consequent en- erenced, and it will be noted that Figures 1 and 2 are
hanced performance. similar, except that in Figure 2 a single primary winding
It is, accordingly, a broad object of the present inven- 20 is utilized, which is tapped at a point 21, for con-
tion te provide a novel signal amplifier, of improved nection to screen grid 5. The point 21 is so selected as
performance and operating characteristics. J!) to load the screen grid 5 preferably about 18.5%, or in
It is a more specific object of the invention to provide ihe T::,ngc 5% to 25%, as in Figure 1. Figures 1 and 2
a mode of t·~trode operation, in an amplifier circuit, such represent then equivalent alternative constnlctions, Fig-
that ultra linear input-output relationship is accomplished. ure 2 being the simpler and more economical, but being
It is a still more specific obje.;t of the invention to subject to the limitr.tion that the screen grid 5 and the
provide an audio amplifier circuit employing one or more 20 plate 2 are at identical D. C. static potential.
tubes having screen electrodes, in v;hi,=h the one or more Figure 3 shows the connection of two tubes in push-
tubes are provided with loaded screen electrode circu;ts, i>u1i arrangement, following the teaching of Figure 2 of
loading being selected to provide low intern31 tube im· th:; c:cc'Jmpanying drawings. It will be clear to one
pedance, high output, and low intermodulation distortion skilled in the art that the arrangement of Figure ] may
and harmonic response. ::!5 b.~ push-puU, instead of single ended, if desired.
It is a further specific object of the invention to pro- 1n the system of Figure 3, the alTangement of Figure
vide screen grid tube operation in an amplifier, the screen 2 is converted to push-pull or double ended operation.
grid of the tube being loaded between abOllt 5% and The tubes 30, 31 are respectively provided with cathodes
30%. 32, 33, cOlltroi grids 34, 35, screen grids 36, 37 and
The above and still further feat~lres, objects and ad- 30 anodes 3!'l, 39. The cathodes 32, 33 are directly inter-
vantages of the present invention will become evident c{;l1nected, and via bias resistance 40 to ground. The
upon consideration of the following detailed description resistance 49 is by-passed for audio signals by a con-
of several specific and preferred embodiments of the denser .rI.. It is evident that other arrangements for
invention, especially when taken in conjunction with the providing bias may be used also.
accompanying drawings, wherein: 35 Connected between control grid 34 and ground is a
Figure 1 is a schematic circuit diagram of a first grid leak 4J, and between control 35 and ground a grid
embodiment of our invention, operated single ended; leak 42. The control electrodes 3<1 and 35 are driven
Figure 2 is a schematic circuit diagram of a modifica- in balanced relation with respect to ground, by signals
tion of the arrangement of Figure 1; applied in opposed phase to leads 43, 44. Alternately,
Figure 3 is a schematic circuit of a push-pull ampli- 4Q other input coupling schemes can be used.
fier corresponding generally with the embodiment of Primary winding halves 45, 46 are connected in series,
Figure 2; respectively, with the anodes 38, 39, the junction of the
Figures 4 and 5 are schematic circuit diagrams of fur- winding halves 45, 46 being supplied with a terminal
ther modifications of our invention; and 47 for connection to a source of B+ voltage. A sec-
Figure 6 is a plot illustrating certain operating char- 45 ondary winding S is coupled with primary halves 45, 46.
acteristics of the arrangements of Figures 1-5, inclusive. The screen grids 36, 37 are connected at points 48,
Referring now more specifically to the drawings, the 49 of winding halves 45, 46. The points 48, 49 are
reference numeral 1, in Figure 1, identifies a tetrode selected to provide for the screen grid'S 18.5% of the
vaCUllm tube, h1!ving an anode 2, a cathode 3, a control loading provided for anodes 38, 39. It wiI! be clear
grid 4-, nnd a screen grid 5. A bias resistor 6, by-passed n that the specified percentage is not critical, but preferred,
for audio signals by a condenser 7, is connected between and that a range of values about 5% to 25% may be
cathode 3 and ground. A grid leak resistor 8 is con· selected, without serious operational variation, as will
nected bet·."'ee~l contrel grid 4 and ground, and signal become clear when the operations I characteristics of thc
to be amplified is applied to iead 9, and hence between amplifier are described hereinafter.
control grid 4 and ground. The circuit as specifically In the modification of our invention illustrated in Fig-
described to tbis point is conventional. ure 4 of the accompanying drawings the loading of the
In the anode cirC1.1j~ of the tube 1 is connected the screen grid is accomplished by connecting a choke 50
primary winding !@ of an audio outpul transformer 11, across primary winding la, aod the screen grid is '.:on-
provided with a :;econu~_ry winding 12. The winding 10 neeted to a point 51 of the ehoke SO such as to accomplish
is in series with a B+ source (not shown), but con.- til' the desired loading. The choke thus acts effectively as
neeted to terminal 13. A further primary winding 14 an allto-transformer. The result accomplished by means
is provided, in transformer 11, in series with a separate of choke 50 is accomplished in the embodiment of ollr
voltage source B' +. In the alternative, it will be evi- invention illustrated in Figure 5 0f the accompanying
dent that windings 14 and HI may be connected in par- drawings by means of a resistor 52, connected across
allel, and to a single common B+ source. /:" winding 10 and tapped at a peint 53' which is 18.5%
The l.Vinding 14 is provided with fewer turns than is from the B+ end. The tap 53 is connected with screen
the winding lS, so that its i;npedance at any frequency grid S. ]n other respects the embodiments of our inven-
is rreferably about 18.5% of the impedance of winding tion illustrated in Figures 4 ancI 5 follow closely the em-
lQ. ali-hollgh this value is optirnl:m and not critical, and bodiment of Figure 1, and accordingly, identical parts
a ;iatisfactory r&nge of values may be taken to extend ;-0 and elements in the several Figures 1, 4, 5 are identified
frem t5% to 25%, for tubes of the 6L6, or the 807 by corresponding numerais of reference.
type. It is conceivable that a ciifferent percentage of It will be clear that each of the embodiments of our
loading might be reqaired for tubes of different design invention which- is illustrated as operating single ended
parameters than those tested by us to date, but in all may be modified to operate in push-pull or balanced re-
cases the value is so selected that extremely linear com- 7:1 fatlon, by adopting the general circuit configuration of
 2,710,312
5 6
Figure 3, modified to include the appropriate screen grid control electrodes in balanced relation to a balanced
loading expedients of Figures 1, 4, 5, respectively. source of high fidelity audio signal, means for biasing
Referring now to Figure 6 of the accompanying draw- said first and second control electrodes for alternative
ings, there are illustrated four plots, A-D, inclusive. anode current flow in response to alternate half cycles of
Plot A shows the range of values of tube impedance for 5 said high fidelity audio signal, a balanced output circuit
all ranges of screen grid loading, from 0 to 100%. connected between said anodes, said balanced output
Screen grid loading of 0% represents zero impedance in circuit comprising a primary winding of an output trans-
series with the screen grid, while screen grid loading of former, a balanced output circuit connected bew/een said
100% represents triode operation, i. e. equal impedance screen electrodes, said last mentioned output circuit in-
in series with anode and screen grid. The latter opera- 10 cluding a portion of the turns of said primary winding,
tion is normally accomplished by direotly connecting the said amplifier having operating characteristics as a flmc-
screen grid to the anode of the tube. It will be clear tion of screen gtid loading such that tube impedance has
from the plot that the internal impedance of the specific substantially a minimum value for a relatively critical
tube for which the plot was made decreases rapidly as value of said loading, value of watts of undistorted out-
screen loading increases from 0%, until a value of screen 15 put has 8ubstantially a maximum for said value of load-
loading of about 18.5% is attained. For further increase ing, percentage of high level distortion caused by inter-
in loading but relatively slight decrease in tube impedance modulation has substantially a minimum value for said
is attained. value of loading, and percentage of low level intermodu-
Plot B indicates the relation between watts W of un- lation distortion has substantially a minimum value for
distorted output and values of screen loading between 20 said value of loading, and wherein the points of connec-
0% and 100%. It will be noted that the value of output tion of said screen grids to said primary winding are sub-
is reasonably level out to about 18% loading, and there- stantially such as to provide said relatively critical value
after decreases to a rela'tively low value for triode of said loading.
operation. 2. The combination in accordance with claim 1 where-
Plot C indicates the percentage of high level distortion 25 in said screen grid loading is substantially 18.5%, and
caused by intermodulation. It will be noted that distor- wherein said vacuum tubes are equivalents of the 6L6
tion at 18% screen loading is about 10%, and is about type.
the same as exists for normal tetrode operation. This 3. The combination in accordance with claim 1 where-
type of distortion increases rapidly thereafter, as triode in said vacuum tubes are equivalents of the 6L6 type.
operation is approached. Plot D, which indicates per- 30 4. The combination in accordance with claim 1 where-
centage low level intermodulaiion distortion, shows about in said vacuum tubes are equivalents of the 6L6 type and
.6% distortion for normal tetrode operation, and de- are operated class ABl.
creases to about .4% for 18.5% screen loading. For 5. The combination in accordance with claim 1 where-
further increase of loading relatively slight improvement in said vacuum tubes are equivalents of the 6V6 type and
in low level intermodulation distortion is accomplished. 35 said screen grid loading is approximately 5 per cent.
H will be clear from the plots A-D, inclusive, that a 6. The combination in accordance with claim 1 where-
considerable range of screen loading, deviating from in the tube types employed are selected from commercial
18.5% loading, will provide high quality operation, and types 6L6, KT-66, 5881, and 807, and wherein said
the useful range may be described as extending from 5 % screen grid loading falls in the range 18 per cent to 26
to 30%, 18.5% being clearly an optimum value, for the 40 per cent.
particular tube for which the plot of Figure 6 was made.
While we have described and illustrated one specific References Cited in the file of this patent
embodiment of our invention, it will be clear ,that varia- UNITED STATES PATENTS
tions of the general arrangement and of the details of Robinson ______________ Apr. 3, 1934
construction which are specifically illustrated and de- 45 1,953,775
2,216,465 Wilhelm _______________ Oct. 1, 1940
scribed may be resorted to without departing from the Blumlein ______________ Oct. 22, 1940
true spirit and scope of the invention. In particular the 2,218,902
2,273,511 Bruck _________________ Feb. 17, 1942
invention may be applied to any tube type which is suffi- Cannon ________________ Aug. 7, 1945
ciently analogous in operation ,to the screen grid tube, 2,380,923
2,651,685 Tharp _________________ Sept. 8, 1953
and particularly to pentodes and beam power tetrodes, 50
which provide screen electrodes. OTHER REFERENCES
What we claim is: RCA Receiving Tube Manual, page 21; pub. 1947 by
1. A push-pull high fidelity audio amplifier having a R. C. A., N. Y. Copy in Div. 69.
first vacuum tube and a second vacuum tube, said first Terman text, "Radio Engineering," 3rd ed., pages
vacuum tube having a first anode, cathode, control elec- 55 299-307; pub. 1947 by McGraw-Hill Book Co., N. Y. C.
trode and screen electrode, said second vacuum tube hav- Copy in Div. 69.
ing a second anode, cathode, control electrode and screen
electrode, means for connecting said first and second
 June 7,1955 D. Hafler & H. Keroes 2,710,312
ULTRA-LINEAR AMPLIFIERS

Filed May 20, 1952

.fIq.l
10
/15

~
II

.frq.3 JiIq.4
10

INVENTORS:
DAVID HAFLER &
HERBERT I. KEROES

BY:~~
THEIR ATTORNEY

'A' '8' 'C' 'D' Percentage of primary turns from B+ to screen tap
Load OutW % 1M % 1M 0 32 45 55 63 71 77 84 89 95 100
- .. -
--- .
., .
160n 20 40 0.8

120n 15 30 0.6 \ - .~
f-B
"
.
,

".~ ....... ..... ) .'

"

80n 10 20 0.4 "

......
...
'y' ...:: ... t-~ -'-
\ ' - r- ...... ".-
40n 5 10 0.2 .......... \. .. ...... ,- - ·\:.C 0 ---
on o o 0.0
"- A. . . .
o 10 20 30 40 50 60 70 80 90 100
See '5' on page 3 Tetrode Ultra·linear Triode
for clarfication . Percentage of screen-load to plate-load impedance
 2,791,6'46
United States Patent Office Patented May 7, 1957

t 2
Accordingly, the primary winding is arranged as four
interleaved concentric sections, two of which, in series,
provide one of the primary halves, and the remaining two
2,791,64(j of which in series provide the other of the primary halves.
ELECTRICAL SYSTEM FOR ULTRALINEAR 5 The secondary winding is composed of three axially con-
AMPLIFIER centric sections, two of which are of equal number of
turns. The latter two sections are connected in parallel,
Berbert I. Keroes, Philadelphia, Pa. and interleaved, respectively, one between the two inner-
Application October 26, 1955, Serial No. 542,825 most primary sections and the other between the remain-
10 ing two. The parallel connected sections provide a low
6 Claimll. (Cl.179-171) impedance secondary winding which is symmetrically ar-
ranged with respect to the primary winding sections, and
evenly distributed thereamong; and which therefore is
electrically in balance with respect to the primary wind-
The present invention relates generally to audio am- 15 ings, i. e. possesses equal capacitance with respect to each
plifiers, and more particularly to high fidelity audio am- half of the primary. Capacitive coupling from the sec-
plifiers operating at high efficiency, and to transformers ondary to each primary half is approximately the same,
employed therein. and since voltage on one side of the section is arranged to
The present invention is an improvement of that dis- be of opposite phase to that on the other side, capacitive
closed and claimed in United States Patent No. 2,710,312, 20 coupling is effectively balanced out.
issued to Keroes et al. That patent discloses an audio am- The remaining section of the secondary winding is
plifier of the high fidelity type, which operates class B (.r formed of two separate conductors, parallel wound to
class AB, i. e. in push-pull, and with a grid bias selected form a "re-entrant" series connection, where the term re-
to aSSUl;e that two push-pull connected tubes shall conduct entrant means parallel wound mutually insulated COll-
in alternation. Applicants, in the above identified patent, 25 ductQrs, each jn series with the other by connecting the
disclose a system in which the amplifier tubes employed terminal point of one to the initial point of the other.
have screen grids, which are connected to relatively criti- One of these conductors is connected in series with the
cal points of an output transformer. whereby to optimize low impedance secondary to provide a secondary of inter-
simultaneously a number of operating characteristics of mediate impedance. Both re-entrant conductors in series
the amplifier. This operative cQndition is denominated ao with the low impedance secondary provide a high inl-
ultra-lineal' operation, herein. J\pp1ic~nts also disclose pedance secondary. The secondary windings so formed
that the screen grids may be driven from a tertiary wind- each remains symmetrical with respect to the primary
ing of an output transformer. Screen grid d.rive by means windings, the re-entrant conductors being located between
of a tert.ary winding may be C$sentia! where the tube primary sections No.2 and No.3.
types employed require different D. C. voltages at the 115 The tertiary windings are interleaved, between the pri-
anodes and 'screen grids, respectively. mary pairs, each tertiary winding immediately adjacent a
The schematic circuit diagram of a system of the gen- primary layer most directly connected to the B+ term i-
eral type above briefly described appears to be relatively nals, and is usually a single layer, although more than one
simple and straightforward. The actna! physical arrange- layer may be employed. This location assures minimum
ment of windings, in an act~lal transfOI'mer, is another 40 capacitive cunent to the tertiary sections, and minimum
matter, and it is with the problem of physical arrange- leakage reactance, since the direction of the screen wind-
ment and connection of the several win(lings of a push- ing is the same both in rotation and in starts and finishes
pull output transformer, in which tertiary screen grid as for the adjacent prnnary plate section, and since the
windings are employed, that the present invention is con- tertiary winding is shielded from the remaining adjacent
cerned. 41) primary section by a secondary section. which is essen·
Briefly describing the invention, a pair of electronic· tially at ground potential.
amplifier tubes are connected in push-pull relation, and The primary windings are so arranged that their mutual
are l;>iased preferably for class B or class AB operation. inductance adds to the series inductance of the separate
The tubes are driven in balanced relation, and drive an winding sections, to establish maximum inductance in
output transformer having a push-pull prinl,ary winding, 60 each primary half for the amount of copper and iron em-
and a secondllfY winding. A tertiarY Winding is assem- ployed in the transformer.
bled on a common core with the primary and secondary The physical distribution and winding relation of the
windings, and is employed to drive the screen grids of the tertiary windings with respect to the primary sections,
tubes congruently with the anodes. but in association with which effects minimum voltage with Tespect to one ad-
a lower impedance. Accordingly, the screen grids follow 1111 jacent primary section and a shielded relation to the other,
the A. C. excursion of the anodes, but at a reduced A. C. implies minimum leakage reactance between windings,
level. The screen grids may be supplied with D. C. volt- and minimum capacitive transfer. The fact that each
age from a separate source than that which supplies the tertiarY winding is between two primary windings of oppc.r
anodes, where the anodes and screen grids require differ- site phase tends to. balance out, or to minimize, caoacitive
ent D. C. operating vol~ages. 80 current flow between the primary and tertiary windings.
In accordance with. the present invention, the primary Since the secondary windings are in balanced relation to
winding is arranged as four interleaved concentrie sec- the primary sections, capacitively, and since leakage re-
tions, superposed. The innel'most .ayer of the innermost
actance is minimized, the entire system of windings is
section (No.1) and the outer layer of the outermost sec-
tiQn (No.4) are connected, respectively, to the anodes of illS intercoupled capacitively to a minimum extent, and the
a pair of p\lsh-pull connected tubes. The remaining sec- leakage reactance between each pair of windings is mini-
tions, i. e. No.2 and 3, starting at the core, are connected mized.
to a source of anode voltage, the innermost layer of sec- It is" accordingly, an ~bject of t?e pr.esent .inventio,D
tion No.2 and the outermost layer of section No.3 being to provH.ie a novel ultra-hnear amplifier, In which a pall'
so connected. Sections No. 2 ~nd 4 are connected in' 70 of p.ush-pull connected screen grid amplifier tubes is
series to one anode, and sectiOJ;lS No. 3 and, No. 1 to the provided with optimum coupling between anodes and
other anode, in the re(:ited order. screen &rids, by means of a tertiary winding, which is
 ~;791,846
3 4
part of an output transformer having minimum inter- current between windings 31, 37, is minimized, as well
winding capacity and leakage reactance. as capacitive currents between windings 35 and 38, be-
It is still another object of the present invention to cause relative A. C. voltage excursions are eliminated.
provide a novel high fidelity audio transformer having Moreover, leakage reactance is also minimized, because
tertiary windings, in which minimum capacitive"coupling 5 the windings are axially co-extensive and immediately
and minimum leakage reactance exists, between primary adjacent. Moreover, secondary winding 37, which is ad-
windings and the tertiary windings, and between second· jacent to ground in potential, acts as a shield between
ary windings and the tertiary windings, as well as between screen winding 37 and primary winding 36, and minimizes
primary and secondary windings. capacitive transfer from winding 36 to tertiary winding
The above and still further objects," features and ad- 10 37. Similar considerations apply to tertiary winding 38,
vantages of the present invention will become apparent which is shielded from winding 32, for capacitive transfer,
upon consideration of the following detailed description by secondary winding 41.
of one spech'1c embodiment thereof, especially when taken The arrangement of secondary windings is i1lustrated
in conjunction with the accompanying drawings, wherein: in Figure 3 of the accompanying drawings, and includes
Figure 1 is a schematic circuit diagram of an ultra- 15 two parallel connected windings 40, 41, which together
linear amplifier, which illustrates the physical arrange- make a 4 ohm winding, Two further bifilarly related
ment of the several windings of the transformer; windings 42, 43 are included. The winding 42 may be
Figure 2 is a schematic circuit diagram of the primary connected in series with windings 40, 41 taken in parallel,
and tertiary winding arrangement of a transformer in to form an 8 ohm winding, or winding 43 may be con-
ac<;;ordance with the present invention; 20 nected re-entrantly with winding 42 to form a 16 ohm
Figure 3 is a schematic circuit diagram of a secondary winding.
winding in accordance with the invention; and The secondary "'1inding 40 is physically located inter-
Figure 4 is a schematic circuit diagram of a push-pull mediate primary windings 31, 36, and secondary wind-
ultralinear amplifier, having no regard to winding arrange- ing 41 intermediate primary windings 32, 35. Voltage
ment. 25 induced in the separate secondary windings by the sepa-
Proceeding now by reference to the accompanying rate pairs of primary windings, oppose each other inso-
drawings, and particularly Figure 4 thereof, the reference far as due to capacitive coupling, and hence balance out.
numerals 1 and 2 denote, respectively, a pair of push-pull The physical relations of the windings, i. c. the fact that
connected tetrodes. The tetrode 1 includes an anode 3, the secondary windings are co-extensive with the primary
a screen grid 4, a control grid 5 and a cathode 6. The 80 windings axially, and located intermediate primary pairs
tetrode :1 includes an anode 7, a screen grid 8, a control of opposite phase, reduces, respectively, capacitive cou-
grid 9 and a cathode 10. The cathodes 6 and 10 are pIing and leakage reactance.
connected together, and to ground via a bias resistance It will be noted that the winding directions of winding
It, which is shunted by a large condenser 12, which repre- sections 31, 31 arc opposite, i. e. one starts from the
sents essentially a short circuit for aU frequencies within 35 right and the other from the left, and that the terminal
the high fidelity audio band, i. e. 30-20,000 C. P. S. point of winding 31 is connected to the initial point of
The control grids 5 and 9 are connected to ground via winding 32. "Correspondingly, the rotation of winding
grid leaks, 13, 14 respectively, and are coupled to a high is the same, so that the magnetic fluxes produced by the
fidelity audio source (not shown) via balanced coupling separate sections are additive.
condensers 15,16 respectively. 40 Since the screen windings 37,38 are immediately adja-
The anodes 3 and 7 are connected to opposite ends of cent the primary winding layers of highest D. C. poten-
a high impedance primary winding 17, of a transformer tial, i. e. adjacent the supply lead, the total A. C. capaci-
J 5, the center point 19 of which is tapped, to provide a tive charging current to the screen winding is minimized,
B+ + terminal, 21. A secondary winding 20 is coupled and insulation requirements are reducted. At the same
to the primary winding 17, and serves as an output wind- 45 time the secondary winding operates as a screen, to at
jng. A tertiary winding 22 is also coupled to primary least partially shield the screen windings 37, 38 from
~inding 17, and drives the screen grids 4 and S, develop- the primary sections 31,35.
mg A. C. voltages which duplicate those across primary The overall arrangement of windings, considered in
winding 17, except in respect to magnitude. The tertiary their physical and electrical relations, provide a high
winding 22 is center tapped at Z3, to enable balanced 60 fidelity transformer having a screen driving tertiary wind-
connection to a B+ terminal 23. ing, wherein the latter does not detract from ultraIinear
The D. C. voltages applied to terminals 21 and 23 are, operation, by virtue of capacitive coupling therewith of
or may b~ of different magnitudes, and the relation of other windings, or by virtue of leakage reactance.
the number of turns in the windings 17:, 22 may be While I have described and illustrated One specific
selected to establish ultralinear operation, in the sense of fill embodiment of my invention, it will be clear that varia-
United States Patent No. 2,710,312, hereinabove referred tions of the general arrangement and of the details of
to. construction which are specifically illustrated and de-
. Referring now to Figure 2 of the accompanying draw- scribed may be resorted to without departing from the
lOgS, the reference numeral 30 denotes an anode voltage true spirit and scope of the invention as defined in tbe
terminal of a half primary transformer winding, including 60 appended claims.
two sections 31 and 32, connected in series, and so wound What I claim is:
on a common core that the inductances of the windings 1. An ultralinear amplifier baving a first electronic
add, and the" mutual inductance further adds, whereby amplifier tube having a first anode, cathode, screen grid
maximum inductance is associated with the primary half, and control grid, a second electronic amplifier tube hav-
for a givm amo;.tnt of iron and copper in the transformer. 65 ing a second anode, screen grid, control grid, and cathode,
The terminal 34 is a further anode voltage terminal, and means for biasing said amplifier tubes for alternative
is connected to winding sections 35, 36, in series. The conduction in succession, an output transformer for said
anode terminals of the transformers are denoted P. amplifier, said output transformer having two first pri-
The screen winding 37 subsists between windings 31 mary winding sections and two second primary winding
and 36, and is co-extensive with and wound in the same 10 sections, said first and second sections arranged in alter-
sense as, and starts at the same point as, the winding 31, nation 'and co-axially, whereby said first primary winding
and is immediately adjacent thereto. Similarly, screen sections are separated by a second primary winding sec-
winding 38 is co-extensive with winding 35, and is wound tion, and said second primary winding sections are sepa-
similarly and initiates and terminates similarly, and is rated by a first primary winding sections, said first pri-
immediately adjacent thereto. It follows that capacitive 111 mary winding sections being wound in opposite d.irec-
 9,791,84.8
& 8
tions, said second primary winding sections being wound connecting said second and fourth sections in series, and
in opposite directiom, the sense of winding of said sec- sCtOnd and fourth sections relatively wound to provide
tions such as to maximize primary inductance and to maximum inductance.
minimize leakage reactance and capacitive coupling, and S. The combination in accordance with claim 4, wherein
two tertiary windings respectively connected to said 5 is provided a tertiary winding, said tertiary winding hav-
screen grids and coupled to said primary windings, each ing a first winding element 'and a second winding element,
of said tertiary windings being physically located inter- said first winding element comprising at least one wind-
mediate a first and second primary winding. ing layer superposed on the outer layer of said second
2. The combination in accordance with claim 1 wherein section, and wound identically in sense and extent with
each of said tertiary windings is co-extensive axially with 10 said outer layer of said second section, said second ele-
one of said primary winding sections. ment comprising at least one winding layer wound imme-
3. The combination in accordance with claim 2 wherein diately under the inner layer of said third section, and
the point of highest D. C. potential of one of said ter- wound identically in sense and extent with said inner layer
tiary windings and the highest D. C. potential of one of said third section.
of said primary .windings are immediately adjacent one 15 6. The combination in accordance with claim S wherein
another. is provided a secondary winding in said transformer,
4. In combination, a push-pull amplifier having a push- said secondary winding including a first, second, third
pull output transformer, said push-pull output transformer and fourth secondary section, said first and second sec-
having an output winding having first, second, third and ondary sections connected in parallel, said first section
fourth multi-layer winding sections, said sections arranged 20 located intermediate said first tertiary element and said
co-axially, means for connecting an anode voltage supply first primary section, said second secondary section lo-
to the outennost layer of turns of said second section, cated intermediate said second tertiary element and said
means for connecting an anode to the innermost layer fourth primary section, and said third and fourth sec-
of said fourth section, means connecting an anode volt- ondary sections re-entrantly connected in series with each
age supply to the innennost layer of said third section, 25 other and with said parallel connected first and second
means for connecting an anode to the outermost layer secondary sections and 'located intermediate said second
of said first section, means for connecting said third and and third primary winding sections.
first sections in series, said first and third sections rela-
tively wound to provide maximum inductance, meatlS for No references cited.
 May 7,1957 H. I. KEROES - 2,791,646
ELECTRICAL SYSTEM FOR ULTRALINEAR AMPLIFIER
Filed Oot. 26, 1955

ocooooooocco
40 36 37
p

.
SCRE-EN
llr!L2.~~...!-!~~~~:l::SUPPLY
B++

:~F~~~F~~~8o~.~O~483~t3~1~~~~~~;.
o .
o 0
0 0 0 0 0 0 0 0
0 0 0 0 COO 0
~2
30
4.n.
34

~~~)O~oi~35~crl~~:t:B++
r SCREEN
SUPPLY

52 0----'
----,-..,-.,....,.....,
Pz 0 - - - - 1
'--t---eS+
41

f'Iq.5

21

18
INVENTOR
HERBERT I. KEROES

BY~~ ATIORNEY
 An Ultra-Linear Amplifier
DAVID HAFLER'� and HERBERT I. KEROES*

Presenting a new output-stage connection in an otherwise conventional

amplifier which provides a degree of listenability which is well above average.

T HAS BEEN CU.I MED that there is no importance-it is the· most stringent favorers, Therefore, the screen grid
I more room for improvement of power test of all.
output stages since other elements of On the basis of listening tests (defi­
seems to be the element which gives the
tetrode its advantages and its disadvan­
a complete sound system-particularly nitely not on the basis of measurements) tages compared to the triode. In fact,
the electro-mechanical ones-are far in­ the audio school has been divided into when the screen is connected to the
ferior., There is a prevalent belief that two camps-triodes versus tetrodes. plate, the resultant tube is a triode which
"one good amplifier is only marginally There has been shifting between the pop­ is excellent in many respects though
rlifferent from another." The proponents ularity of the two, but there has always handicapped by limited power output
of this line of. thought imply that sig­ been a distinct cleavage. When the tri­ and low permissible dissipation. Control
nificant improvement in power amplifiers ode-without-feedback was judged su­ of the screen is a logical step toward ex-
is extremely difficult to achieve, and perior to the tetrode-without-feedback,
with this idea the authors agree, but the the tetrode school added feedback and
authors disagree as to the need for fur­ reaffirmed the merits of this tube type.
ther improvement. Obviously, the weaker This was again superseded by the triode­
links do need improvement, but this with-feedback, but the beam tetrode still
alone is no reason for abandoning the has its followers, presently in the cate­
further development of stronger links gory of a defensive minority among the
in the chain of audio reproduction-the audio elite.
power amplifier and primarily the power The very fact that each tube type has
output stage which is the prime gener­ ardent supporters is evidence that each
ator of distortion in the purely electronic has definite points of merit. Possibly the
part of the audio system. devotees of each type listen for different
Present thinking is very parallel to qualities of reproduction, and this causes
the views of the 1935 era when it was divergence of opinion. The triode fan
felt that the principle need was for better usually emphasizes "smoothness" or
program sources and that the trans­ "sweetness" of sound. The beam power
ducers and audio amplifiers had reached advocates seek "crispness" or "clean
a stage of near perfection which could sound." Each group obviously desires (8)
hardly be improved. Now, what audio- . sound which simulates the original, but Fir:. 1. Arr.1ngements for energizing the screen
phile would be satisfied with the repro­ each' rejects the elusive and unmeasure­ grid to improve tube line.arity.
rluction standards of sixteen years ago able distortions which characterize the
when playing the new LP's or high tube type preferred. by the opposition tracting the favorahle attributes of the
grarle tape? By analogy, therefore, as camp. A new type of tube, none of which tube and discarrling the unfavorable.
well as for the never-ending search for has been put on the market for, many Experimentally it was found that the
a never-attainable perfection; we must years, might be the thing which could goal of improved operation could be
continue to seek improvement in ev�ry reconcile these diverse views of listeners achieved through energizing the screen
link of the audio chain. who all look for the same thing but seek with d.c. through a special winding on
The old standards for evaluating am­ it in different ways. the output transformer and combining
plifier quality have fallen into disrepute. The requisites for such a new tube the effects of both plate and screen cur­
'
It can be audibly demonstrated that a can be listed readily: rent in the output transformer. This is
wide pass band and low harmonic con­ 1. Low internal impedance, such as is Illustrated at . (A) in Fig. 1 with an
tent do not necessarily mean that the am­ olTered by the triode. alternative ann simpler method shown
plifier satisfies the critical listener. Newer 2. High power sensitivity of the tetrode at (B). It has been found that the
criteria have been developed such as so as to minimize drive problems. screens must be fed from a low�im­
intermodulation distortion analysis and J. Lower harmonic' and intermodulatioll pedance source or the benefits' of this
square wave testing, both of which sim­ distortion than either triode or tetrode arrangement cannot be realized. This
ulate dynamic conditions to some extent at both high and low levels of opera­ eliminates the' possibility of doing the
tion.
and take into consideration that music 4. Sufficiently high efficiency to permit
same job with 'resistive bridge networks
and speech are not. of a static nature. . adequate output without undue bulk or or voltage dividers.
These new tests produce' higher cor­ cost. The screen, load impedance is some­
relation between experimental data and what critical if optimum results are de­
listener preference. Therefore, modern Since no such tube is available, the sired. As the ratio of screen load im�
amplifiers sound better than the ones.of only recourse is to seek a mode of op­ pedance varies from zero (tetrode op­
a few years ago as a general rule. How­ eration of existing type tubes to approxi­ eration) to unity (triode operation),
ever, these tests do not always separate mate the desired qualitie<; and then to important effects show up:
the wheat from the chaff. Amplifiers see whether the theory is justified by 1. The internal impedance takes a sharp
which measure 'well do not necessarily listening tests. drop and then levels o IT.
sound well although an amplifier which 2. Maximum undistorted output drops
shows up as ppor on measurements will Llne"isin, the Output St3,e . slightly at first, then decreases rapidly.
. not sound well. Excellent measurements The physical difference between the J. Intermodulation distortion at high
are a necessary but not a sufficient con­ triode and tetrode is, of course, 'the level operation drops to a minimum
dition for qUillity of sound. This means screen grid. This gives the tetrode its and then soars upward.
that the listening test is the one of most 4. Low level IM decreases somewhat and
efficiency on t)le one. hand, but also in­
then holds almost level.
creases the plate. resistance and con­
• Acro Prod'lCts Company. 369 Shurs tributes toward ' the "tetrode. sound" The situation is demonstrated graph­
Lall�. Philadelphia ·28, Pa. which is so violently disliked by triode ically in Fig. 2 where it can be seen

AUDIO ENGINEERING • NOVEMBER, 1951 15

 'A' '8' 'C' '0' Percentage of primary turns from B+ to screen tap
Performance of th. Amplifier
Load Out W % IM % IM 0 32 45 55 63 71 77 84 89 95 100
0.8
--- - All stages of the amplifier have been
160n 20 40 - - ad j usted for minimum intermodul ati o n,
- B .,
120n 15 30 0.6 \ .--:: and the IM curves based on sine�wave
power output are shown in Fig. 4. These

80n 10 20 0.4 '} .,

.....
'-
...

r- '."'" '-
-� ..... ....
�.
curves were run using trequencies of 40
and 2,000; 40 and 7,000; 40 and 12,000;
100 and 2,000; and 60 and, 7.000, all
40n 5 10 0.2 .•... -I� ..... .. " _
..
'"\..c. '7'
0
1-, :.� mixed four to one. The IM is almost
"'- A....... identical under all conditions of test
indicating that it is completely inoe?,end.
on o o 0.0
ent, of frequency, at least up to 20 watts
o 10 20 30 40 50 60 70 80 90 100
Tetrode Ultra·linear Triode output. This factor possibly accounts for
Percentage of screen-load to plate-load imp'edance the superlative listening quality of the
amplilier.
Fig. 2. Comparison of Ultra-linear operation with triode and tetrode operation using Undistorted power, less than 2 per
a push- stage without feedback. cent IM. is in . excess of 20 watts. This
power is delivered '1mdistorted within
that over a narr o w band of operation from 10 to 100,000 cps with extremely 1 db over the range from 20 to 20,000
where screen load impedance is about low phase shift and no resonances within cps. This power curve (Fig. 5) is not a
18.5 per cent oj plate load impedance, this band. respons e curve run at high po w er 1evel.
the new arrange me nt provides the h i gh Th e complete ampl i fi er circuit is rela­ Instead it represents clean power avail­
power output of tetrodes with low in­ tively simple, inexpensive, and efficient. able at these frequencies. This is par­
ternal impedance such as is normally With a 370 - vo lt power transformer at ticularly important with today's program
obtained from triodes, while distortion 130-l)1a peak requirement, power output· sources. The dynamic range :of some of
figures are equal or better than the ex­ is almost as high as for a tetrode ampli­ the best LP's is reputed to De in excess
tremes of operati oo . We have achieved fier and twice that of a triode amplifier of 100 db. It is necessary to· have pow�r
a new tube type without designing a with cathode bias and the same power. to handle this range, .and this po w er is
new tube. This tube is neither triode
nor tetrode, but its improved linearity
over either of those types justifies the
designation' "ultra linear."

TIwI Complete Amplifier Circllit

In app lying the ultra-linear output ar­
rangement to c omp le te amplifier circuits,
it was found that the simple v er s ion of
ta
(B), Fig. 1 could be u se d to advantage.
By feeding d.c. to the screens through
a prop erl y placed tap on the primary of
th e output transformer, the operating
,
conditions are pre serv ed , and the clo se
coupling ¥t ween screen and plate is
advantageous when feedback is carried
around the )stage. The disadvantage of
this simpler arr a ng emen t is that screen
and plate must ope r at e at the same d.c.
potential . In the pa r ticular arrangement
used the screen and plate are operated
at the same potential (350 volts plate to
cathode) without exceeding dissipati on t10 ... ... tl1
r�quirements, ei ther Guie�cent or at 40 _ 40

maximum output. This new output

coupling arrangement reduces screen
dissipation at high levels and is a safe -=-
mode of operation with .respect to tube
life.
A circuit arrangement has been de­
signed to take fuil advantage of the Fig. 3. O¥er.1! sc:hematic: of the Ultra· Linear amplifier and i5soc:ialed power 5upply.
ultra linear output stage. This circuit,
Fig. 3, takes into account the necessity supply. No adjustments are necessary required over a wide f requ en t y band.
for complete stabi li ty under feedback for balance either of the phase inverter New st and ar ds of audio fidelity are
conditions so as to eliminate tendencies . or of the output-stage plate current, and r a p idly ma k i ng obsolete the five or ten
toward transient instability under any there are no critical values of capacitors watt amplifier w hi ch cannot even deliver
type .of load, including the varying im­ or resistors required. The amplifier is its rated power at f requ ency extremes.
pedance of loudspeaker systems. driven to 20 watts of output with an in­ Another factor of considerable im­
This complete circuit offers li{learity put of only 0.7 volt. portance in evaluating am p l i fi er per­
of operation of a very high order. It is Feedback is carried around the co m­ formance cannot be se en from the curves.
based around a special output trans­ plete circuit in an ext ernal loop. There This is overload characteristic. The
former, the Acrosound TO.300, which are 20 db of feedback in this loop as a mp li fier has been given listening tests
is 6600 ohms' primary. impedance and has measured under load conditions (about under overload conditions with a p ad on
taps at the optimum point indicated in 30 db based on open circuit gain), and the output so as not to deafen the partici­
Fig. 2. A special' seven-section sym­ a safe m ar gi n 15 maintained. A small pants. Peaks which would require a 40-
metrical winding arrangement placed on ca p a cito r across the feedback resistor in­ watt amplifier are transmitted without
a substantial grain-oriented I"mination. creases the feedback in the region above irritation even though the ou tp u t can be
of unique shape permits a ratio of pri- \ 100 kc to smooth the high-frequency .re­ seen to c li p on the 'scope. The overload
mar)' inductance to leakage reactance in sponse. This capacitor is not required to recovery is rapid and h as no noticeable
excess of 15,000 to 1. The response of keep the amplifier stable thou gh it does hangover, so a dip ped peak has n(} ti.me
the transformer alone is within ;t 1 db add to the stability margin. to penetrate the ear. Some amplifiers

AUDIO ENGINEERING • NOVEMBER, 1951 16

 +t

."
I I IIIIII I I I I Ill" I I IIIIIII I
o
J I I I 11111 I 1 .... .. 0 D
V
V ".,,,MU," UNOISTORT( 0 "0"'[ R OUT "UT
•• FREOUENCY ,/
�

.% OlD. %4 WATT!;
!lU. 01$TO"TION. 2,\ TOTAL �AR"ONIC
t
,. . . . .
,. ,-
�. .
',- .0 60 20.000
,... �QUI:NCV IN CYCLES "lEA 5�CONO "oItr::C"J�NCY IN CVCLE' P£R I SECONO

Fig. 5. Undistorted power vs. frequency Fig. 6. Frequency response, showing effects well above and below
Note the 1dB/div. amplitude scaling. the audio range. Note the 1dB/div. amplitude scaling.

break up on a peak, and for seconds converted to this output arrangement Another audible feature in the bass
thereafter the sound is distorted badly with gratifying results. This conversion rang'e is that the ampli fier does not have
because of poor recovery. In the ultra­ permits 30 watts of o ut pu t plu s the mort! b as s, but it has [mf/er bas s. Other
linear amplifier transient instability has other benefits inherent in the increased amplifiers, of good quality in terms of
been eliminated--changes in amplifier linearity of the output stage. mca�urements, by comparison were gen­
characteristics caused by overload do not erating harmonics and intcrmodulation
make the circuit unstable; and. lherefore, Listening 're*fs products. This was· also apparent on'
recovery is almost instantaneous. Most scratchy "dirty" recordings which
The m:1jority of listeners agree read­
fe edhackamplifiers fail miserably under cleaned up on the ultra-linear amplifier
ily to the superiority of this circuit .
overload listening tests. while remaining mushy and i rr i ta ting on
None felt that other equipment was bet­
Figure 6 shows the voltage gain ver� others.
ter although some could not recognize
sus frequency. Obviously, most present differences on the program sources used. In the treble region the consensus of
day amplifiers are flat through the audio However, during the course of the tests, opinion is that the amplifier sounds
band. However, it is the band outside of certain recordings were found which "smoother." The scrtltch level of shellac
the audible region which makes some of demonstrated differences' vividly; and records is less irritating while the high
the difference between one amplifier after this finding, even the less discrimi­ frequency sounds, particularly of a per­
and another. In this circuit it is evident nating listeners could identify the ultra cussive type, cut through the scratch
that smooth Rat bandpass extends more and seem far more prominent. This
than two octaves on each side of the seems due to the fact that intermodula­
traditional 20 to 20,000 band. This enor­ tion between scratch and mugic is di­

n n n
mous band width is necessary to elimi­ minished, and the two assume much
nate phase shift over the customary more pleasant proportions.
region and to provide good transient The authors believe that for sheer
20 cP' to Kc 30l(e
response. listening pleasure the ultra-linear ampli­
The square wave performance of Fig. PHASE _'5" PHASE _ .0· PHASE _·2" fier represents the best that can be
'I testifies to the transient response. Evi­ achieved at the present state of the art.
Fig. 7. Squue'w3ve performnce of the Ultra­
dently, a circuit with response flat within Others who have had an opportunity to
linear amplifier.
2 db for a decade on each side of the hear and try the circuit agree with this '
audio band should show a presentable and these beliefs will not ue shaken untij
square wave at most frequencies. How­ linear amplifier on "blind" tests and something comes along which' sounds
ever, the low phase shift, fast rise time, could recognize its superiority. better, or at least sounds as good and can
and insignificance of ringing in this Listeners agree tlmt the bass region is be built for lower <:;ost
circuit as ind i cated by the square waves more articulate. better detincd.:and bet ter
shows that more than just the frequency damped than in other amplifiers.' This
response is excellent. In addition, square damping is not a function of internal im­ PARTS LIST
waves were checked on a speaker lORd pedance alone but also relates to stahility C, 500 �u. 6 Y. eledro\ytic
with practically identical results, thus undt'r dynamic conditions. For example, , C,. C" C" C, 0.2 �f, 6OOv. paper
demonstrating that performance of the no low frequency cutoff is required in C. 100 Ill, 50 Y. electrolytk
amplifier is unaffected by a load of vary­ the preamplifier as no ill effects Rre C, 120 lA�f, mica
ing impedance. . audible due to momentary overloads C" C. 20-20 "I. 450 v. electrolytic
Other' circuit configurations can be from turntahle rumble, s w it chi n g CIf. ell 40-40 )1£. 450 v. ele<;trolytic
R" R•• R" R. 0.47 mcg.· Yz - watt
used with this ultra-linear output stage. thumps. anu similar disturbRnces. Cer­
R, 2(}() ohms. ¥4 watt
However, they should have a phase char­ tain types of signals such as organ pedal
R.I,R",R .. WOO ohms, Yz watt
acteristic permitting suhstantial feed­ tones combined with rumble wil! cause 0.1 meg. 1 watt; 5%
R" R,
back. and they should have the lowest other amplifiers to break up even at R, 0.56 mcg•. Yz watt
possible distortion for the early stages. levels as low as a few watts in the mid­ R" 560 ohms, Yz watt
The popular Williamson circuit ha s been freQuencies. R", R" 47.000 ohms, 2 watt, S%
RII, R" 0.12 meg, 0 watt
Rn 350 ohm�. 10 watt
R" 12.00J ohm!. Yz watt
R .. 22.000 ohms, t watt
e- R.. 22•• 0 ohllls; 2 watt

. -
60 la 7,000
tOO .. i.ooc DIt T, A croTO -3ro outPut trans­
former. Primary: 6600

·1 I 40 la 7,OOC
40 .It.ooc f}f/Z ohms plate to plate,
taJlJll·d for screen: se<;·
ondary : (G, 8. and "

'r.IM !
V:40a
r 2.000
Fig. 4. lntermodula­
tion distortion. using
sever�1 test frequen· T,
ohms.
Power transformer: 370�-
�V 370 v. at 130 maj 5 v. at
des with , CO"st,"t
2 ratio of 4:1. 2 2; 6.J v. at 3 2.

�V
L, Filtrr choke; 8,Hy at 130
t ma.
v, 651.7
� V. 65N7
o • 11 '2 III to 14 :!. V" V • 6L6
. WATTS OUTPUT V• 5V4

AUDIO ENGINEERING • NOVEMBER, 1951 77

Registered at the General Post Office, Sydney, for transmission by post as a periodical. Single Copy, One Shilling

Volume 20 May 1955 Numher 5

---._.--. -
--- --_.
-- .--- --
- -
- -
- -- -
-- -- �--------
---- -- .-,-. --
--.--
.-- --------_._---._._._-

ULTRA LINEAR AMPLIFIERS

First article of a series by F. LANGFORD-SMITH

and A. R. CHESTERMAN

This fir,st article gives a general introduction to sufficiently detailed to satisfy an amplifier designer.
the subject, with its history, some of its characteris­ The present article IS the first of a series to meet
tics, and then an investigation into the effects of the this need.
tapping point on the power output and distortion.
3. Description
The effects of high resistive loads are also
investigated.. A typical circuit diagram is shown in Fig. 1. It
1. Name
differs from pentode operation by the screens being
connected to tapping points T, T, on the trans­
former primary. 'It is always advisable to fit screen
The name has been the subject of much controversy. suppressors R3 and R4 and usually also condensers
"Tapped Transformer" and "Triode Tetrode" opera­ Cl and C2 between the plates and the transformer
tion have been suggested in England, while "Partial end of the screen suppressors, to eliminate a form
Triode Operation" has been used in the Radiotron of instability. Typical values are 0.001 or 0.002
Designer's Handbook (Ref. 1). None of these /LF and 47 to 220 ohms. Grid stopper resistors
seem to have taken on, so that we are left with the ( lOK) are also desirable.
title "Ultra Linear", which will be used throughout
this article in its abbreviated form, Ui. In this article, "pentode operation" is to be taken
as including beam tetrode operation.
2. History UL operation is a step between pentode and
triode operation. When the tapping points are
A claim has been made that the circuit was moved down to the B+ terminal, the valves operate
originated jointly by R. Lackey and R. R. Chilton as pentodes. When the two tapping points are
of the Australian Radio College and distributed to moved to the plate ends of the windings, the valves
students in the form of notes in 1933. Unfortun­ operate as triodes; when the tapping points are in
nately, the early records were destroyed by fire, and between, the operation is intermediate between
no written description has been seen by the authors. triode and pentode operation. Any desired tapping
If any one of our readers can produce the relevant points may be used - those closest to the B+ end
documents, we would be happy to publish the fact. give characteristics closely resembling pentodes and
then as the tapping points are moved towards the
The UL circuit (not, of course, under this name) plate ends the characteristics become more and more
was patented by the late A. D. Blumlein in 1937 like triodes, until the extreme case is reached where
(Ref. 2). It did not receive much publicity until full triode operation is obtained. It is usual to
1951, when D. Hafler and H. I. Keroes (Ref. 3) refer to the tapping point as so many per cent.
published a fairly detailed description under their of the total impedance, and this practice will
newly-coined appellation "Ultra Linear". Since be followed in the present series of articles.
then, it has attracted a good deal of attention and However, sometimes a tapping point is referred to
is now being widely used in place of pentode opera­ as a percentage of the total number of turns. One
tion in high fidelity and oth�r good quality ampli­ may be converted to the other by using the
fiers. Some of the many references are given at the relationship.
end of this article. However, no satisfactory treatise
on the subject has yet appeared with information Percentage tU1'ns = 10 V percentage impedance ( 1)
 54
For convenience, Table 1 has been calculated.
Table 1 - Tapping Point (from B + end)
Percentage impedance Percentage turns
0% 07'0 (pentode)
5% 22.4%
10% 31.6%
15% 38.70/0
18.5% 437'0
20% 4 4.7%
25% 50%
35% 59.2%,
50% 70.70/0
75% 86.6%
10070 100% (triode)
A complication with UL operation is that the
screen voltage swings around the applied voltage.
For example, with 250 volts applied, and a trans­
former tap of 20% impedance, the alternating screen
voltage will be 44.770 of the alternating plate volt­
age (Table 1). If the plate load resistance is 5000
ohms (single-ended) and the power Output is 5 watts,
the r.m.s. plate voltage will be about 158 volts and
the r.m.s. screen voltage will be about 70 volts.
Pig. 1. Circuit diagram of a typical UL amplifier.
The two most important effects of UL operation
are on power Output and total harmonic dis­
tortion. The power Output with UL operation crease in distortion. At the 200/0 tap the power
is always less than that with pentode operation and output is from 520/0 to 59% of the pentode maxi­
greater than that with triode operation, for constant mum power, and the THD from 0.4% to 0.650/0,
applied voltages. As will be shown later, the depending on the bias and load resistance.
maximum signal cathode current with UL is less (b) The optimum tapping point for highest
than with pentode operation, and the cathode current fidelity is between 17% and 200/c, and it is not at all
efficiency is approximately the same, so that if the critical.
plate voltage is increased to give the same input (c) The optimum tapping point for public
power for both cases, then the Output power will address work is probably 5%. This gives from 730/0
be the same for both UL and pentode. to 780/0 of the pentode maximum power Output,
4. Optimum bias and load resistance depending on the bias and load resistance, and from
Up to now all experimenters seem to have 0.5% to 2 % THD. Of course, if the plate and
assumed that the load resistance and bias for UL screen voltages can be increased, it might be possible
would be the same as for pentode operation. A to use the 200/0 tap for P.A. with no loss of power
graphical approach indicated that the optimum load Output.
resistance would be different, so tests were carried (d) Whether adjusted for minimum distortion or
out on type KT66 as a typical example. for maximum power, UL with 15% to 200/0 tapping
5. Type KT66, 300 volt operation points gives lower THD and higher power output
Many measurements were made on type KT66 than triode operation - see Fig. 2.
with 300 volt plate and screen supply, and those are (e) The optimum bias for maximum power out­
summarised in Fig. 2. Curve' 1, marked "Max. put is greater for UL than for pentode operation.
power output", shows power output against per­ It increases from - 28 volts (pentode) to - 30 and
centage tap, and each point is the condition of bias - 34 volts (UL) and - 34 volts (triode) - see Fig.
and load resistance giving maximum power Output 2. These values of maximum power Output are
irrespective of distortion. Curve 2 shows the power irrespective of distortion and would not be used in
output when the bias and load resistance are selected practice for pentode operation, since by allowing a
for minimum total harmonic distortion. On both slight reduction in power output there would be
these curves the load resistance and grid bias for a considerable reduction in distortion. However,
each point are marked on the curves. Curves 3 and the condition for maximum power output can be
4 give the THD for each of the two upper curves. used for UL operation, since the distortion is low.
It will be seen that the difference in power output (f) The optimum bias for minimum THD de­
between curves 1 and 2 is fairly small (less than creases from - 34V with pentode operation, to
0.5 db) and in most cases it is only necessary to - 26V with all UL tappings and for triode
consider the minimum distortion curve (2). The operation.
following conclusions can be derived for type KT66, (g) The optimum load resistance for maximum
for constant plate and screen voltages. power outp,ut is greater for UL than for either pen­
(a) Lowest distortion occurs at about the 200/0 tode or triode operation, increasing from 6K (pen­
tap. It is deleterious to go beyond this point, on tode) to 6K, 7K, and 8K (UL), and back to 5K
account of the reduction in pow�r output and in- and 6K (triode) - see Fig. 2.
Radiotronics May, 1 955
 55
26
�6K,-28V THD.
24 12%
\
'\--4'S K,-34V
22 11
�\\
20 \\ 1o
I � � 6K, 30Y
18
I '�'\..- 4.SK,-30V 9
I "-.�7K,-34V
I
16 I -7K,30V 8

14
=\=SK,-26V �8K.-34V !:-. 7
�8K,-34V
V) -\-- SK-26V"'-
,
I-
I- 12
......
� � 6
-cC
�
! 5 K,-26V .� �!..'lA "4X
10
I 6K,-26V -� �IVER Ou 5
I «)�O. F�
MIN.
_T� - 50r6K'-
r�
THD.--- -
8 I r-- r---� 4
� r--r-- '\
,
6 ,
3
5K,=?
4 2
\
I', (3) THD. FOR MAX. POWER --
" --- --- --- --- ---
2 G.
...--
...... (4) MIN. THD
",
-
..... -
-- - -
- --- -- - --- --- - --r --
•
- - --
I--
0
0 10 20 30 40 50 60 70 80 90 108%
PENTODE TAP TRIODE

Fig. 2. Effect of tapping point on power output 6. Screen dissipation

tlnd total ha'rmonic distortion. Type KT66 with
Eb .E(,2 300 V. Optimum bias and load re-
The static screen dissipation will normally be
slightly less than that for pentode operation, because
= =

sistance for each point.

of the increased bias. Since the screen makes some
contribution to the total power output, the dynamic
(h) The optimum load resistance for mInImum screen dissipation is not given by the screen input
THD increases from 4.5K (pentode) to 5K for UL power. The maximum signal screen dissipation
10% to 20�/o, up to 6K at 25 % tapping, and down for identical operating conditions in each case,
to 5K for triode operation - see Fig. 2. is highest for pentode' operation, lowest for
The detailed test results on which the curves of triode and UL is intermediate between the two.
Fig. 2 were based are shown in ,Fig. 3. The effect For this reason pentodes may sometimes be used
of load resistance on power output is quite small with UL operation under conditions when pentode
over the range shown, but its effect on distortion. operation would cause the screen dissipation rating
is very marked. The effect of bias voltage is also to be exceeded. The screen power relationships may
more marked on distortion than on power output. be expressed:-
Note that the minimum distortion point for pentode
operation is quite critical, both on bias voltage and Screen dissipation = screen input po wer-
load resistance (although not shown in Fig. 3),
whereas the 20% UL tapping gives almost constant po wer output (2)
distortion over the range of load resistances shown, It is usually not convenient to measure the screen
for any bias voltage between - 26 and - 30
power output and, so long as the maximum signal
volts. This is a very valuable feature.
screen input power is below the screen dissipation
Note the surprising low distortion possible with
rating, the valve will be operating well within its
pentode operation and a load of 4500 ohms-1.77'o
ratings.
THD at - 34 volt bias, but the operating point
is ,rather critical. Fig. 4 show's the screen current plotted against
The readings for distortion are only relative owing tapping point for 300 volts and - 30 volt bias.
to the difficulties in reading a THD meter at very The lowest curve shows the zero signal current per
low levels, and the distortion in the signal voltage valve, while the two upper curves show the maxi­
applied to the grids (up to 0.2%). There is mum signal screen current for 5K and 8K load
necessarily some cancellation and accentuation of resistances. It will be seen that the 5% tap reduces
harmonics due to harmonic phase relationships be­ the screen current to 57% and 55% respectively
tween the input source and those produced in the of the pentode value. Fig. 4 applies to a pair of
amplifier. The Appendix gives the detailed con­ typical valves, but wide tolerances occur in the
ditions of testing. screen currents drawn by all beam tetrode valves.
Radiotronics May, 1955
 56
7. Plate current curves of Fig. 5 there is undoubtedly some
Table 2 gives values of plate currents per valve, effect from in-phase and out-of-phase relationships
obtained by subtracting measured screen currents between the harmonics in the signal source and the
from cathode currents, at 300 volts and -28 volts UL stage, so that the e.xact sh.ape of the. distortion
bias. characteristic is uncertain, haVing a maximum pos­
sible tolerance of ±0.26 in the reading of the total
Table 2 harmonic percentage at any point (i.e. twice the
Plate Current
distortion in the source).
The UL 100/0 curve is given in Fig. 5C and has
Operation
I Zero Signal
I Maximum Signal the same general characteristic as for the 5 % curve,
Load 5K Load 8K but better linearity. It is practically linear up to
Pentode 4 9.1 7 1.5 5 9. 75 4 watts. The measured THD reaches a maximum
10% 4 8.1 63. 9 57 .3 of 0.77% between 8 and 14 watt�, taIling to 0.68%
(minimum) at 17.5 watts and tlSlng , to 1 % when
20% 47.75 61.0 56.5 .
Triode 44.5 5 0.9 48.7 the peak grid voltage reach�s zero. In practice, ":'Ith
a high impedance source, It would ?ot be . pOSSIble
to drive the grid to zero voltage Without introduc-
The differences between the zero signal plate cur­ ing distortion into the driver voltage.
rents are due to the resistance in the output inductor. .
The UL 20% curve is given in Fig. 5 D and IS
Table 2 shows that with UL operation the plate practically linear within the limit of accuracy o� t�e
current rise with maximum signal is less than that measurement up to 15 watts. The voltage gain IS
with pentode operation for the s�me operating con­ constant at 7.15 times right up to 15 watts. The
ditions, but the power output IS less, so that . no measured THD shows a steady rise up to 0.6% at
conclusion can be drawn.. Table 3 (below) gives 15 watts, being about one-tenth of that . for �entode
similar data in respect of the cathode current. operation. The whole performance IS qUIte re­
8. Linearity markable.
9. Efficiency
Linearity curves are given in Fig. 5, where 1; is
for pentode operation, and B, C, �nd D vanous ' cathode current efficiency for pentode opera­
The
tion was measured under the conditions giving maxi­
UL conditions. The THD at the gnds was reduced mum power output. Three different conditi�ns f?r
to 0.131"0 for Fig. 5 only. The linearity characteris­
UL operation were also measured, and are gIven In
tics were measured using peak reading valve volt­
Table 3.
meters for both grid and plate voltages, in order to
The pentode and the first yL condition .given
get a true result*. The disadvant�ge. of this pro­ above are the respective conditIons for maximum
cedure is that the power output indICated by the
power output given in Fig. 2. Th� second �L
valve voltmeter is appreciably less than the actual
condition is for the bias and load resIStance whICh
power output when the THD exceed� about 1 %.
give maximum power ou.tput with pentode operation;
As a consequence, i t has not been practICable. to a ? d
it is therefore not optimum for power output or
the power output readings to the pentode llneanty
efficiency with UL operation. The third UL con·
curve. The approximate power output may be
clition is intermediate between the other two UL
calculated by multiplying the power indicated on
conditions; it also is not optimum for power output
the curve by a factor of 1.1 for 5 % THD.
or efficiency with UL operation. . .
The pentode curves for maximum power output . Table 3 shows that, by selecting the UL condition
are given in Fig. SA. Note the "shelf" in the
giving maximum power output, the efficiency �s
distortion curve from about 15 to 19 watts output.
49.8% compared with 51.2% for the pent?de. ,!,hIS
The linearity curve deviates from the straight line
is quite a remarkable performance, eS eCla�ly sl?ce
the pentode output includes about 11 E7'0 dIsto�tI?n,
about 2 watts power output, and the slope of the
curve at maximum power output (input voltage 17.:
with the large third harmonic component giVing
volts) is only 34 % of that at low levels. ThiS
an effective increase in the power output for a
reduces the effective feedback at maximum power
limited swing (Ref. 12). On the other hand the
outp� . . UL output had only about 1 % total harmonic distor­
The UL 51"0 curve is given in Fig. 5 B. T?ls. IS
tion with a negligible effect on the power output.
practically linear up to 2.5 watts and the devIatl�n
is distinctly less than for the pentode, although ItS
Under the conditions for maximum power output
shape is rather unusual. The THD curve reaches a in both cases, the efficiency with the 10% tap was
maximum of 1 % between 9 and 13 watts, then slightly higher than for th� 200/0 tap.. .
Thus UL operation, uSing condltlons for maXI­
comes down to about 0.8 % (minimum) at 18 watts
before rising to 1.4% at zero grid.t Although this mum power output, will . give nearly t�e same
power output from . a speCIfied d.e. powe� Input as
form of distortion characteristic is generally unde­
sirable the THD at any level is much below will pentode operatIon. It merely reqUIres some­
that for pentode operation. In the distortion what higher plate voltage and somewhat less plate
current than for pentode operation. . .
1'The zero grid line is where the peak inpu t voltage
is equal to [he bias .
In addition, a specified valve, operating With
.. An ar ticle on linearity characteristics will appear in a constant d.e. input power, will give nearly the same
future issue. power output with UL operation as it will with
Radiotronics May, 1955
 57
pentode operation (each being operated under its with pentode operation, the plate dissipation will
own optimum conditions), and the total plate plus be greater by the amount that the screen dissipation
screen dissipation will be about the same. Hence, was reduced-in most cases this is only a very
�ince the screen dissipation with UL is less than slight amount.
Table 3
Type KT66, Supply Voltage, 300 Volts
Ik (mA) Power Power
Condition no sig. max. sig. input output Efficiency
Pentode 6K, - 28V 105 165 48 .8 W 25W 51.2%
UL 20% (1) 8K,-34V 60 99 29.6 14.8 49.8
( 2) 6K,-28V 108 122 36.4 14 38.5
(3) 7K,-30V 88 116 34.5 14.2 41
2"
KT66
E b': EC2= 300V
< Ecl=-30V
E
I I5 \
\
I­
Z \
UJ \
\ \
c::
c::
\ \8K
GI 0
\ "
z 51<' . "-
UJ •
r-...
UJ
c::
......
.......
U - .. �
Vl 5 . MALX. SIGNAL
- - -- -- - - - .51.<_
ZERO. SIGNAL BK

0
0 10 20 30 40 50 60 70 80 90 10070
PENTODE TAP TRIODE

Fig. 4. Effect of tapping point on maximum signal . These values appear to check reasonably well with
screen current for a single valve, for load 'I'esistances the gain reduction ratio derived from Williamson
of 8000 and 5000 ohms, and Eb = EC2 = 300· V. and Walker (Ref. 5) on the assumption that this
Zero signal screen current is also given. can be treated as pure voltage feedback_
1
10. Negative feedback Ratio = ------- ...... . ......... (3)
The UL amplifier has internal negative voltage x p.. RI-
feedback which decreases the total harmonic distor­ 1+-.
tion and reduces the output resistance and the gain, p..t rp+ RI,
when compared with pentode operation. Each of where ratio = 1 for pentode operation
these characteristics is intermediate between the X fraction of output voltage fed back
values for triode and pentode operation.,
=

to screen
However, the feedback cannot be treated as pure , 'percentage turns of tapping point
feedback, because it reduces the maximum power -:- 100
output and requires additional bias and higher load p.. = pentode mu
resistance for optimum operation. p..t = triode mu (screen connected to
The non-linearity known to exist in the screen plate)
characteristic is not the cause of these two pheno­ rp plate resistance of valve
menon, which are fundamental, although it must
=

and RI, = load resistance for single valve

have some effect on non-linear distortion. A further ! X load resistance plate-to-plate.
point in support of our conclusion is that the dis­
=

tortion with UL is not intermediate between pen­

11. Overload characteristics and distortion
tode and triode, but is lower than either.
Measured values of the voltage gain are The overload characteristics as observed on an
given below for KT66, Ebb EC2 = 300 V,
= oscilloscope differ fundamentally from those of pen­
ECI = -30 V and RI- 6000 ohms.= todes, although they bear some resemblance to those
Operation Gain Ratio of pentodes with a high degree of feedback. Pen­
Pentode .. .... .... ... .. 10.2 _ _ 1 todes show slight flattening of the peak some dis·
5% tap ..... ... ....... . 9.15
_ 0.9 tance before reaching maximum power output, the
10% tap .. .. . ...... . ..... 8.6 0.84 minimum visible being about 3 or 49'0.UL amplifiers.
209'0 tap . ... . ....... .. ... 7.7 0.75 without feedback show no visible deformation until
Radiotronics May, 1955
 8

a "fiat" appears at about 0.670, and as the drive is

N' o
increased this extends In a straight line chopping
off the whole top of the peak.
This method is very much more sensitive as an
indication of total harmonic distortion for UL than
for pentode operation,
UJ N
C
o a: 'i'
UJ 12. Characteristics deferred for a later article
a: �
f-
o
o C 'i'
a. :I:
co
(a) Input impedance (Miller Effect), see also
N Ref. 11.
IUJ
-oC)
N<
I� (b) Output resistance. This is intermediate be­
tween pentode and triode operation.
I� 'i' V)
(c) Non-linear distortion. Tests ofIM distortion
:;:� and measurement of individual harmonics to sup­
0:: 110
UJ
� o plement total harmonic distortion as used 111 this
o d 'i'
a. :I: article.
dl
7 (d) The effects of a loudspeaker load. This may
11
be simulated by a dummy load having a high
III impedance and a choice between 450 and zero phase
angle, in addition to the normal purely resistive load.

0::
(e) Effect of tolerances in the valve· characteris­
UJ
�
tics, especially screen current, on the optimum tap­
o ping point and other perfonnance characteristics.
a.

(f) Graphical examination of the composite

characteristics with load lines.
III

- I-H 13. General comments

o·
:I:
f-.
( a ) Compalison with pentode operation.
o
M
-. I
The p�incipal advantage of UL is that the distor­
tion IS so very much lower that the total circuit
-+{­ feedback can be less, leading to a higher stability
+F
:I: 0""
o· 0 margin or in many cases making possible tbe use
f- >O III
of a cheaper output transformer for the same

� ---I
1[; stability margin. Another important advantage 15
that the lInearity characteristic for the 20% tap
UJ
- 0::
o
� is practically straight up to the grid current point,
o
o
..J
::l _ cl \
M
I giving constant amplifier gain even without feed­
:x: _ ,
- f- .. co back. In this respect it appears to be as good as
1+ . N
I
triode operation, and results in the full amount of
- �� - -0
N UJ feedback being applied at all levels. Pentodes suffer
I C)
from drooping linearity characteristics and . only have
M ..J
1 0 about one-third of the full amount of feedback at
N>
M
high levels where it is most wanted.
:1-- I V)

o�
-0 'j'dl When compared on the basis of power output
:I:
:.:
co
"" co on a constant supply voltage (e.g. Fig. 2) UL opera­

o·
�
:.:
�� � �'l' tion shows a serious drop in maximum power output.
This drop in power output occurs because, on the
:I:�
f- III III positive peak of the grid driving cycle, the screen
7 voltage IS reduced below that which would occur
0::
ci UJ with the pentode operation, so that the maximum
:I: �
N
M
o
UJ
o I
o a.
f­ o
Z M
UJ I
0.. co
N
, -
•
1
Fig. 3. Po wer output and toted harmonic distoftion
o co �� � � � � ! � 2
versus g1'id bias voltage, f01' selected values of load
M N
S��VM �nd�no ij3MOd 1·esistance.

Radiotronics May, 1955

 350 350 59
I
A t- PENTOOE I-- UL 5'0 TAP
B I
I-r-f- 5K, -30V r--I-- 5K, -32 V
If THO. I THO
I 'JW -J,
.

V % ..;. 0
/0
300 If v- 300 rl 6
17-5W "'"'?-
6 I
I
I 1/ I
/ 15W J
I I 1/ I I
250 '( 5 250 1.l12.5W I 5
....., . / r"'
-. Cl. If 1..1
Cl.
I 1,( IOW ( I
I I Cl.
Cl.
I
VI
200
.. I
I
--
4 1J)200 /
�7·i'll
�� f-
�
c-
4

i�
, I-
I- ...J
...J c-
0
o
>
JI/ > I c-
I '/ :-
� 150 f) SW
--

� 150 I I 3 3
Cl. IJ TH . lA I
I­ '/ -r-"" I-
::::>
I
:::I
I 1/ I 0 ,
o
( "
0 2-5W'./ I
100 J er. 2 iOO 2
'( 1 Cl

1 I 0
I
11 er.
LIJ
N
I1 1/ THO_
50 SO
I. - l/
1 1/ - � f-- �
f-I�f--"
._ -

l..I -. -
t- - -
-c--
i""" c-
III V-
I ,......
-

o I o o o
o RMS 5.0 10.0 15.0 20.0 . 25.0 o RMS 5.0 10.0 15.0 20.0 25.0
o Peak 7.07 14.14 21.21 28.28 35.35 o Peak 7.07 14.14 21.21 28.28 35.35
Input volts - each side Input volts - each side

350 350
I 11 I I
C1- UL 1 0 % TAP I D
t- UI, 2 0 % TAP I
I-r-t- 6 K, -32V 17-5W F.oo;� I t-r-t- 6K,-30V - -

THO. c-r-f-
I THO.
11/ 1
I
% 5WY
%
300 . 15W r- 6 300
/
6

r- If'L
�,(·12·5W I -SW
I /
1 1
250 5 250 5
'-),10\11 I OW�
- -/i· 1-1-
�
.......
Cl. L. -- -- -
Cl.
I H-I-
-- 1 c-
I '/' j
'-' '-'
Cl.
7-5W
If,h5W
Cl.
. �

11

�i
Vl2 00 /. -
4 tIJ 200 4
I-
� c-
I-
...J
/
...J /) 0
o S VJ
c-
> SW �
> r-

� 150
-1- -

I/
, �� r-
3
I-
� 150
1/ 0
3
1.1
-

er.
Cl. c- Cl
I­ I I-
::::> 1/
:::I
2·5W,c I 0 2·5W;C 0
er.
o --

-- -- -- --
LIJ
N
100 I1 2 100 I 2
--
I
-f- -

1 - --
'-c-
I-
1/
-f-
I--�
--

50 THO. 50
-f- � --
-f- -- - --

"" - --
-I..... / THO.
I1 -
....
--

- -I-
-I c-c-
f-f- ....
,
- -- .-- - -- --

I
-

o o o o
o RMS 5.0 10.0 15.0 20.0 . 25.0 o RMS 5.0 10.0 15.0 20.0 . 25.0
o Peak 7.07 14.14 21.21 28.28 35.35 o Peak 7.07 14.14 21.21 28.28 35.35
Input volts - each side Input volts - each side

Fig. 5. Linearity cttrves, with total harmonic dis· B. UL 5%, E,., = -30 V, R" = 5000 ohms .
tortion, for KT66 with EIJ Ecz 300 V: C. UL 10%, ECl -30V, RI, 6000 ohms.
'
= = = =

A. Pentod e ECI -28V, RI, 5000 ohms.

= = D. UL 20ryc, Ec , = -28V, R,. = 6000 ohms.

Radiotronics May, 1955

 60
OSC I LL ATO R l oo n

1,000 C/S

LOAD
R E G U L AT ED ....+........
.. oooo4
...
r--go, l I L t. R E S .
VO LTAGE
RL
4'S K to
lOOK

Fig. 6. Circuit used for deriving test 1'esults pub­ ( c ) Operation.

lished in this article. No te that the The point for mml1DUm distortion is given, at
0.001 j1-F condensers are in different positions to least for type KT66, by a tapping point from 1 7 %
those shown in Fig. 1. t o 20% impedance, and the optimum grid bias and
load resistance will both be greater than for pentode
signal plate current i s reduced. However, the effi­ operation on the same supply voltage. Comments
ciency may be made to be almost the same as for on increasing the screen voltage have been made
pentode operation, and, circumstances and valve
in (a) above.
voltage ratings permitting, i f the UL amplifier is sup­ When used as a public address amplifier with the
plied with the same power input as the pentode, S % tap, the loss in power output is only about
by increasing the UL supply voltage, approximately 2 0 % , and yet there is an outstanding reduction in
the same power output will be given by both and non-linearity when compared with pentodes.
the combined plate and screen dissipation will be
roughly the same. 14. Appendix
UL operation not only reduces the distortion, but 14.1. Test conditions
it makes the load resistance less critical over the The test circuit used for these ' measurements is
'
limited range used in these tests. It is hoped to shown in Fig. 6. The oscillator harmonics were
investigate this matter more fully at a later date. reduced by a filter, and the amplifier non-linear dis­
The screen dissipation is less with UL operation, tortion was reduced by the use of a large amount
so that conditions of operation are sometimes per­ of feedback, so that the total harmonic distortion
missible with UL which are not permissible with at each grid did not exceed 0.20/0 (0. 1 3 % for the
pentode operation. The only disadvantages of UL linearity curves). The in put transformer T 1 was
compared with pentode operation appear to be the made with a C core and an inductance of 100H, with
slight loss i n voltage gain, the reduced power output a set-up ratio of 1 : 2 primary to whole of secon­
( unless the supply voltage is increased) and . the dary. The leakage inductance was 1 6mH primary to
slightly increased cost due to the tapped transformer whole of secondary, and . 36mH primary to half
and the stopper condensers (although the latt�r are secondary.
sometimes used with pentodes).
( b ) Comparison with triode operation. A tapped inductor 11 was used in preference to a
UL operation has higher power output, greater transformer, tapped at S , 1 0, I S , 20, 2 S , 3 S , S O
efficiency, less total harmonic distortion and requires and 7 S % o f the impedance o n each side. Its in­
less driving voltage. When overloaded, it suffers ductance plate-to-plate was 3 S 0H at low level,
from a particularly flat form of flat top even without S OOH at 240V S O c/s, with leakage inductance from
any external feedback. one half-winding to the other half 1 0mH. The
Operation on a loudspeaker load has not been d.e. resistance was 440 ohms total.
covered by the present series of tests, but it is hoped The large number of taps and switching facilities
to publish further results in the near future. However, gave a tendency towards instability, which was
a graphical check has been made on the only valve avoided by the use of grid, screen and plate stop_ ·
type for which UL characteristics were readily avail­ pers, and by capacitors of 0.00 1 j1-j1-F between plates
able (type S881) , leading to the conclusion that and screens.
high load resistance will cause high distortion. If
this is substantiated i n later measurements, i t leaves The input voltage . was measured separately on
the triode output ' stage .still as the ideal when its each side of the secondary T1 , using a valve volt­
performance on a loudspeaker load (and nOt a fixed meter, but the values were almost identical. An
resistive load) is considered. However, this must oscilloscope was connected in turn from each grid
be at the expense of larger power supply, and to earth and the side showing earliest signs of
larger input voltage. " flattening" was used as an indicator. The input

Radiotronics May, 1 955

 61
was decreased until there was no visible deforma­
tion of a greatly enlarged waveform peak.
The output voltage, which was used for deriving
the power output was measured with an Avo Model
8 rectifier type meter. This is the type of instru­
ment commonly used in power output meters. This
was checked for accuracy against a Western dynamo­
meter type voltmeter. ' For pentode operation the
error was between + 2% and + 3.5 9'0 from 100
to 240 volts, falling to zero from 2 70 to 300 volts.
For UL operation the error was .about "+ 2 % from
100 to 230 volts, zero at 250 volts, -1% at 2 70
,fOltS, a n d zero again at 300 volts.
Before measuring distortion, the input and output
voltmeters and the oscilloscope , were disconnected,
since they had an appreciable effect on the distortion.
The linearity tests were carried out with r.m.s.
calibrated peak-reading valve voltmeters in both
input and output circuits.
The supply voltage for the present series of tests
was approximately 304 volts, giving very close to
300 volts between positive electrodes and cathodes.

15. References

1. Radio tron Designer's Handbook, 4 th ed., p. 5 7 0.

2. Blumlein, A. D., Bri t ish Patent No. 496,883

( 5 / 6 / 1 9 3 7 ) . U.S. Patent No. 2 , 2 1 8,902 ( 2 2 / 1 0 / 1 94 0 ) .

3 Hafler, D., and H . 1 . Keroes, " A n ultra-linear am­

plifier". Audio Eng. 3 5 . 1 1 ( Nov., 1 9 5 1 ) , 1 5 .
4 . Hafler, D. , and H . I . Keroes "Ultra-linear operation
of the W illiamson amplifier ". Audio Eng. 36.6 ( June,
1 9 5 2 ) , 2 6.

5 . Williamson, D. T. N . , and P. J. Walker, "Amplifiers

and superla tives-an' examination of American claims for
improving linearity and efficiency", W.W. 5 8 . 9 ( Sep t.,
1 9 5 2 ) , 3 5 7.

6 . Whi te, S . , "The White Powtron Amplifier". Audio

Eng. 3 7. 1 1 ( N ov. , 1 9 5 3 ) , 3 2 . Uses ultra-linear amplifiers,
with 2 channels, 20 wat ts main amplifier, with frequency
dividing ne twork prior to bo th amplifiers. No distor tion
figures quo ted. Employs mainly nega tive voltage feedback,
but also small degree of negative curren t feedback; this
is claimed to eliminate "power dis tortion" caused by
variation in loudspeaker impedance.

7. Frieborn, J. K., "High quality circui ts". Radio Elec·

tronics 24.9 ( Sep t., 1 9 5 3 ) . I ncludes Brociner U L - l
'( ultra-linear) , Bell 2 2 0 0 ( combined plate and cathode
loaded ) , Bogen DB 20 ( combined plate and cathode
loaded with dis tortion curves ) , and S tromberg-Carlson
A R - 4 2 5 ( pentodes with overall feedback-with distortion
curves ) .

8. Marshall, J., "Junior Golden-Ear Amplier". Radio

Electronics 2 4. 1 1 , ( Nov., 1 9 5 3 ) , 5 5 . Modified ultra-
linear with push-pull 6V6.

9. HaIler, D., "Ultra-linear operation of 6V6 tubes".

Radio and TV News, 5 1 . 6 ( June, 1 9 5 4 ) , 4 3.

1 0. Ayres, W. R., "Audiology". Audio Eng. 3 7 . 5 ( May

1 9 5 3 ) , 1 4 , especially pp. 68-69 and Fig. 4'.

1 1 . Marshall, J., "Ex tending amplifier bandwith",

Radio Electronics 2 4 . 1 0 ( Oct. 1 9 5 3 ) , 4 0.

, 1 2. Radio tron Designer's Handbook, 4 th ed., p. 564,

second paragraph.

Radiotronics May, 1955

♦ Verso Filler Page ♦
Registered at the General Post Office, Sydney, for transmission by post as a periodical. Single Copy, One Shilling

Volume 20 June 195� Number 6

... _--. .... ..... ... _ ..... ... ... .. _-_ .•.••...... ... ......... _----

SQUARE ' WAVE TESTI NG

by F. Langford-Smith and A. R. Chesterman

1. A square wave generator. capacitance on the high frequency performance, but

at 20 Kc/s, on our A.W.A. Model A56031 oscil­
In selecting a circuir for development and test
loscope, the difference was hard to see. At fre­
it was necessary to choose between squaring a sine
quencies greater than 20 Kc/s it would be necessary
wave and some form of multivibrator. Both methods
to use the 200 ohm resistors.
are capable of giving good results, and the cathode­
controlled multivibrator was selected for first trial ( 3) Radiotron Type' 12AT7 was used as multi­
as being comparatively simple and inexpensive, vibrator in place of type 2C51 in the original
while capable of giving a very rapid rise time. circuit, for greater availability and essentially the
The circuit first tried out (Fig. 1) is that described same performance, thanks to the two 6AQ5 pentodes
by G. W. Gray of the RCA Laboratories Division used as cathode load impedances. Thus the same
in Radiotronics, Vol. 17, No. 8 ( August, 1952), valve type is used in the multivibrator and the
page 13 3. cathode-follower Output.
This article is well worth reading by those ( 4) The capacitances of the frequency controlling

50 c/s to 1 Mc/s. In our case the desired range .

interested - the circuit gives square waves from capacitors are derived from Fig. 2 of Gray's article,
but the graph is so small that it can only be read
is from 50 c/s to 20 Kc/s. The rise time is stated with an accuracy of about 20%. There is additional
to be 0.05 micro-second, which is completely in­ inaccuracy from the tolerances in the capacitors them­
visible on an oscilloscope, being one five�hundredth selves, so that quite a large total error may be
of the width of the flat top at 20 Kc/s. expected. In our case the frequencies before cor­
Some of our readers may q.uestion the omission rection were:-
of grid resistors in the multivibrator, and they are
0.004 fJ·F 28.1 Kc/s.
referred to the original article for a full description.
The operation is fundamentally sound since the 0. 005 fJ-F 11. 5 Kc/s.
plate currents are limited. '0. 02 fJ-F 5.4 Kc/s.
Certain minor changes were made in Gray's 0.1 fJ-F 1140 c/s.
original circuit diagram (his Fig. 4 on page 13 5) :-
1.1 fJ-F 57 c/s.
(1) A pair of 6AQ5 pentodes were used in place These may all be lowered to the correct frequency
of cathode resistors on the multivibrator valves to by adding a shunt capacitance of suitable value. Of
provide constant current, and so to eliminate the course, for most applications, a slight error in
slight tilt in the negative half. This is the modifica­ frequency is not very important ..
tion suggested by Gray (his Fig. 3), except that
we used a separate cathode bias resistor for each Operation.
pentode, one ,being variable. Any lack of symmetry
The output voltage is about 18 volts peak-to-peak.
between the positive and negative pulses can be
The wave form at the higher frequencies was very
adjusted by altering the .variable 5K Q cathode
good.
resistor. Also, when the multivibrator valve is

and bottom tilt 1 0/0. In addition, there was a slight

The 50 c/s waveform showed a top tilt of 2.5%,
changed, the variable cathode resistor may have to
be adjusted.

to an additional 2.5% of the peak-to-peak vol­

rounding at the far end of the flat top amounting
( 2) The 12AT7 cathode follower load resistances
were changed from 200 to 2200 ohms, to inc,rease tage. This performance is entirely satisfactory for
the output voltage. This increases the effects of measuring overshoot and recovery time in amplifiers.
 66
SQUARE WAV E GENERATOR

+3S0V
l,ooon 3W 27mA

IOpOO.n
IW

'ii

MULTI-VI BRATOR

20 KC 'OOS]JF
0---1
10 KC '008f-lF

5 KC
0---1
�
o---i ' . 11 f-l F 'IMn

I�
5 0 c.p.s. '·2,.. F
o---i
6A05

SYNC.

INPUT

-
·005fJF
•

CATHODE FOLLOWER L-.-.t-

12AT7
--

OUTPUT

-
•
-
•
-
•
1-
-
•

Fig. 1. Circuit cliagt'am of sq�tare 'Wave genet·ator.

Radiotronics June, 1955
 67

with overshoot greater than 5 % requires a band­

The rise time is so rapid ( 0.05 tJ-see. ) that it is a A rise time of 0.05 microsecond in fln oscilloscope
more severe test than is normally used for testing
amplifiers. The 5 0 cls top and bottom tilt simply width of
mean that with extremely high fidelity amplifiers 0.45
it is necessary to make a comparison between the = 9 Mc/s.
input and output waveform. 0.05
The oscilloscope used in our RadiotroniCs Labora­
All 12 AT7 valves do not work equally well as
tory is the A.W.A. Model A56031, which is fitted
multivibrators in this circuit, and it is suggested
with a vertical amplifier designed and factory­
that the two 12AT7 valves be interchanged and adjusted to give good square wave response. The
left i n the position giving the better 50' cls wave­ push-pull amplifiers are neutralized to control the
form. Finally, the output may be taken from either
overshoot and gai n at high frequencies. When
cathode of the push-pull cathode followers, and here
neutralized correctly they give no overshoot on
again the better side may be selected., . '

, square wav ' es.

The performance at the htgher frequencies is not
In order' to check an oscilloscope, connect the
much affected by interchanging valves. square wave generator, working at 20 Kc/s, directly
We plan to develop a more elaborate squa �e
to the oscilloscope, Llsing short leads. The generator
wave generator some time in the future for use In should be allowed pknty of time to warm up and
our own laboratory, but we honestly feel that for settle down to steady operating conditions. The
most applications this simple circu it is all that is effect of the volume control in the generator on
necessary. the Output waveform is practically nil, so it may
be set to any desired position. On the other hand
2. Oscilloscope for use with a square �ave the stepped attenuators in most oscilloscope vertical
·generator. amplifiers have quite noticeable effect on the wave­
It is important to have an oscilloscope with a form, and this effect should be minimized by setting
vertical amplifier with response up to at least 500 it to the h ighest gain position ( i.e., no attenuation ) .
Kc/s, and without any overshoot in itself. The , In all cases it is w ise to check the variations
great majority of conventional oscilloscopes in use in square wave response with different stepped
in Australia do not give any indication when over­ attenuator settings. If the variation� are noticeable
'
shoot commences,' and are useless for such measure­ it is advisable to seiect the attenuator position
ments owing to their limited high frequency giving minimum overshoot - probably zero atte n�la­
.
response. A 'push-pull vertical amplifier is �nuch tion in most cases - and to use the same pOSItIOn
preferable to a single-ended one, but there IS . no for all square wave tests wherever possible.
.
necessity for it to be a d.e. amplIfier. There IS a 3. Measurements with square waves.
relationship between the rise time and the ban? ­
Square waves are usualJy treated as a special form
width of an amplifier, which may be expressed ' 111
of pulse ( Fig. 2 ) , which rises from a base line,
the form ( Ref. 1 ) :
the height of the pulse being the total peak-to-peak

rB = 0.35 to 0.45
voltage ( AB ) . This differs from the conventional
(1) sine wave approach, and all voltage measurements
( 0.35 when the overshoot is lass than 50/0) are peak-to-peak.

= rise time in microseconds from 100/0

The most useful measurements to be made on
where l'
feedback amplifiers are described below. These are
to 900/0 of the step function response
of a low-pass amplifier without over­
normally carried out with a resistive load, or with
both a resistive load' and a loudspeaker load, since the
shoot, loudspeaker may have a noticeable effect at certain
and B = bandwidth 111 Mc/s for response 3: db frequencies. If a resistive load alone is used, it
down. should be shunted by the value of capacitance which
will give the. maximum overshoot. This, value may
For instance, in the case of an oscilloscope 3 db be determined by some form of gecade capacitance
down at 0.5 Mc/s, with overshoot less than 5 0/0, box, going upwards in capacitance steps to the
the rise time is limit suggested in Table l.

=
0.35
= 0.7 tJ-see.
Table 1.
l'
Shunt capacitance to give 600 phase angle at
0.5
20,000 c/s.
An old type oscilloscope, 3 db down at 50 Kcls
(= 0.05 Mc!s ) , with no overshoot, would have
Impedance' ( ohms) ( Capaci tance ( tJ-F )
2 .0 7
a rise time 3 .0 5
4 .0 3 .5
0.35
l' = 7, tJ-secs.
6 .0 2.5
8.0 2
0.05
10 .0 1 .5
Note that 7 m icroseconds IS equal to 28 % of the 12.0 1
flat top at 20 Kc/s. 15.0 1

Radiotronics 'June, 1955

 68
B r-------, E -

A G H
BASE LINE

Fig. 2. A square wave shown as a rectangular pulse Fig. 4. Rise time for a pulse or square wave .
rising from a base line.

�C J
OVERSHOOT
E K B
TB
RECOVERY
TIME
F

�--H

G
Fig. 3. Square wave response of an amplifier showing
- Fig. 5. Square wave response of an amplifier
- showing
overshoot and damped transient oscillation . top and bottom tilt.

An amplifier should be sufficiently stable to be quency response the greatest value of overshoot
used with any likely form of loudspeaker crossover occurs with a square wave frequency of 5 Kc/s,
network, most of which present highly reactive while with very wide range amplifiers the greatest
loads at certain frequencies. A case is known where overshoot may occur with a square wave frequency
a Williamson type amplifier was stable on a single of lO Kc/s. The test frequency chosen should
loudspeaker, but unstable on a dual system. normally allow a reasonable amount of flat top
to show. With care in design, it is possible to keep
overshoot below 5 0/0.
It is hoped to carry out tests in our laboratory
with various amplifiers under all practicable loading
conditions, and to cover the subject at some later
date by another article in Radiotronics. 3.2. Transient recovery time.
This is the time in microseconds for the damped
3.1. Overshoot. oscillation following the overshoot peak to die

The overshoot (BC in Fig. 3) is measured as

down - see Fig. 3. In well-designed amplifiers the
recovery time should be a small fraction of the
the distance moved beyond the flat top, Ha. It flat top at 5 Kc/s. Arbitrarily, we m.ay take one
is expressed as a percentage of AB, that is the quarter of the flat top at 5 Kc/s, equivalent to 25
peak-to-peak voltage. It may be measured at any m}croseconds, as the maximum permissible recovery
desired frequency. The overshoot is theoretically time.
constant at all applied square wave frequencies, with
an amplifier flat from zero frequency to infinity.
3.3. Rise time.
However, the overshoot peak occupies only a very
This is usually measured as the time in micro­
seconds for the wave to rise from 1 0% to 900/0
small portion of the half-cycle at low frequencies,
and is therefore difficult to observe. Our usual
practice is to use a square wave frequency of 5 or of the peak-to-peak value - see Fig. 4. The relation­
1 0 Kc/s, where the overshoot may readily be
_

ship between rise time and bandwidth is given in

observed. Usually, in amplifiers with restricted fre- eqn. ( 1 ).
Radiotronics June, 1955
 69
3.4. Top tilt. 3.5. Bottom tilt.

If the tilt of the lower half-cycle differs from

This is a measure of the low frequency charac­
that of the upper, both values should be stated.
teristics of the amplifier. It should be carried out

is expressed by the ratio of the "sag" (EP m

at the lowest available frequency - here 50 c/s. It References.
1 . G. E. Valley and H. Wallman, "Vacuwn Tube
Fig. 5 ) to the peak-ta-peak voltage, expressed as
Amplifiers", M.LT. Radiation Series, McGraw­
a percentage.
Hill Book Company. Section 2.

The measured top tilt at 5 0 cls on the 4.5 watt Table 2.

( February, 195 5 ) , was 8.5 % of which 2.5ro was

single-ended amplifier described in Radiotronics Frequency Length of flat top
( c/s) ( micro-secs.)
due to the square wave generator and the balance 50 10,000
of 6ro was due to the amplifier. This is considerably 1,000 500
better performance than that of a: typical radio 5,000 100
receiver, but is not so good as can be achieved 10,000 50
in a push-pull h igh fidelity amplifier. 20,000 25

Radiotronics June, 1955

 70

"ULTRA LINEAR" AMPLIFIERS

b'y F. Lcmgfo·rd-Smith cmd A. R. Chesterman.

This article treats 6V6-GT operation and makes a comparison between UL operation and the divided
plate and cathode loading method used in the QUAD II and other amplifiers.

1. Typ e 6V6-GT UL operation. given by the other figures, indicating a bias of

-2 1 volts as optimum for power output and giving
Fig. lA shows the power output versus. tapping reasonably low d istortio·n.
point for plate and screen 2 85 volts, and bias � 19
volts, for selected values of load resistance. Fig. 1B The load resistance of 8000 ohms plate-to-plate
shows the total harmonic d istortion. was selected as optimum, giving an output of 10.4
Figs. 2 and 3 show the same inforrhation for watts at 0.7 2 % THD, even though an output of
bias values of -2 1 and -2 2.5 volts. From these 11.2 watts was obtainable with a load resistance of
it appears that the 5 9'0 tap is the best all-round 10,000 ohms. The 'reason for the choice is partly
compromise, giving minimum distortion for -21 to make it less sensitive to increases in load resis­
volts bias and 8000 ohms load resistance. tance such as. always occurs w ith a loudspeaker load,
Fig. 4 shows the power output and distortion and partly to make the transformer simpler and w ith
against grid bias. These confirm the impression fewer primary turns.

7�o
6V6GT
\
A B
Eb = EC2 = 285V THO.
EC1 = -19V
UL Operation 1I-'
15 6
\ ,
'\

-- -
I- -
1-. 1--i_ --

14 i· -i-
I- --f-;-�-
- -

,� \
f- f.-
I \"
\
13 \�\
- .. -

1\
\
I

12 1\\ '. \\
5 T
1--\.- --1- ,

i
- -

11 \ \
,--

1- -\ __ L
I
--.

�1- -
�::: t-= -\ I -1L..
_\ _ -- ._.
1- --

�,
- - - ---

0, \
\
. --1'--1-, \ I-

101< SK 7K 20K
- - -- -

\ 1\
10 ,
4

:� ,,� \1 )
,
IL-/- .. f- I- i- f-- -
/\
n- -1-r- �\+
-- - --- --

" 'i.)<
9 l- -f- -1- - - I- t--- l-t--. l-f-
/\ 'I, " /
-

t--.:�
.J

8
l'- r- I--
: .�
" �"
At--It. -r-.. f":1:::.
....

" 3
�
I

·i le K
I-Tf f-

20K
5K61< t-- l- t-t.:::'� " ,
; \
7 I \
p... r- r- f::::: �:::. �.1- f- �--1-
6
I-I- - .

i
1--
-

. \-- :_+---
r-c� �
..

\
_.

5
1\ /'
. .

\
2
1-I- \
8K
- --

\ \
4

3
;��
RI-'-- ,
\
1

i
, 5K 6K 71<
\1 I
•
i ,

r- h.. "- \ "

2 ,
�
-h.,l- K
8K 20K
5K 6K
. ...
,,--

1--I-
10K
15%
o
5'0
1..-1.- o
Ptnt. 10% 15'0 20�o Ptnt. 10%
Tap FIG,I Tap

Fig. 1. 6V6-GT pttsh-pull UL operation, Eb = E"2= 285 V, ECl = -19V, peak grid voltage eqttals
bias; (A) Power Otttput vemts tap; (B) Total harm01zic dist01"tion ventts tap.

Radiotronics June, 1955

 71

\
6V6-GT
5V
A \ B
Eb=£C2= 28 ,
IV
ECI =-2
15
UL OPE RATION
\
6
\
\
\
1\
14
10K
�
'\ .� 81( I

I\. \ \' /7t< I

13
- -- , .l \
5
12 \ �" / 6K 1 \
:
-"" \'x / \
1', � \ .)
'. ,
! 101< I

i
11 ...... "
..j \\ i '\
'\ '\ \ " '- . I \ 201<
\V � .'" 10K 8K 7K
10 :

L
"- 4 \

I '" � ."- . ..... ' 1 11 i

......
I
·1 " I 1
20K � " :--... ......." j
9 I
1' .... . THD.
5K i'" ' I'-'t"-- . .... 'j .....
-c--

r--- '� -. ..... ....... 1 r:,::: . .....

....... I \
:

r--.. � ..:::: :::::.. r--t- ' ....... t---...J..�

...... 3
1 \
/ 1f -I-- ��� \
61\ 5K 20Kv ---�
\
1-
\
--
\
:
I
\
5
\

'l
;
2
\ \
1
81< \ 1\
5K I< 7K
4 I - --
\1 \1 \
I 1
[7,1\ \ '\
3
� ....-'�
,
I I ; .'
i--I- - . -

.�
.- , ,:,
1
2 V� , I-...- - . -. -

'�F.L.:l:"- .:
1 .... - -.
; ,
� \ - -
"/ / I

61< 5t< I 81< IOK' I'

1 !

o o
5 01 10
0 10
0 0,
Pent. 15% 20% Pent. 10 15 20%
.Tap
FIG. 2.

Fig. 2. 6V6-GT pttSh-pttll UL operatio1Z, Eb Ec2 285 V, ECl -21 V, peak g1'id voltage eqttals bias,.
(A) Power Otttpttt verms tap,. (B) Total harmo1Zic distortio1Z verstts tap.
= = =

If high power output had been unimportant, a published value (-19 volts ) .
tapping point of 15ro or 20% might have been The curves and other data are for operation in
selected to make the load resistance less critical for the conventional way with the peak signal vcltage
distortion. However, this d istortion is less than that equal to the bias ( "zero grid" ) . Of course, in any
with pentode operation. The whole question of practical amplifier, it is not possible to drive to
d istortion w ith high impedance loads, such as occur zero grid w ithout grid current and consequent dis­
w ith loudspeakers, will be treated in full detail tortion, and the consequent reduction in effective
in a later article in this series, and comparisons made power output applies to- all types of operation. The
between UL, triode and pentode operation. output resistance u nder these cond itions without
The selected condition gives 71% of the output external feedback is 15,400 ohms plate-to-plate. With
obtainable with the same valves as pentodes under the 20ro tapping, this figure would become 9,600
the same condition s' except that the bias is the ohms.

Radiotronics June, 1955

 72
Conditiom for fixed bias ( values are both valves) : Supply voltage ............. . . . . . . . 306 volts
Supply voltage ( plate to cathode ) . .. 285 volts Cathode bias resistor ( each valve ) * . . 430 ohms
Bias voltage .. .. . .. . . . ... . .. .. . .. -21 volts Plate to cathode voltage ( max. s ignal) 285 volts
Load resistance (P-P) . .. .... . .. .. . 8000 ohms Grid bias ( max. signal ) . .......... 20.2 volts
Cathode current: zero signal 58 mA Load resistance (P-P) 8000 ohms
max. signal 85 mA Cathode current: 'zero signal 83 mA
Power output 9.8 watts max. signal 94 mA
Total harmonic distortion . . .... . ... 0.8 % Plate current: zero signal 75.4 mA

Total input power ( based on cathode max. signal 82.8 mA

current) , max. signal . . .. . ...... . 24.2 watts Screen current: zero signal 7.6 mA
Efficiency (based on cathode current) . 40.5 % max . signal 11. 2 mA
(Compare this with the tetrode efficiency 46.5 ro) Power output 9.6 watts
Conditions for cathode bias ( values are for both Total harmonic distortion ... . .... . . 0.98 %
valves) :
Total input power (based on cathode
" With each cathode resistor adequately bypassed. Alter­ current), max. signal .. .... .. . . .. . 26 .8 watts
36 %
natively a common cathode resistor of 220 ohms could
be used. Cathode current efficiency . . .. . . ....

15

HA -- I

14
6V6-GT
1 B
Eb= Ee2= 28SV

+
I
1--- - --. - r--I- --I-I- THD
.\
ECI =- 22·SV \
13
UL
f- - -c-l- 1
-i 1\
5
8K
�\
12
'.
201<
- f- _..
1\ 11.
5K
"-
\:�
K
\ ["
" ! '\h.
10 4

. ....
"- \. I

\
1

" f".. r-- .... . t- .:- i

'\ i'.. ."- "-
\
BK
9 :'-
.... ., --. -..
1
� 61<' i"--�-....r---
..
-
��:...... 7K t-l-- \
\
8
i"--t--.
- I.:::,.
-t-I-: J
J� 3
�11\
\1 \ \
I-"r--:: 7K
Ii- 6
1 i

i-- SK
i- t- -!- : K
20K �.
\�1\ ( K
\
I
6

5
I
2 \�\ �. K \
5K
I
I I�\\' I-L. I
f-f- _ .
\ f->-I-
- \' �7K
1 1\ ["
4 - - f--- - 8K
t
10K --
IOK
8Kt"\ \
3 I
� ......
2 . 'to.. 20K
I

--- r-.. 1--

I
f-

5% 10 % 15% 20% 10% 15%

FIG. 3

Ftg. 3. 6V6-GT push-pull UL operation, Eb EC2 285V, (Eel -22.5 V, peak grid voltage equals
bias; (A) Powet· output verstts tap; (B) Total harmonic disto'f'tion versus tap.
= = =

Radiotronics hme, 1955

 �
Q
-
;...
....
o 6V6-GT
.... A B
12 Eb=Ec2=285V
g
....
(")
UL 5 % �ap
THD,
Cl)

"6% SK
10K
7
- ' 1-- f7
" 1'5
lOK /
/ -- -- i--- -- / I
81< 1·4
// [.....-- 1/ � 6K
-
7K
-
/ /' ...I'- -- . 10K
1·3 /
10 �-; /�I< -;.7' / '/
// / 1'2 // 17 --- 7K
/ / / // I THD,
6K
/7K V / li/ / v-- 8K 9%
H
/ /' -- -- >-, .- I
20K A //
/ I2,OK / // /
9 �-- 1·0 8
V y/ SK SK V / /IOK / 1
,,/ V61< /1--- 0-9 / / ./ / I1
...
.. :/ V I 7
..
V/" / /
o
V V / / /' /" /
0-8
/ . /6K/ 1// 71< /8K
� 8 ",. 6
V5K ...
0-7, :... -� _/....1-"" ,...,
/' I ,/ V ....-

V V v" 5
0·6
//
7 O'S
'"81<
4
J - -
'
0-4
3
0'3
6 2
0·2

0·1

5 o o
-19 -
1 9 -5 -20 -20·5 -21 -21,5 -22 -22·S -19 -19·S -20 -20·S -21 -21·5 -22 -22'S -19 -19·5 -20 -20'5 -21 -21'S -22 -22-S
FIG. 4 &ias Bias Bias

......
I: Fig. 4. 6V6-GT push-pull UL operation, Eb =
EC2 =
285V, tap 50/0, peak grid voltage equals bias:
;s
�Ctl (A) Power output versus bias; (B) Total harmonic distortion versus bias; (C) Total harmonic distortion
.... versus bias for 20,000 ohm load, with 8,000 ohm load for comparison. Note that THD scales in B
\0 and C differ.
"l\
\.It
-.....J
IJ>
 74
2. Comparison between UL and divided p late 3. Comments on UL operation.
and cathode loading.
The UL amplifier is quite d istinctive, neither
Qiyided plate and cathode loading is used in the
other amplifiers ( Ref. 1 and 2 ).
a pentode nor a triode, bllf with its own
QUAD 11 and marked characteristics. Its power output is definitely
The transformer hart-primary is in two sectIons, part less than that of a pentode, as is clearly shown for
connected from cathode to earth and part from type KT66 by Fig. 2 of Part 1 of this seriest. This
. plate to B + , while the screen is bypassed to earth. effect has often been obscured by the choice of
It is obvious that there will be an a.c. voltage conditions to give optimum performance for UL
operation and then, without any other change except
between screens and cathodes in the same way as
the connection of the s'c reens on the transformer,
the UL amplifier - in fact, it can be kept to the measuring the power output for pentode operation.
same value if desired. Thus divided loading opera­ As shown in the earlier article, this would not give
tion has a performance in some ways similar to that. optimum pentode operation, so that the comparison
of the UL operation. The principal difference is is not a fair one. This remark applies particularly
that divided loading has additional negative voltage to the curve published by Hafler and Keroes.
.
feedback in the cathode-grid circu it.
Secondly, the · reduction in distOrtion with UL
operation is much greater than the reduction in
gain. Taking Figs. 5 A and B from Part 1 of this
For equivalent results with div ided loading the
cathode winding should have the same proportion
article, which are both for 5 000 ohm loads, the
of the total impedance as for UL operation. For ratio of gains from pentode to UL 5 % operation
example, with type KT66 this impedance ratio is approximately 1 .48 times ( from the slopes of
should be about 20 % . In effect, the transformers the tangents to the linearity characteristics ) . On
will d iffer only in that the wi�d ing for the former the other hand the ratio of total harmonic distortion
is broken at the tapping point. varies from approximately 1 .4 times at low levels,
to over 3 . 3 times at 18 watts, for the same power
output in both cases. The same. effect holds under
The choice between UL and divided loading will
all conditions which we have measured, for type
be influenced by the reduced gain with the latter.
. 6V6-GT as well as KT66.
In most other respects the two methods are quite
similar, and resu lts for one apply very closely to the Thirdly, the optimum load resistance for Ui
operation, in all cases so far tested by us, is greater
other.
than that for pentode operation.
The measured impedance ratio in the QUAD II
Fourthly, the shape of the plate characteristics of
amplifier is 3 .5 % , which is considerably less than
the valve is distinctly d ifferent from those of either
the optimum ( 20 % ) found in our tests of type triodes or pentodes.
KT66. It seems l ikely that the preceding stage
gain is insufficient to perm it optimum operation of These points all ind icate that UL . operation is
the output stage, but this will be checked later by distinctly different from either triode or pentode
direct. measurement. operation, and should be regarded as a separate
phenomenon.
It is hoped to publish, at some future date, test
results using a ·special output transformer permit­
References.
ting divided loads with a choice of several impedance
ratios. 1 . W. N. Williamson and P. J. Walker, "Amplifiers
and Superlatives", W.W. 58 .9 ( Sept., 1 952 ) , 357.
2 . P. J. Walker ( letter ) , "Ultra- linear Operation",
t Radiotronics, May. 1955. W.W. 60. 1 2 ( Dec., 1 954 ) , 593 .

Radiotronics June, 1955

Registered at the General Post Office, Sydney, for transmission by pOst as a periodical. Single Copy, One Shilling

Volume 20 July 1955 Numher 7

____ Aa _______ ••• _ •••• _ _ _ ____ •
___ ____________
_ _____
_ _________ ___ ____
____ • •___________

ULTRA LINEAR AMPLIFIERS

, PART 3

b), F. Langford-Smith and A. R. Chesterman

This article gives the Output resistance and screen power Output.

1. Output resistance

:=I=t I<T,661
IIOK
Curves of measured output resistance for type I ! I
KT66, plate-to-plate, with 300 volt supply are given 1
t- Eb" Eca. 300V
in Fig. 1 for twO extreme values of bias (Curves A
and C). Curve B is the calculated curve, using the lOOK
method outlined below, using the same bias as for
curve A. Curves A and B agree, within the instru­
ment errors, for tappings of 12% or higher, thus
901<
indicating that the calculation is valid,
,
The method used for measuring output resistance 8
is given elsewhere in this issue.
80K
This measurement applies to the whole set-up,
including the output transformer, plate and screen
series resistors and 0.001 fLF condensers from plates I

to screens. In our particular case, since the multi­ I

701<
tapped inductor Ll (Ref. 1) did not have a second­ I
a.
ary winding, an A & R UL 20 watt Output trans­ I
a.
former was connected with its primary in parallel
1
with LI, and some inaccuracy thereby introduced in UJ 60K
U
readings for Ro above 10,000 ohms plate-to-plate. z
I

Since the 100 ohm plate stopper resistors are in­ � 1

Vl
cluded in the measurement, the measured plate-to­ I'
in SOl<
UJ
plate resistance should be decreased by 200 ohms a:
I
to give the valve output resistance.
... I
:::l
The output resistance may be calculated, using the I'
expression based on that derived by Williamson and
� 401<
:::l
Walker (Ref. 2.): o I

301<
Ro= ----- (1) 1

1 + XfL/ fLl 1

where Ro = plate resistance (per valve)

201<
fL = pentode mu at working point
rp = pentode 1'7l at working point
\
x = percentage turns of tapping + 100 A 8 C
and fLt = triode mu (screen connected to plate), 10K
\
r'"
-- . C
Fig. 1. Output resistance plate-to-plate versus tapping r-. - - �-
'

r- - '-- ,- .
for type KT66, push-pull operation, supply voltage A&8
300 V,. (A) Measu1'ed at --.:. 26 V, RL 5000 ohms,.
'
0
0 10 20 30 40 50 60 70 80 90 100%
(B) Calculated at -26 V,. (C) Measured at PENT. TRI.
IMPEDANC( TAP
- 34 V, RL= 8000 ohms.
 78
The two valves · used for the measured curves had Measured results are shown 10 the curves of
the following average characteristics at Bb BC2 Fig. 2, for type KT66 under stated conditions.
.'
= =

300V, BCl -26V: = Curve A shows that the percentage of total power
IL =227; 1'" 44, 300; ILt
= 7.7; IL/fLt 29.5. =
=
contributed by the screen varies from 2.3% at the
The calculations are tabulated below: 5 % tap, steadily increasing to. 7.81'0 for triode
operation. Curve B shows that the total power
f
x
contributed by the screen under the selected con­
Impedance Output resistance
ditions increased ·from less than 4% with a 5000
rado per valve . plate-plate
ohm load to 5.85% with a 10,000 ohm load.
0% 0 44,300 88,600
The impedance (Rs ) into which a single screen
5% 0.224 6,830 13,760
works was calculated from the known screen power
10% 0.316 4,300 8,600
Output.
15% 0.387 3,580 7,160
20% 0.447 3,120 6,240 Let I.� = screen a.c. current per plate
100% 1.0 1,450* 2,900 '
and Bs = screen a.c. voltage per plate.

;, This checks well with the direct measurement (triode) Then FO. (screen) � BsI.� = 1/ R. •.
1500 ohms.
Therefore Rs = B.•/Is.
References
1. Part 1 of this series, Radiotronics, May 1955, p. 60, For example, with type KT66, 300 volts,
Fig. 6. RL =
5000 ohms plate-to-plate, 13 watts, 20% tap:
2. Williamson & Walker "Amplifiers and Superlatives",
Bs 57 V per valve
W.W. 58.9 (Sept., 1952), 357.
=

2. Screen power output Measured Is = 4.5 mA per valve

With UL operation the screen' contributes a Screen power Output per valve = BsIs = 0.256 W
percentage of the total power output. The screen
Rs Bs/Is 12,700 ohms per valve.
power output may be determined for a single valve
= =

from the expression:- Also screen power output per valve = (Is) 2Rs
screen a.c.· voltage X screen
=

P.O. (screen)
0.256W, agreeing with the result obtained by the
=

a.c. current
other method.
where all these values refer to those measured on
a single valve. The screen a.c. voltage may either Note that this calculated value of Rs, the im­
be measured directly or calculated from the plate pedance into which a single screen works, is con­
a.c. voltage and the mrns ratio of the transformer. siderably greater than the proportion of the load
Before the screen a.c. current can be measured, it resistance RI, which is reflected through the trans­
must be separated from the screen d.c. current by
former to the tapping point. This result is to be
means of a choke-capacitor network.
expected from fundamental network analysis, in
. The total power for a pair of push-pull valves is
which twO generators feed a common load resistance
twice that for a single valve. The fact that the
in phase; each generator "sees" a higher load re­
screen was delivering power and not receiving
power was checked by noting that the screen a.c. sistance than the actual load. For those who are
voltage at the screen itself was greater than that interested, a simple analysis along these lines will
on the supply side of a series resistor. appear in a fumre issue.

I-
� A K T66 I-I- B KT66 I- t--
� I--
0 1- Eb = E 2
C '"
300V Eb=E ='300V I-l-
'Q. f-- C2
1- Ecl=- 26V Ecl:- 26V
t--t--
4-
:;, c::
'" 1-
RL: S,OOO.n
f--
UL 20 % TAP
t--I-
Q.� 10%
4- ...
:;, I-
e o �
>-
.0'
I--I--
..... �J.-
0-0

� .
�
-
5%
-I--I-- -� >-
01-
o Q.
-�l- L--I-I--
... 0-
C :;, V ..
� '"
u V
...
�
Q.

o 10 20 30 40 50 60 70 80 90 100 SK 6K 7K BK 9K 10K
PENT. Taps I Percentoge ImpedanCe!: TRI. Load Re!:sistance

Fig. 2. Percentage of output power supplied by the screen, for type KT66;
(A) vet'sus tapping p oint; (B) verstts load resistance.

Rariiotronics July, 1955.

 79

SOME EFFECTS OF NECATIVE FEEDBACK

ON OUTPUT RESISTANCE
by E. Watkinsol1

1. Measuring amplifier output impedance Thus Ro I is the resultant of three parallel resistances,

What appears to be a simple method of measuring RL3 + 1'1)3

the output impedance of an amplifier, with or with­ RL3, 1'1)3 and , and the effect of the
Out feedback, is shown in Fig. 1. A voltage E is �AIA2/-t3
developed across the output load through a resistor R
which is much greater in resistance than the output RL3 + 1''l13

impedance of the amplifier. By determining the feedback is to add a resistance . --- in parallel
current I flowing in R and the voltage E, the parallel
impedance of the amplifier output impedance Ro with RL3 and 1'p3'
and the load resistance RL can be obtained. Since
the value of RL is known, or' can be determined by The conventional solution of output impedance
the same method, Ro can be simply calculated. in the presence of feedback shows that the effect
Analysis of this circuit shows that

E E E�AIA2/-t3 of feedback is to add a resistance of --- in

J= - + -+ BA1A2/-t3
RL3 1'p3
parallel with the output circuit. Especially in the
where RI + R2 » R L 3 case of triode output valves, for which RL3 and 1'p3
may be of the same order, the difference between
R2 these twO solutions' may be significant.
�= ---

R1 + R2 The reason for the difference is that in Fig. 1

A stage gain
= the voltage E is set up across RL3 and 1'p3 in parallel,
and gmlR2« 1. whereas in the conventional analysis E is a voltage
in series with RL3 and 1'p3 as in Fig. 2. Since only
Since the amplifier output resistance with feed-
the latter con �ition simulates the effect of voltages
_

developed in the voice coil of a speaker when

E
"hanging on" at its resonant frequency, it is the
back Ro' =

correct method for determining amplifier output

1
E impedance.
Ro'
When the circuit of Fig. 2 is used for Output
impedance determination, the source impedance of
+ +
E must be much less than that of the series con­
nection of RL3 and 1'p3, which may be difficult when
1 measuring on the secondary side of the output
transformer, but which can be readily arranged
on the primary side. A suitable step-down trans­
+ + former may be used for measurements on the
secondary side.

rv

FI G. 1. All D. C. Sourcu Omitted

Radiotronics July, 1955.

 80
2. Output impedance of other amplifier stages reduced the Olltput voltage of the second stage

Although the effect of negative feedback on the 1J..2RD2

output impedance of the output stages of amplifiers
is commonly discllssed, its effect on the apparent out­
put impedance of other stages is never referred to.
R,.2
As shown above, it is necessary, in determining
1 + �/lIA:!,Ll� ----­

OlltP llt impedance, to use a reasonably close approxi­

'l'p2 + R,.�
mation to actual operating conditions if a correct
analysis is to be made. For example, the method of
,Ll2R'J2
inserting a voltage in series with the load as shown
t Ei' /1) - -- --
in Fig. 2 is not applicable to stages other than the 1'1'2 + R,,'2
Output stage because in operation sllch serIes \101-
to
tages do not appear. R/,2
However, changes in the impedance of R" are 1 + t �AjA:�/J..2 ---­

met with, for example, if a following stage is driven 'rp� + R/J2

FIG.2. All D.e. Sourc($ Omitted

into grid current or if R" has a reactive component. Thus, when a reasonable degree of feedback is used,
Thus a determination of the variation in output the effect of the shunting resistor on the output of
voltage with changes in the value of R" may be the driver stage is small, which is another way of
useful. saying that the output impedance of the driver
stage is low. The overall feedback has therefore
In Fig. 1 let a resistor be connected in parallel
reduced the output impedance of the second stage.
with Rr•2 which halves its effective resistance. In
Similar results would be obtained by calculating
the absence of feedback the additional resistor will
the effect on any other stage.
reduce the voltage output of the second stage
It is of interest to note that the . output impedance
,Ll2 R/,2
from Ei. At -----
of stages other than the output stage is reduced by

+ R,.2 either negative voltage feedback or negative current

1'p2
feedback so long as this is taken from a stage
,Ll2 R,.2 subsequent to the stage under investigation.
to
This determination of interstage output impedance
is nevertheless merely a different method of examin­
i.e., when 1'p2 » RL2 as for pentodes the output of
ing the effects of negative feedback upon, say, fre­
the stage is approximately halved.
quency response if Rr, is changing with frequency
Let the input voltage, to give the same output or distortion if R" is changing with amplitude. It
when feedback is applied to the amplifier, be Ei.', has no fundamental significance with respect to the
then with feedback the additional resistor will behaviour of negative feedback amplifiers.

Radiotl'onics July, 1955.

 115

ANALYSIS OF TWO GEN ERATORS· FEEDING A LOAD

by F. Langford-Smith

In a previous article (Ref. 1) reference was made Then 122 R = Ea/ 2

to some of the effects of two generators feeding a R Ea/12
load, in that case the generators being the plate and
Ea
screen of a pentode working with the "Ultra Linear" R x
connection of the ·screen.

The following treatment is quite general, treating

finally R Ra ........ ........ (1)
the effects of two generators· feeding a resistive
= ---

12
network. In Fig. 1 there are two constant voltage
generators El and E2, each in series with a resistance Eqn. (1) shows that the load into which E2
works is not Ra, as might be expected, but a
resistance greater than Ra. Taking a numerical case
for illustration, El 120 volts, E2
= 110 volts, =

RI =
250 ohms, R2 500 ohms, R a = 1000
=

ohms. It is shown in the Appendix that 1 1 =;

80 mA, 12 =
20 mA and 11 + 12 100 mA. =

Applying equation (1), the load resistance into

which E2 works' is given by

R = 1000 (100/20) =
5000 ohms.

Thus in this case, with E2 contributing one-fifth of

the total current, the load resistance into which E2
works is 5 times Ra.

Appendix
In. Fig. 1, let El = 120 volts, E2 =
110 volts,
RI = 250 ohms, R2 = 500 ohms, Ra = 1000 ohms.
(RI and R2 respectively), feeding a load resistance El = I 1Rl + (11 + h) R a ...... ...... (2)
Ra. The case. treated here is when both generators E2 = 12R2 + (It + 12) R a .. ........... (3)
.are supplying power in the positive direction; it El - E2 120 - 110 10 volts =

does not cover the case where one generator IS

IIRl - hR2
being driven "backwards" by the other.
I lRl - 10 + 12R2
Power generated by El El/1 11 (10/250) + 12 (500/250)
11 0.04 + 212
Power generated by E2 E2/2 It + 12 = 0.04 + 31 2 ........... (4)

Total power generated El/1 + E2/ 2 Applying (4) in (3):

110 500 h + (0.04 + 3h) 1000
Power loss In RI I1 Rl
110 3500 h + 40
Power loss in R2 1 2R2 70/3500 = 20 mA.
0.04 + 2 X 20/1000 80 mA.
Power dissipated in Ra Ea (I1 + 12)
It + 12 = 100 mA.

References
Power in Ra contributed by E2
1. F. Langford-Smith and A. R. Chesterman, "Ultra
Let the power in Ra contributed by E2 be repre­ Linear Amplifiers" - Part 3, Section 2, Screen
sented in the alternative form 122R, where R is the Power Output, Radiotronics 20.7 (July, 1955),
load resistance into which E2 works. 78.

Radiotronics October, 1955

♦ Verso Filler Page ♦
 WIRELESS
ENGINEER
Vol. 32 AUGUST 1955 No.8

"L1tra-Linear Amplifiers

I N recent years, and particularly in the C.S.A.,

the so-called ultra-linear amplifier has become
popular among quality enthusiasts.
great deal has been written about it and a good
A
be an ordinary tetrode or pentode one. If it were
at the anode end of the winding. the screen-grid
would be joined to the anode. and the stage would
be a normal triode one.
many performance figures have been published What happens is that a portion of the alternating
which do seem to show some reduction of non- voltage developed on the anode is fed back to the
linearity distortion as compared with similar screen grid and the amount of feedback depends
triode, tetrode or pentode amplifters 1 • No on the position of the tapping on the winding.
serious attempt at explaining why this result By varying the amount of feedback, a gradual
should occur seems to have been made, however. transition between the limiting triode and tetrode
The arrangement is shown in Fig. 1 applied conditions is obtainable. ~ow a triode is generally
to a single output valve (a pair in push-pull is thought to be more linear than a pentode and it is
usually adopted) and it can be seen that it differs sometimes said that this can be accounted for
from a normal output stage only in that the by the linearizing action of feedback. A triode
screen grid is connected to a tapping on the can be regarded as merely a tetrode with 100%
output transformer primary. If this tapping feedback to the screen-grid.
were at the h. t. end of the winding, the screen- If this were all, the ultra-linear stage would
grid would be joined to h.t. and the stage would represent an intermediate condition between the
tnode and the tetrode and one would expect it
to be less linear than the triode and more linear
than the tetrode. On this basis. there is nothing
to account for its being more linear than either.
The flaw in this argument is the supposition
c that the ordinary ideas about negative feedback
hold good when they are applied to this special
case of feedback to an electrode other than the
input electrode. The normal ideas have been

I
VI
1
v, R
developed on the basis of the feedback and the
input being applied to a common electrode so
that the feedback voltage and the input voltage
I
are additive. It is well known that feedback
j ,I then always tends to improve the linearity.
When separate electrodes are used, however. it
can be shown that feedback will introduce
Fig. 1 Fig. 2
-~----~----~~----------------~~~~--------­ curvature into an otherwise straight charac-
'·'.-\mpWlms and Superlalh'es" bv D. T. :i. WUlIamson and P. J.
Walker, Wi,eless Wo,Ul, Seplmlbcr i952, p. 357. teristic.

\VIREL1!:SS E~GINEER, AUGUST 1955 199

 A modified form of the ultra-linear amplifier linear action, but the feedback grid is still linear,
is shown in Fig. 2; in this. a fraction P of the so that the equation for anode current is
anode voltage is applied through the CR coupling ia == (a l + blV l + Cl V 12 + d l V 13)(a 3 + b3 r 3 )
to the suppressor grid. It is well known that in a and, as before, V s = - ~iaRa. Then
pentode the cathode current is . a 3 (a l + bl VI + Cl V 12 + d l V 13 )
iR = a l + b1 V 1 ta = f+-pb 3R a (a l + bll-~l + Cl V l 2 + dlV~3}

within the limits 0 f a 1mear · approxlma . t'Ion. It 'l~ = A al- +-blV +.C-l -Vl 2.-2"-.--.-s-
- -I.. - + d 1 V 13
also well known that it is independent of V 3 1 + B(b 1 V I + cll' 1 + d 1 It I }
which serves only to control the division. of current wh€r~ A and B have the same values as before.
........................................................................................... This equation can be expanded to the form
ia == ALaI + (b l - alB) VI - (1 - a 1 B)(b 1 2B - c 1 ) V 12
+ {B Zb13{l - alB) - Bblcl{l - 2a l B) - B(ald l -j- blc l ) + d l } V 13 ••• ]
.................................................................... <....................... retaining terms only up to the cube.
between screen-grid and anode, so that It is at once evident that it may be possible to
ia = ik (as + baTt's) choose P so that the coefficient of V\3 is zero.
Therefore, we can write It is not worth while to attempt to work out the
ia = (a l + b l V l )(a 3 + baV 3 ) •• •• (I) condition for this. The equation would be com-
In Fig. 2, the anode voltage is -iaRa, where plex and of little practical utility because we
Ra is the anode load, and so V 3 = - PiaRa, and have assumed a linear control of anode current
(1) becomes by the feedback grid and this is unlikely to be
. aa(a 1 + bl V 1) present in practice.
'a = 1 + pbaRa (a + blV l I) ...........................................................................................
= A[a 1 + bl (1 - alB) VI - b/·B {I - alB} VIZ + bl 3 B2 (1 - alB) VIS • • • • • ] (2)
.....................................................................................................
where A = aa/(I + a l b3 PRa} "'ith feedback to a grid other than the signal
and B - b3 pRa!(1 + alb s ~Ra) grid, however. the possibility exists of being able
so to adjust the amount of feedback that a
Feedback to the suppressor grid thus introduces particular harmonic can be eliminated from the
curvature into an otherwise linear characteristic. output. In the ultra-linear amplifier, two valves
If the input VIis a sine wave, the anode current are used in push-pull 50 that all even harmonics
will contain all harmonics of it. are, ideally, eliminated.
Feedback to the screen-grid of a valve is, of The third harmonic, which is normally the
course, not quite the same thing as feedback to most important remaining one, can, again ideally,
the suppressor grid. However, as is well known, be eliminated by a critical adjustment of feedback
the mutual conductance does depend upon the to the screen-grids. The distortion present
screen voltage and so the equation for this should thus be confined to odd-order harmonics
condition must be rather like (I); there must be a higher than the third. Superficially, therefore,
tenn involving the product of the control- and the ultra-linear amplifter would appear to be an
screen-grid voltages. If this product term is improvement on ordinary types. We say super-
present, then the equation for screen-grid feedback ficially because we have not examined the
will be of the same general form as (2). amplitudes of the residual harmonics and the
It would thus appear that feedback to an ability to eliminate a third harmonic would confer
electrode other than the input electrode is a bad no 1;>enefit if it meant that the higher harmonics
thing. However, we have only so far shown that became of comparable amplitude.
it makes an otherwise linear stage non-linear. There is no reason to suppose that this is the
There is the possibility that the kind of non- case, however, for the evidence of measurement
linearity which it introduces is inverse to that does support the claims made for the ultra-linear
which occurs naturally in valves. If this amplifier. Our aim here has been merely to show
possibility is a reality, then a critical degree of the mechanism by which this improvement is
feedback to screen or suppressor grid would achieved, for we have felt that it has Temained
improve the linearity of a stage. too long unexplained.
Let us suppose that the input grid has a non- W.T.C.

200 ·WIRELESS ENGISEER. ACGUST 1955

 The Ultra Linear
Power Amplifier
An adventure between triode and pentode

by Rudolf Moers

1
 Who am I
Born in 1955 in Veldhoven and now living in Eindhoven in the Netherlands.

Education: Primary Technical School  electrical engineering

Secondary Technical School  electronics
High Technical School  electronics
Summary of work experience:

Halin  Analog video modification (RGB-keying)

Analog audio circuits with semi-conductors
Philips Optical Disk Mastering  Compact Disk mastering electronics
Compact Disk signal processing electronics
And a lot more
Philips Medical Systems  Diaphragm control of Röntgen camera
Philips Electron Optics  Vacuum pump control for electron microscope
Secondary Technical School  Teacher electronics, theory and practice
ASML  Architecture of electronic hardware
Infrastructure of cabling and racks with
electronic boards and supplies.
Hobby : electron tube amplifiers and radio’s
2
Contents of this presentation
1. Introduction and history

2. Comparison of the static characteristic for Triode, Ultra Linear and

Pentode

3. Comparison of the powers for Triode, Ultra Linear and Pentode

4. Network analyses of the Ultra Linear Power Amplifier

a. Repetition of the pentode characteristics

b. Repetition of the pentode quantities
c. Current source and Voltage Source equivalent circuits of the
Pentode
d. Current source and Voltage Source equivalent circuits applied
Ultra Linear

5. Determination of the screen grid tap

3
Contents of this presentation
6. Test equipment

7. Practical evidence 1 of the network analyses of the Ultra Linear

Amplifier

8. Comparison of practical powers and efficiency of an amplifier in

Triode mode, in Ultra Linear mode and in Pentode mode

9. Practical evidence 2 of the network analyses of the Ultra Linear

Amplifier

10. Comparison practical frequency behavior of an amplifier in Triode

mode, in Ultra Linear mode and in Pentode mode

11. Comparison practical non-linear distortion of an amplifier in Triode

mode, in Ultra Linear mode and in Pentode mode

12. Bibliography
4
1. Introduction and history
David Hafler & Herbert Keroes (not the inventors) published
their Ultra Linear story in 1951.
Publishing in 1959 of the Dutch book “Radio Technique part 1”
written by A. J. Sietsma of the Philips company.
The Philips company has never published an Ultra Linear story, but A. J. Sietsma
made a homework exercise about screen grid negative feedback for students.

Rudolf Moers solved this homework exercise about negative feedback in 2006
with his own formulae which gave the same results as A. J. Sietsma.
5
6
Screen grid tap of the
primary transformer
winding : x

vg 2,k v g 2, k
x x
vak vak

v g 2,k  x  vak

0.0 ≤ x ≤ 1.0

x = 0 : pentode
0 < x < 1 : ultra-linear
x = 1 : triode

7
2. Comparison of the static characteristic for Triode, Ultra Linear
and Pentode

Same load lines

with different
scales for Vak

triode
Constriction 40V Constriction 100V
pentode

Constriction 50V
Ultra-linear 8
3. Comparison of the powers for Triode, Ultra Linear
and Pentode in theory.

Vap,triode << Vap,ultralinear

Vap,triode << Vap,pentode
Iap,triode << Iap,ultralinear
Iap,triode << Iap,pentode

Va,pentode = Va,ultralinear
Ia,pentode = Ia,ultralinear
By this:

Pap,triode << Pap,ultralinear

Pap,triode << Pap,pentode

Pa,pentode = Pa,ultralinear

9
 Comparison of the powers for Triode, Ultra Linear
and Pentode in practice.

In Menno’s first book some design examples are shown which use with jumpers to configure
the circuit into triode, ultra-linear and pentode.

Power results:

2x EL34 with transformer VDV6040PP: ptriode = 13W, pultralinear = 33W and ppentode = 40W
4x EL34 with transformer VDV3070PP: ptriode = 30W, pultralinear = 70W and ppentode = 80W

ptriode = 13 W  20 W  pultralinear = 33 W versus pultralinear = 33 W  7 W  ppentode = 40 W

ptriode = 30 W  40 W  pultralinear = 70 W versus pultralinear = 70 W 10 W  ppentode = 80 W

Reason : vap,triode << vap,ultralinear < vap,pentode

The constriction of the vg1,k -curves in the anode characteristic Ia = f (Vak)

near the Ia –axis is slightly more with ultra linear than with a pentode.

10
4. Network analyses of the Ultra Linear Amplifier
4.a. Repetition of the pentode characteristics

Anode current 11
Screen grid current

12
13
4.b. Repetition of the pentode quantities

Anode steepness is also called mutual conductance gm.

I a
Definition of anode steepness : S with constant Vak and Vg2,k
V g1,k

ia
For small signals : S with constant Vak and Vg2,k
v g1,k

I g 2
Definition of screen grid steepness : S2  with constant Vak and Vg2,k
V g1,k
ig 2
For small signals : S2  with constant Vak and Vg2,k
v g1,k

14
Definition of anode Vak
amplification factor :
 with constant Ia and Vg2,k
V g1,k

v ak
For small signals :  with constant Ia and Vg2,k
v g1,k

Definition of screen grid V g 2,k

amplification factor :  g 2, g1  with constant Ig2 and Vak
V g1,k

v g 2,k
For small signals :  g 2, g1  with constant Ig2 and Vak
v g1,k

Anode penetration factor : Da = µ-1 = 1/µ (Anode Durchgriff )

Screen grid penetration factor : Dg 2   g12, g1  1 /  g 2, g1 (Screen grid Durchgriff)

15
Definition of anode Vak
AC internal resistance : ri  with constant Vg1,k and Vg2,k
I a

v ak
For small signals : ri  with constant Vg1,k and Vg2,k
ia

Definition of screen grid V g 2,k

AC internal resistance : ri 2  with constant Vg1,k and Vak
I g 2

v g 2,k
For small signals : ri 2  with constant Vg1,k and Vak
ig 2

16
Barkhausen’s anode formula :   S  ri

Barkhausen’s screen grid formula :  g 2, g1  S 2  ri 2

µpentode as triode ≈ µg2,g1

At the anode:

Anode AC internal resistance: ri (or plate resistance)

Anode AC external resistance: ra (external load at the anode)

Screen grid tap of the primary transformer winding : x

17
 I a
S
V g1,k

ΔVak
μ
ΔVg1,k

Vg1,k
I a

Vak

ΔVak
r i
I a ΔI a
Vg1,k
Vak

Anode current
18
 I g 2 ΔV g2,k
S2  ri2 
V g1,k ΔI g2

Vg 2,k
Vg 2,k

I g 2

Vg1,k Vg1,k

ΔVg2,k
μ g2,g1 
Screen grid current ΔVg1,k
19
 I a
S
V g1,k
I g 2
S2 
Vg1,k

∆Vg1,k for S = ∆Vg1,k for S2

I a I g 2
Vg1,k  
S S2

S2
I g2   Ia
S

S2
ig 2   ia
S

20
4.c. Current and Voltage Source equivalent circuits for the Pentode
For triodes:
According to the definitions, AC voltage vg1,k causes anode current : ia1 = S·vg1,k
According to the definitions, AC voltage vak causes anode current : ia2 = vak / ri
vak
Superposititon of ia1 and ia2 gives : ia  S  v g1,k  apply Barkhausen’s   S  ri
ri
 v 
The triode equation cc : ia  S   v g1,k  ak 
  

For pentodes:
v ak
Factor contributes to the anode current slightly because µ is large

see anode steepness characteristic Ia = f (Vg1,k).
v g 2,k
Factor contributes significantly to the anode current because µg2,g1
 g 2, g1
is small, see screen grid steepness characteristic Ig2 = f (Vg1,k).

 v g 2, k v ak 
The pentode equation : ia  S   v g1,k  
  g 2, g1  
21
  v g 2, k v ak 
The pentode equation : 
ia  S  v g1,k  
  g 2, g1  

S2
Equal control grid base for anode current and screen grid current : ig 2   ia
S
 v g 2, k v ak 
Apply this in the pentode equation : ig 2 
 S 2  v g1,k  
   
 g 2, g1

After some mathematical magic tricks we get the current source and voltage source models.
 v g 2,k  v ak
Anode current source : 
ia  S  v g1,k  
  g 2, g1  ri

 vak  v g 2,k
Screen grid current source : ig 2  S 2   v g1,k   
   ri 2
 v g 2, k 
Anode voltage source : 
ia  ri    v g1,k    v ak
  
 g 2 , g 1 
 v 
Screen grid voltage source : i g 2  ri 2   g 2, g1   v g1,k  ak   v g 2,k
  
22
23
4.d. Current source and Voltage Source equivalent circuits
applied to Ultra Linear
vg 2,k v g 2, k
x x
vak vak
v g 2,k  x  vak

0.0 ≤ x ≤ 1.0

TARGET:

A = vo /vi = f (x) and rout = f (x)

24
25
Without formulae we see directly :

ig2 flows through part x of the primary winding  partly contribution to power

ia flows through part (1−x) +x of the primary winding  full contribution to power

With formulae derivation from the equivalent circuits we achieve :

Anode voltage : vak  v g 2,k  ia  1  x   ra  ik  x  ra  ia  1  x  r a

vg2,k = −(ia + ig2 ) ∙ x ∙ ra and is Kirchhoff’s first law ik = ia + ig2 for AC

vak
Total AC current :  ia  x  i g 2   itotal
ra

The total AC current itotal is not the same as cathode AC current ik.

26
With the art of magic
formula tricks …….  v g 2, k v ak 

The pentode equation : ia  S   v g1,k  
  g 2, g1  

v g 2,k  x  vak

S2
ig 2   ia
vak S
 ia  x  i g 2   itotal
ra

 S2   x  vak vak 

itotal  1  x    S  vg1,k  
 S     
 g 2, g1

vak   itotal  ra

v ak S  x  S 2   ra
.………… we achieve at the anode: Aa  
v g1,k  x 1
1     S  x  S 2   ra
 
 g 2, g1  
An easy formula derivation in small steps is available.
27
 v ak S  x  S 2   ra
Aa  
v g1,k  x 1
1     S  x  S 2   ra
 
 g 2, g1  

ns
vo   vak and vi = vg1,k
np

A
vo n
 s 
S  x  S2  ra
vi np  x 1
1     S  x  S 2   ra
 
 g 2, g1  

28
 vo,open
AC output resistance : rout  (Thevenin’s theorem)
io, shortcircuit
2
 ns 
When we have vo,open then RL = ∞ with a 
r    RL   .

 np  Again with the art of
2 magic formula tricks ….
n 
When we have io,shortcircuit then RL = 0 with ra   s   RL  0 .

 np 

vo,open ns 1
 
vi np  x 1
  
vo,open  
 g 2, g1  
rout 
io, shortcircuit np  S 
io, shortcircuit   1  x  2   S  vi
ns  S 

2
n  1
rout  s  
.………… we achieve at the output: n 
 p  S  x  S    x  1 
2  
 g 2, g1  
An easy formula derivation in small steps is available. 29
Summary
vo n S  x  S 2   ra
A  s 
vi np  x 1
1     S  x  S 2   ra
 
 g 2, g1  

2
 ns  1
rout   
n 
 p  S  x  S    x  1 
2  
 g 2, g1  

x is the variable and the other quantities

are almost constant (in theory).

30
5. Determination of the screen grid tap

31
32
33
 Table 1. Measured values of line 1
Vak (V) Ia (mA) Ig2 (mA) Vg2,k (V) ΔVak (V) ΔVg2,k (V) V g 2,k
adjusted read on measured adjusted to [300V – Vak] [300V – Vg2,k] x
Ia-axis achieve the read Ia Vak
0 0 0 0 300 300 1.00
25 0 0 25 275 275 1.00
50 0 0 50 250 250 1.00
75 0 0 75 225 225 1.00
100 0 0 100 200 200 1.00
125 0 0 125 175 175 1.00
150 0 0 150 150 150 1.00
175 2.6 0.1 175 125 125 1.00
200 8.5 0.7 200 100 100 1.00
225 19.2 1.6 225 75 75 1.00
250 35.6 2.9 250 50 50 1.00
275 55 4.6 275 25 25 1.00
300 79 7.0 300 0 0 unknown
325 110 9.2 325
350 140 12.1 350 Not further than Not further than Not further than
point W point W point W
375 170 16.5 375
400 200 21.0 400

The adjustment of Vg2,k happens automatically of course, because the screen grid is connected to the anode.
The screen grid primary transformer tap x = 1.00 but that will surprise nobody, so pentode as triode.
34
 Table 3. Measured values of line 3

Vak (V) Ia (mA) Ig2 (mA) Vg2,k (V) ΔVak (V) ΔVg2,k (V)
adjusted read on measured adjusted to [300V – Vak] [300V – Vg2,k] V g 2,k
Ia-axis achieve the read Ia
x
Vak

0 0 0 unknown 300 unknown unknown

25 6.5 3.8 206 275 94 0.34

50 13 12.5 237 250 63 0.25

75 19.5 16 251 225 49 0.22

100 26 13 254 200 46 0.23

125 32.5 10.4 259 175 41 0.23

150 39 8 263 150 37 0.25

175 45.5 7 269 125 31 0.25

200 52 6.5 275 100 25 0.25

225 58.5 6.5 281 75 19 0.25

250 65 6.5 288 50 12 0.24

275 71.5 6.5 294 25 6 0.24

300 78 7.1 300 0 0 unknown

The average value of all screen grid primary transformer taps xaverage = 0.25. This value is mentioned at line 3.
For this specimen KT88-1 we have pure ultra-linear at x = 0.25.
35
 Table 5. Measured values of line 5
Vak (V) Ia (mA) Ig2 (mA) Vg2,k (V) ΔVak (V) ΔVg2,k (V)
V g 2,k
adjusted read on measured adjusted to [300V – Vak] [300V – Vg2,k] x
Ia-axis achieve the read Ia Vak
0 1 54 300 300 0 0.00
25 60 30 300 275 0 0.00
50 60 30 300 250 0 0.00
75 61 28 300 225 0 0.00
100 63 22 300 200 0 0.00
125 65 19 300 175 0 0.00
150 68 14 300 150 0 0.00
175 70 12 300 125 0 0.00
200 72 9.5 300 100 0 0.00
225 74 8.5 300 75 0 0.00
250 75 7.8 300 50 0 0.00
275 76 7.2 300 25 0 0.00
300 77 7.0 300 0 0 unknown
325 78 6.5 300
350 79 6.3 300 Not further than point Not further than point Not further than
W W point W
375 80 6.0 300
400 80 6.0 300

The adjustment of Vg2,k happens automatically of course, because the screen grid is connected to Vb.
The screen grid primary transformer tap x = 0.00 but that will surprise nobody, so pentode as pentode.
36
6. Test equipment

37
38
Working point: Given at Vak = 300V for KT88:

Vak,w = 300 V S = 11.5 mA/V

Ia,w = 80 mA ri = 12 kΩ
Vg1,kw = − 26 V µ = 138
Vg2,kw ≈ 300 V S2 = 1.15 mA/V
µg2,g1 = 8
Input signal for each
value of x : ra = 7000 Ω
vg1,k = 3.72 VRMS

v ak S  x  S 2   ra
Aa  
v g1,k  x 1 
1    S  x  S 2   ra
  
 g 2, g1 

2
 ns  1
rout   
n 
 p  S  x  S    x  1 
2  
 g 2, g1  
39
7. Practical evidence 1 of the network analyses
of the Ultra Linear Amplifier
Table 6

x vg1,k vak vg2,k pa vRL pRL v |Aa|

(VRMS) (VRMS) (VRMS) v g 2,k (W) (VRMS) (W) Aa  ak
xmeasured  v g1,k calculated by
vak formula
measured
0.00 3.72 158.0 0 0.00 3.60 4.01 3.20 42.4 50.6

0.10 3.72 60.8 19.2 0.32 0.53 1.51 0.45 16.3 16.8

0.20 3.72 47.7 21.3 0.45 0.33 1.22 0.29 12.8 13.3

0.30 3.72 40.7 22.3 0.55 0.24 1.01 0.20 10.9 11.3

0.40 3.72 36.1 23.2 0.64 0.19 0.92 0.16 9.7 10.1

0.50 3.72 33.4 23.6 0.71 0.16 0.83 0.14 9.0 9.3

0.60 3.72 30.9 23.9 0.77 0.14 0.78 0.12 8.3 8.7

0.70 3.72 28.9 23.2 0.84 0.12 0.73 0.11 7.8 8.1

0.80 3.72 27.3 24.2 0.89 0.11 0.69 0.10 7.3 7.7

0.90 3.72 26.1 24.6 0.94 0.10 0.65 0.09 7.0 7.3

1.00 3.72 25.0 25.0 1.00 0.09 0.62 0.08 6.6 7.0

40
 v ak S  x  S 2   ra
Aa  
v g1,k  x 1
1     S  x  S 2   ra
 
 g 2, g1  

41
rout, pentode

vo1  vo 2 
rout 
io1  io 2

2
 ns  1
rout   
n 
 p  S  x  S    x  1 
2  
 g 2, g1  
rout,ultralinear

rout,triode

42
 8. Comparison of practical powers and efficiencies of
an amplifier in Triode mode, in Ultra Linear mode
and in Pentode mode

My first electron tube amplifier according to a design from Menno’s first book.
43
Schematic diagram of my first electron tube amplifier

i g 2,1
i a1

ia 2
i g 2, 2

44
 input power
anode dissipation

anode efficiency
delivered anode power

input power

anode efficiency
anode dissipation
delivered anode power
pa = 25 W

input power

anode efficiency
anode dissipation
delivered anode power

45
9. Practical evidence 2 of the network analyses
of the Ultra Linear Amplifier
Table 7.
vak (VRMS) vg2,k (VRMS) v g 2,k given x of power transformer VDV6040PP
x measured 
v ak
31.1 12.5 0.402 0.400
100.0 40.1 0.401 0.400
193.6 78.0 0.403 0.400

46
ia1  ia 2  ia,measured  59 mARMS and ig 2,1  ig 2, 2  ig 2,measured  14 mARMS
Substitution of these currents in: itotal  ia,measured  x  i g 2,measured
itotal  59  0.4 14  59  5.6
itotal  64.5 mA
Anode AC external resistance: ra  1  raa,VDV 6040PP  1  6000   3000 
2 2
We have seen that the total anode power is 25W pa, EL 34  12.5 W
2 2
vak vak
pa , EL 34  12.5   vak  12.5  3000  193.6 VRMS
ra 3000
vak
For determination of itotal apply : itotal  ia ,measured  x  i g 2,measured 
ra
vak 193.6
itotal    0.0645 A
ra 3000
itotal  64.5 mA
The effect of this fictive itotal is an anode power of 12.5 W for one power pentode.

The effect of (ia,measured  x  ig 2,measured ) is an anode power of 12.5 W for one power pentode.
47
10. Comparison of practical frequency behavior of an
amplifier in Triode, Ultra Linear and Pentode mode.
Triode Ultra-Linear Pentode

15 Hz 21 Hz 18 Hz
80 kHz 70 kHz 50 kHz
Differences in frequency and phase characteristics are small.
The parasitic capacitances due to the Miller effect are not negligible
for triodes, but their anode AC internal resistance is low.
The parasitic capacitances due to the Miller effect are negligible
for pentodes, but their anode AC internal resistance is high.
The limiting of 3dB-high is mainly due to the limiting bandwidth of the 48
transformer.
11. Comparison practical non-linear distortion of an
amplifier in Triode mode, Ultra Linear mode and
Pentode mode

d total
d2
d3
d5
d4

d total
d2
d3
d5
d4

d total
d3

d2
d5
d4
49
Final summary
These formulae match in
theory and practice.
vo n S  x  S 2   ra
A  s 
vi np  x 1 
1    S  x  S 2   ra
  
 g 2, g1 
2
n  1
rout   s  
n 
 p  S  x  S    x  1 
2  
 g 2, g1  
x is the variable and the other quantities
are almost constant (in theory).
50
The contribution of the anode AC to the delivered output power is much more than
the contribution of the screen grid AC because in practice: x ≈ 0.4 and i g 2  0.2  ia

vak
 ia  x  i g 2   itotal
ra
It is possible to determine screen grid tap x for an ultra-linear application for each
sample of a pentode after measuring the anode characteristics for both the triode and
pentode configurations of that pentode.

The delivered output power at ultra-linear is slightly less than with pentode configuration.
The delivered output power at ultra-linear is much more than with triode configuration.
The configurations triode, ultra-linear and pentode have an almost equal audio bandwidth.

The ultra-linear configuration gives almost the same low non-linear distortion as the
triode configuration.

The ultra-linear configuration has the “high power advantage”

of the pentode configuration and the “low non-linear distortion
advantage” of the triode configuration.
51
12. Bibliography

52
electron's

ELECTRON TUBES PAGE

 

This page is dedicated to explaining the fundamentals of electron tube technology.

Electron Tubes are electro-mechanical devices that use electrons to perform work.
 
 

1.    ELECTRONS
Note: The following text may be heresy to modern scientists, however this was the state
of the art in 1959.
 

The RCA Receiving Tube Manual RC-19 (1959) describes "electrons" in this way:

"All matter exists in the solid, liquid, or gaseous state.

These three forms consist entirely of minute divisions known as molecules, which, in turn, are composed of atoms.

Atoms have a nucleus which is a positive charge of electricity, around which revolve tiny charges of  negative electricty
known as electrons.

Scientists have estimated that electrons weigh only 1/30-billion, billion, billion, billionth of an ounce, and that they may travel
at speeds of thousands of miles per second.

Electron movements may be accelerated by the addition of energy.

Heat is one form of energy, which can be conveniently used to speed up the electron. For example, if the temperature of a
metal is gradually raised, the electrons in the metal gain velocity. When the metal becomes hot enough, some electrons
may acquire sufficient speed to break away from the surface of the metal. This action, which is accelerated when the metal
is heated in a vacuum, is utilised in most electron tubes to produce the necessary electron supply."
 
 

2.    PUTTING ELECTRONS TO WORK IN ELECTRON TUBES.

The RCA Receiving Tube Manual RC-19 (1959) and the RCA Transmitting Tube Manual TT-4 (1956) propose that:

"Electrons are of no value in an electron tube unless they can be put to work.

Therefore, a tube is designed with the parts necessary to utilise electrons, as well as those required to produce them.

Now, the transfer of electrical energy through a circuit involves control of two factors - rate and direction.

The rate of energy transfer is determined by the number of individual electron charges moving unidirectionally through the
circuit in a given interval of time and is proportional to the applied voltage.

The direction in which the electron charges move is determined by the polarity of the applied voltage.

In electrical circuits, control of the direction of current flow is necessary when the power source produces ac (alternating
current) voltages and currents and the load requires a uni-directional current.
Devices that are used primarily to control the direction of current flow are known as "rectifiers".

All such devices (rectifiers) however are also rate control or rate limiting devices in the sense that they have a finite
current carrying capacity.

Now, electric charges may be transferred through a circuit by several methods.

In one method, kinetic energy is transferred between adjacent electrons within the molecular structure of a conductor. This
method is employed in switches, rheostats, potentiometers and other devices that utilise conductive materials as control
electrodes.

Because the currents through such devices are controlled by mechanical means, the speed with which the amount or
direction of current can be changed is limited by friction or inertia.

In a second method, individual electrons are transferred in one direction through semi-conducting materials such as in
silicon diodes and transistors. This method has the advantage that the rate of current flow may be controlled by electric
fields.

In a third method, individual electrons are transferred through a low-density, non-conductive medium, such as a vacuum or
low pressure gas. This method is used in electron tubes and has the advantage that both rate and direction of current
flow may be controlled by electric fields.

Because these fields, as well as the electrons, have negligible inertia, electron tubes can effect changes in the value and
direction of electric current at speeds considerably higher than those obtainable with mechanically operated devices.
 
 

3.    ELECTRON TUBES

3.1    CONCEPT -  VACUUM TUBES, ELECTRON TUBES, THERMIONIC VALVES
 

The ELECTRON TUBE is an electro-mechanical device that enables an electrical current to be controlled be external
electrical means.

The Vacuum Tube is commonly known around the world by the names "Vacuum Tube, Electron Tube, Radio Tube or
Thermionic Valve."

The RCA Receiving Tube Manual RC-19 (1959) describes it so:

"The Electron Tube is a marvellous device.

It makes possible the performing of operations, amazing in conception, with a precision and a certainty that are astounding.

It is an exceedingly sensitive and accurate instrument - the product of co-ordinated efforts of engineers and craftsmen.

Its construction requires materials from every corner of the earth.

Its use is world-wide.

Its future possibilities, even in the light of present day accomplishments, are but dimly forseen; for each development opens
new fields of design and application.

The importance of the Electron Tube lies in its ability to control almost instantly the flight of the millions of electrons
supplied by the cathode. It accomplishes this control with a minimum of energy.

Because it is almost instantaneous in its action, the electron tube can operate efficiently and accurately at electrical
frequencies much higher tha those attainable with rotating machines.

An Electron Tube consists of a cathode, which supplies electrons, and one or more additional electrodes, which control and
collect these electrons, mounted in an evacuated envelope.

The envelope may be made of glass, metal, ceramic, or a combination of these materials."
 
3.2    DESIGN

In its simplest form, an electron tube consists of a cathode (the negative electrode) and an anode or plate (the positive
electrode) in an evacuated sealed envelope.

The envelope may be made from glass, metal, ceramic, or a combination of these materials.

More complex types may also contain one or more additional electrodes which control and collect electrons. The electrodes
are encased in the evacuated envelope and have the necessary connections brought out through air-tight seals.

The air is completely removed from the envelope to allow free movement of the electrons, and to prevent injury to the
emitting surface of the cathode.

The purpose of the cathode is to furnish a continuous supply of free electrons; the plate collects these free electrons.

Now, rate control requirements in electrical circuits range from an occasional on-off switching to continuous variations
occuring several billion times per second.

The rate at which electrons are collected by the plate (the plate current) is determined by the the number of free electrons
available and by the polarity and the strength of the electric field between the plate and cathode.

Tubes that provide this form of control are known generically as "amplifiers".

Power tube amplifiers are capable of controlling relatively large amounts of energy.

(Note: All triode and multi-grid power tubes are inherently rectifiers as well as amplifiers, because they deliver uni-
directional current regardless of the kind of energy provided by the power source.)

Power tubes and rectifiers are usually operated so that the number of electrons available is constant. Consequently, the
rate of collection or current flow is determined by the characteristics of the internal electric field.
 

WARNING - VACUUM TUBES OPERATE AT HIGH VOLTAGES THAT CAN BE FATAL

OR CAUSE PERMANENT PHYSICAL MALFUNCTION OR DISABLEMENT.

DO NOT ATTEMPT TO CONSTRUCT  OR WORK ON AN AUDIO AMPLIFIER UNLESS

YOU HAVE BEEN TRAINED OR ARE OTHERWISE QUALIFIED TO DO SO - WE WANT
YOU TO LIVE TO ENJOY THE FRUITS OF YOUR LABOUR.
 

3.3    OPERATION

When the cathode is heated, electrons leave the cathode surface and form an invisible cloud in the space around it.

Any positive electric potential within the evacuated envelope offers a strong attraction to the electrons (unlike electric
charges attract; like charges repel).

Such a positive electric potential can be supplied by an anode (plate) (positive electrode) located within the tube in
proximity to the cathode."

The Vacuum tube is called a "Valve" is some countries because it performs the traditional function of the valve - ie a valve
controls or regulates the flow of fluid in a device, usually a pipe or tube, by limiting the volume of flow of fluid to a pre-
determined proportion, or percentage, of the maximum possible rate.

A "valve" is also a "regulator". The function of a "regulator" is to limit the rate, pressure or volume of flow to within a range
controlled within pre-determined limits.

The current flowing through a vacuum tube is regulated by the external circuitry - to control the current flow between pre-
determined limits.

As we will see in this paper, electrical current behaves just like a fluid and this analogy will be referred to several times.
 
 
4.     HOW VACUUM TUBES WORK
4.1    BASIC CONCEPT

The Vacuum Tube is an extremely simple device.

In its simplest form, the 'diode", the vacuum tube comprises a cathode and an anode mounted in a vacuum chamber.

A high direct current voltage is applied between the anode and cathode. The cathode is heated to stimulate it to emit
electrons. The electrons are then free in space in the vacuum tube, and are attracted to the anode, setting up a stream of
electrons, just like water in a pipe.

In a diode the volume of electrons flowing is controlled by the load in the circuit. Diodes are therefore not "valves".

The maximum current that can flow in the circuit ("prospective current") is determined by the capacity of the mains, battery
or generator supply limited by the sum of the internal impedances (reactance) of the vacuum tube, circuit components (eg
transformers) and the load.

Diodes have limited application and are therefore mainly used to rectify (convert) alternating current to direct current, in
which application they offer outstanding performance and reliability if used correctly.

Vacuum tubes are not very efficient, giving up about half the input energy as heat, transferred to the atmosphere and
mounting components by convection and conduction respectively. Electrons generated by the hot cathode but not used by
the load are also released as heat, described as "plate dissipation".

In directly heated tubes the cathode is heated by itself and is called a "filament" because the electrons are emitted directly
from it, just as in a light bulb. Filament warm-up time is quick so current flows soon after the tube is switched on.

In indirectly heated tubes - ie where the cathode is physically separate to the heater, the cathode is called a "cathode", to
describe its true function. Cathode warm-up time is slow, so current flows some time after switch on - usually controlled by
tube design to about 11 seconds from cold to full warm up.

For a comprehensive explanation of how tubes work see 1952 Text US Army TM11-662 and Airforce Text TO16-1-255
 
 
4.2    ELECTRON TUBE TYPES

Electron Tubes are designated by the number of electrodes used in their design and construction.

In directly heated tubes the "filament" is classed and counted as an electrode.

In indirectly heated tubes, the "heater" is not classed or counted as an electrode.
In most cases except the "monode" the Electron Tube contains an ANODE ( the "plate") and a CATHODE (the directly
heated "filament" or indirectly heated "cathode"). The remaining electrodes are called "grids" and are used to control the
flow of electrons between cathode and plate.
Certain types of gas filled tubes may vary slightly from the above configuration.

a)    MONODE  - Single Filament Electrode eg LIGHT GLOBE or LAMP. (Not usually classed as an "Electron Tube" even
though they are typically a vacuum tube.)

b)    DIODE - Two electrodes

c)    TRIODE - Three electrodes

d)    TETRODE - Four electrodes

e)    PENTODE - Five electrodes

f)    HEXODE - Six Electrodes

g)   HEPTODE - Seven electrodes

h)   OCTODE - Eight electrodes

i)    BEAM POWER TUBE - Further refinements produced the Beam Power Tube, which may be a tetrode or pentode
whose performance is enhanced by mechanical manipulation of the electron beam, to produce substantially greater
efficiency, power output and reduced distortion.

Examples of  the Beam Power Tube are demonstrated in the 6L6 family of designs, which include the 6L6GT, 807, 1614,
1625, 5881, 7027A, 7581A/6L6GC and KT66.

Other Beam Power Tubes include 6AQ5, 6CZ5, 6DZ7 (2 x 6BQ5), 6V6GT, 6005, 6550/KT88, 6973, 7581A/KT66, 8417,
KT66 and KT88.

These tubes evolved from the 42 and 6F6 family which were pentodes. Comparison of their performance shows reduced
heater power requirements and substantially improved performance.

Please note the 6BQ5/6DZ7/EL84/7189, 6CA7/EL34, and 7591/7868/6GM5 families are not Beam Power Tubes but are
Pentodes.

Beam Power Tubes may generally be identified by a box-like structure on top of the plate assembly, which extends down
inside the tube between the plate and the control grids to direct electron flow.

For an overview of Beam Power Tube design and application see RCA Beam Power Tubes
 
 
4.3    TUBE FUNCTION

Electron Tubes may be used in a very wide range of applications.

In audio amplifier and modulator applications the primary functions of Electron Tubes are:

a)    Rectifier
b)    Voltage Stabiliser or Regulator
c)    Voltage Amplifier
d)    Phase Splitter or Phase Invertor (push-pull circuits only)
e)    Power Output

Tube signal output is controlled by the concurrent application of direct current (DC) and alternating current (AC) voltages to
the electrodes of the tube.

Most electron  tubes are provided with pins to enable insertion into or removal from a socket, to which external control
circuit wiring is attached, however some more modern types have wires instead of pins, to enable direct wiring into
apparatus.

Despite the socket/pin system suggesting unreliability, electron tubes can provide reliable service for many, many years -
depending upon a range of factors including circuit design.

It is common for electron tubes that have been in storage for 60 years to reliably operate first time when energised.

The electron tube is truly one of the most important devices to benefit mankind
ever invented.
 
 
 
 
 
 

electron's home page

 
 
 

Page last revised 07 April 2012

This page is located at http://www.oestex/com/tubes/electron.htm

 
 

electron's
 

TRIODE OPERATION
of
AUDIO and RF MODULATOR
POWER TUBES
 
 

Readers, please note this page is presented for your education, information and guidance only.

This paper refers only to the characteristics and performance of push-pull tube audio amplifiers without negative feedback.

For reasons detailed elsewhere in my website I have no interest whatsoever in either single-ended amplifiers or trans-stage
negative feedback.

For full ratings and applications of specific tube types in which you are interested please refer to the manufacturer's catalogue.

Copyright in all quoted works remains with their owner, author and publisher, as applicable.

Please note that no warranty is expressed or implied - see footnote notice.

The whole or part thereof of this paper and/or the designs and design concepts expressed therein may be reproduced for personal
use - but not for commercial gain or reward without the express written permission of the author.

All rights reserved.

1.    TRIODE OPERATION OF POWER TUBES

1.1    TRIODE TONAL CHARACTERISTICS

It is well established that many listeners prefer the "soft" and certainly pleasant tonal characteristics of triodes.
However triodes suitable for audio power operation are becoming extremely scarce in the world market and aftermarket prices for
popular types have increased astronomically in recent years.

So it is often more convenient to use tetrodes, pentodes or beam power tubes because of availability, cost or convenience.

An often overlooked design feature is that most popular audio power triodes have directly heated filaments, thus creating a
potential for problems in dealing with hum, reliability and power supply - whereas most popular tetrodes, pentodes and beam
power tubes incorporate indirectly heated cathodes - ie use separate heaters.

Most tetrodes, pentodes and beam power tubes can be successfully operated as triodes in either voltage amplifier or power tube
applications, by connecting Grid #2 directly to the Plate.

In the case of triode operation of pentodes and beam power tubes, Grid #3 or the electron beam shield respectively, are still
usually connected to the cathode or filament, as is the case for normal tetrode or pentode connection.

The characteristics of a pentode/ beam tetrode as a triode are largely determined by the Grid#1/Grid#2 amplification factor which is
dependant not on the Anode/Cathode spacing but the Grid#1/ Cathode and Grid#2/Cathode spacings.

The function of the Plate in a triode connected multi-grid tube is merely to collect all the electrons that have been attracted by the
Screen-Grid acting as an ANODE which, because of its open structure, will collect relatively few electrons - which probably
accounts for the increased allowable screen voltage when so connected.

My personal view is that to ensure tube operation - including Cathode Current, Plate Dissipation and Screen-Grid Dissipation - is
maintained within design limits, the applied DC Plate and Screen voltages should never be greater than the rated Screen-Grid
Voltage.

1.2    SUITABLE TUBES

The Radiotron Designers Handbook 4th Edition 1953 says:

Manufacturer's data is available for triode operation of some popular types of tetrode, pentode and beam power tubes such as
EL34/6CA7, 6V6GT, KT66, KT88, 807, 6146. However, typically they are pragmatically assigned a significantly higher Grid #2
rating than for either tetrode or pentode connection.

The higher triode connected manufacturer's rating should only be used with great caution as it increases the risk of
thermal problems and parasitics.

When tetrodes and pentodes are connected as triodes care must be taken to ensure the Plate and Screen Voltages are not too
low, to prevent the Screen-Grid from becoming the primary anode and thus attracting too many electrons, thus exceeding Screen
Dissipation rating and fusing the Screen-Grid.

For efficient and reliable operation, a minimum Plate and Screen B+ Voltage of at least half the rated Plate Voltage is therefore
recommended. This obviously limits the choice of tubes to those tubes having an adequately high Screen-Grid Voltage rating.

In my opinion, the most likely candidates for triode connection are the:

6AQ5                        (275 VDC)        7 pin

6V6GT                      (275 VDC)        Octal
6CZ5/6973                 (285 VDC)        9 pin
5881/6L6GC/7027A    (400 VDC)        Octal
KT88                         (550 VDC)        Octal
813                           (1100 VDC)       Giant 7 pin

All are well proven fine quality beam power tubes, each famous in its own right.

However with more careful selection and operating conditions, there are many other fine tetrodes and pentodes that are capable of
excellent triode operation - such as the EL34/6CA7 (425 VDC) (Octal) and EL84/6BQ5 (300 VDC) (9 pin) pentodes.

1.3    OTHER OPTIONS

STC BRIMAR, in their Valve and Teletube Manual #8 (1959) state:

"Where Pentodes are connected as Triodes, the Suppressor Grid should be connected to the Cathode, unless otherwise stated"
(end quote)

However as far back as 1940, RCA state that the Suppressor Grid (Grid #3) in Pentodes may be connected to the Screen Grid
(Grid #2) to operate Pentodes as Tetrodes. It follows that in Triode operation of Pentodes, Grids #2 and #3 can be strapped to the
Plate if preferred.

Unfortunately, it will be noted that many of the popular Pentodes, such as the 6BM8, 6BQ5/EL84, 6GW8 and 6M5, have their Grid
#3 internally connected to the Cathode, leaving the designer only Grid #2 available to connect to the Plate for triode operation.

It may well be however that in those Pentode tubes (other than Beam Power Tubes described as "Pentodes" in British and
European catalogues) having a separate pin connection for Grid #3 - eg EL34/6CA7, 803 and 837 - some benefit might be
achieved by connecting Grid #3 to both Grid #2 and the Plate to form a composite equi-potential tube element for triode operation.

This option would certainly ensure a smooth flow of electrons through and beyond the Screen-Grid. Any electrons reflected from
the Plate (Secondary Emission) should get caught up in the main electron stream and not present a problem.

No data sheets are available for this class of operation so some degree of experimentation would be essential.

I have no direct applications experience or information to verify this option.

1.4    Triode Operation - Audio Signal Output:

Of fundamental importance in audio amplifier design, is that when tetrodes, pentodes and beam power tubes are
configured in triode connection, the Screen Grid SIGNAL voltage will ALWAYS appear in the Plate circuit because the
Screen Grid is connected directly to the Plate - ie the electrons collected and diverted by the Screen Grids will appear in
the output of the amplifier under all conditions between zero signal and full signal input.

Thus any distortion and/or effects upon frequency response and linearity of the AC signal in the Screen Grid circuit of
triode connected tetrodes and pentodes will appear in the output of the amplifier.

This phenomenon also happens in ultra-linear connection - but only under linear controlled conditions determined by the output
transformer primary turns and impedance ratios - that are quite different to Screen Grid behaviour under triode connection.

However this phenomenon does not happen with conventional tetrode and pentode configurations because the Screen Grid signal
is bypassed to ground and does not appear in the output signal at all - even with negative feedback from the loudspeaker.
Consequently tube behaviour and amplifier performance are very different when a particular tube type is configured in each of
tetrode/pentode, ultra-linear or triode connections.

However, one major bonus from triode operation is that the changes in relative portions of Plate Current and Screen-Grid Current -
ie signal output power - that occur with changes in Grid #1 Voltage - as illustrated in Fig. 2 above - will not have material effect on
the output signal because the Plate and Screen Currents are mixed together.

In other words, triode operation will provide a more linear output stage than tetrode or pentode configurations.

NOTE: Triode operation of tetrodes, pentodes and beam power tubes produces substantially less power output than when used in
their design configuration - ie for any given Plate voltage, the power output from a triode will be substantially less than that for a
tetrode, pentode or beam power tube.

It should also be noted that Grid #1 bias voltage will be proportionately higher for triode connection, hence driving voltage will be
also proportionately higher. For most applications this increase could be substantial, necessitating a more complex driving stage
arrangement than would otherwise be the case.

The tube manufacturer's data sheets should be consulted before proceeding with an amplifier design commitment.

1.5    Triode Operation - Grid Stopper Resistor:

It is important to retain the grid stopper resistor between Plate and Screen Grid, to prevent parasitic-oscillation in the tube. The
grid stopper resistor must be mounted as closely to the tube socket pin as practicable to minimise RF signal pickup and minimise
inductance in the wiring.

The usually recommended value is around 100 to 500 ohms (although many commercial amps connect directly, with no Grid
stopper at all) however an empirical approach derived from the manufacturers' data described on my ultra-linear operation page
suggests a value of around at least one half the Plate to Plate load impedance presented by the output transformer.

A value (per-tube) of half the Plate to Plate load impedance might be a good starting point to give the electrons sufficient incentive
to travel the further journey to the Plates instead of taking the easy path home. More resistance may be required though to achieve
the desired effect - after all, electrons are lazy critters and always take the easiest route they can, even if it means increasing
density - ie crowding into a small area of a conductor or electron beam. For those mechanically minded, electron flow behaves like
a fluid.

Even if the Grid Stopper resistor has a value equal to the Plate load impedance, the current flow in Grid #2 will still be significant.

Unfortunately, the tube handbooks tell us that the Screen to ground return path must be of low AC low impedance. One way out of
this is to shunt the Grid-stopper with a suitable value of capacitor so that the Screen is grounded via the Plate which is, of course,
connected to ground through the load, but doing this will also reduce the AC separation sought between Screen Grid and Plate -
taking us back to conventional triode connection and the less than ideal Screen Grid operating parameters.

A useful addition is a small bypass capacitor (say 100 pF) shunting the Screen Grid voltage dropping resistor to eliminate any stray
RF.

2.    TRIODE OPERATION WITH SILICON DIODES

A significant practical improvement may be gained by installing a silicon diode between the Plate and Screen-Grid in the manner
described in my OPTIMISED ELECTRON STREAM © TECHNOLOGY  page - ie with the arrow pointing towards the Screen-Grid 
- instead of connecting the Screen Grid directly to the Plate, as is the conventional practice.

It offers the same benefits as in tetrode/pentode connection by removing Screen Grid output signal from the Plate circuit - ie the
Tetrode or Pentode in Triode connection behaves like a real Triode, with electron flow direct from Cathode to Plate, as if the
Screen Grid did not exist.

This configuration enables us to use beam power tubes, with all their benefits, as straight triodes, thereby attaining all the benefits
of beam power tube operation but with the tonal and distortion qualitites of triode operation.

The Grid Stopper resistor and/or Screen Grid voltage dropping resistor must still be kept in the circuit to control DC voltage
behaviour because in triode connection there is far too much DC on the Screen Grids than is needed to do their job. In this case,
less is thus better than more.

Some adjustment may need to be made to the Grid #1 bias voltage to limit Plate current to the design value.

Experiments I have undertaken in this area have resulted in substantially improved high-frequency performance, more natural
sound and superior tonal characteristics compared with the conventional approach.
In my opinion, the most likely candidates for triode connection with silicon diodes feeding the Screens are the:

6AQ5                        (275 VDC)        7 pin

6V6GT                      (275 VDC)        Octal
6CZ5/6973                 (285 VDC)        9 pin
5881/6L6GC/7027A    (400 VDC)        Octal
KT88                         (550 VDC)        Octal
813                           (1100 VDC)       Giant 7 pin

All are well proven fine quality beam power tubes, each famous in its own right.

Most other tetrodes, pentodes and beam power tubes have too low a Screen-Grid voltage rating to be suitable for triode operation
at full rated Plate voltage.

There are many fine tetrode, pentode and beam power tubes that can be used in triode operation, however care is necessary to
avoid applying excessive voltage to the Screen-Grids - such as the EL34/6CA7 (425 VDC) (Octal) and EL84/6BQ5 (300 VDC) (9
pin) pentodes.
 

Note: In high power applications - ie more than 100W RMS, transmitting triode tubes such as 805, 809, 810, 811, 812, 833, 845,
8000, 8005, etc may be a more economical and practical solution than pursuing a triode connection configuration with tetrodes or
pentodes because of a wider choice of suitable tube types and characteristics, simplified wiring and power supply requirements.
Most large tetrodes and pentodes have a relatively low rated Screen-grid operating voltage, indicating a high risk of problems when
connected as triodes.

At the present time, large NOS triodes generally cost about the same as large NOS tetrodes or pentodes offering equivalent power
output.
 

3.    TRIODE OPERATION USING THE SCREEN GRIDS AS THE

CONTROL GRID
This mode of operation offers yet another alternative for the application of Pentodes, Tetrodes or Beam Power Tubes as Triodes.

The method is simple.

Connect the Control Grid #1 to the Cathode

Use Grid #2 as the Control Grid.

This approach requires substantially higher DC negative bias voltage and considerable AC drive signal input voltage than as per
the standard configuration, but offers the benefit of allowing full rated Plate Voltage to be applied.

This is because the Screen Grid does not attract electrons - in fact it is negatively charged so cannot.

Power output should thus be similar to that obtainable in Tetrode or Pentode mode - but with the tonal characteristics of Triodes.
This is because the tube is now a real triode - ie only 3 effective elements (Cathode, Control Grid and Plate).

I am unable to provide details of typical operating conditions but this method has been used in professional broadcast standard
equipment and in theory should be more reliable than the standard approach.

Class AB2 or B with fixed bias is essential for high Plate circuit efficiency and dynamic performance.

If you have experience with this configuration please advise your results.

 
 
 
 
 
Please let me know if you can add to this body of new knowledge and I will add it to this commentary.

Of course the smart thing to do here is to use tubes that are already triodes - instead of messing about with compromises - but
then none of the triodes are beam tubes and thus do not offer the benefits of  beam tube technology.

Also most of us have a junkbox stock of perfectly good Tetrodes, Pentodes or Beam Power Tubes just waiting to be used - so it is
a tough call.

REMEMBER:
- ALWAYS TAKE CARE WHEN WORKING WITH HIGH-VOLTAGE -
DEATH IS PERMANENT!!

 
 
 
IMPORTANT NOTICE

THE AUTHOR MAKES NO CLAIM WHATSOEVER AS TO THE VALIDITY OR ACCURACY OF ANY

STATEMENT, INFORMATION OR OPINION CONTAINED IN THESE PAGES AND NO LIABILITY WILL
BE ACCEPTED FOR ANY ERROR OR OMISSION OF ANY KIND WHATSOEVER.

PLEASE NOTE NO WARRANTY IS EXPRESSED OR IMPLIED AS TO THE WORKABILITY OR

PERFORMANCE OF DESIGN INFORMATION DESCRIBED HEREIN.

For suggestions, critique or discussion re this page contact:

Dennis Grimwood
 
 electron's

 
 

SCREEN GRIDS in
AUDIO and RF MODULATOR
POWER TUBES
 
 

Readers - please note this page is presented for your education, information and guidance only.

This paper refers only to the characteristics and performance of push-pull tube audio amplifiers without trans-stage or loop
negative feedback.

For reasons detailed elsewhere in my website I have no interest whatsoever in either single-ended amplifiers or trans-stage
negative feedback.

For full ratings and applications of specific tube types in which you are interested please refer to the manufacturer's catalogue.

Please note that no warranty is expressed or implied - see footnote notice.

The whole or part thereof of this paper and/or the designs and design concepts expressed therein may be reproduced for
personal use - but not for commercial gain or reward without the express written permission of the author.

© Copyright: Dennis R. Grimwoood - All rights reserved.

Copyright in all quoted works remains with their original owner, author and publisher, as applicable.
 

1.    INTRODUCTION
Traditionally, the design of audio amplifiers has followed fairly clear and well established design principles.

Some of those principles relate to the way in which Screen Grids are used to control current flow in audio amplifier tubes,
particularly power tubes.
Examination of professionally designed commercial circuits spanning more than 60 years' audio technology shows us there has
been very little innovation in the way in which Screen Grids are used - ie little variation in, or departure from, conventional,
traditional Screen-Grid application design concepts.

It is understandable why this is so, because innovative engineering was not encouraged in the consumption driven expanding
global marketplaces of the 1940's through 1970's.

The post WWII market - ie the 50's and 60's era - was one of explosive growth and expansion in consumer and industrial
demand, so it was primarily a seller's market. The market's natural wariness towards "way out" designs was high, so
unconvention was not generally pursued. Few equipment manufacturers were brave enough to vary from the tried and true.
Thus the prevailing audio design ethos was to "follow the leader".

In any event, unconvention often resulted in premature component failure, bad reputation, consumer wariness or rejection, and
typically accompanied by an increase in manufacturing cost - and therefore selling price, with little perceivable benefit to the
consumer.

Furthermore, programme materiel available to the consumer ex radio, phono or tape was generally of such quality that even the
"critical listener" consumer was unable to discern audible differences between "good" and "superior" amplifiers.

Numerous documented scientific experiments since the 1930's demonstrated that most listeners were unable to discern the
difference between a live and recorded performance from behind a screen. To the masses, there was no difference, so why pay
more?

Audio amplifier equipment design became more or less a "variation on the theme" exercise in applications engineering, in much
the same way as we now see design technology expressed in the configuration of CD players, DVD players and personal
computers.
 

Vacuum Tubes in Amplifiers:

A very critical factor in tube amplifier design is the vacuum tube itself.

Electron tube manufacturing tolerances and acceptance test specifications are fairly wide, particularly in some types such as
large power tubes, resulting in an audio amplifier design requirement for individual tubes to be individually adjusted, or "tuned" to
specific circuit parameters for optimum performance - such as in high-power push-pull applications.

However amplifier manufacturers were reluctant to incorporate simple user adjustments into their products because that tempts
(and provides the means for) the curious user to "play" with optimising controls such as Grid #1 bias or Grid #2 regulated
supply, thereby ensuring poor performance, overheating, premature failure, or even self-destruction of the amplifier. Extra
optimising adjustments also add considerable manufacturing cost to a base product, imply the amplifier is "dicky" or
"temperamental", reduce reliability, and may offer the end user no real perceptible benefit apart from setting up the product to do
what it is supposed to do in the first place and what alternative products do (or appear to do) without adjustments.

Although the absence of adjustments may lead to less than optimum performance, it does generally provide the consumer with
a more reliable piece of equipment. One disadvantage, commonly found in parallel-push-pull amplifiers, is that 4 tubes or more
may be supplied by a common bias supply, necessitating carefully matched tubes for reasonable dynamic performance and
reliability. However, this arrangement ensures that whenever a single tube requires replacement, all four must be replaced
together to preserve balance. During the 1950's thru 1980's, obtaining an accurately matched set of tubes was often a
challenging task.

However in the long run, this simplified approach to tube selection provided the preferred choice for a safe solution and
equitable warranty protection to both manufacturer and user.

Since the 1950's, the choice between cathode bias or fixed bias was often determined by the lower manufacturing cost of
cathode bias and the self-protecting effect of cathode bias, so fixed-bias tended to be used only where high power was needed
such as in public address or guitar amplifiers.

Cathode-bias was a natural evolution from the "back bias" used widely in early radio receivers, where the field coil of the
loudspeaker served both as a filter choke and a convenient DC bias ("C" supply) voltage source. When the field-coil loudspeaker
was replaced by the cheaper to manufacture "permanent magnet" style loudspeaker around the early 1950's, back-bias designs
more or less disappeared from the face of the earth.

Back-bias requires more complex wiring than cathode-bias so it was seen as an unnecessary complication to wiring costs, for
no realisable benefit to the end user.

Cathode-bias also offers inherently higher reliability than back-bias or fixed-bias, as well as providing a self-compensating
effect for aging tubes. Cathode-bias also offers a lesser Plate Current swing from zero to maximum signal, thereby enabling
power supplies having poor regulation to be incorporated with no apparent reduction in tested performance.

However, even with the simplicity of cathode-bias, many manufacturers still incorporated a single common bias resistor for at
least two tubes in push-pull, resulting typically in tube mismatch - ie still requiring a matched pair of tubes for optimum
performance.

The advent of loop negative feedback - ie negative feedback from output to input - further assisted some manufacturers to
provide even poorer power supplies and driver stages, because audible hum could not be heard.

Loop negative feedback also facilitates the use of poorer quality lower-cost output transformers and wider tolerances on tubes
and components, relying on the feedback to restore performance to an acceptable standard.

These "improvements" resulted in a performance situation where some amplifiers with feedback performed no audibly better
than earlier design amplifiers without feedback - except under steady state conditions on the test bench into constant resistive
loads.

The later introduction of silicon rectifiers and voltage doubler power supplies enabled further cost-reduction at the expense of
transient performance. Advantages such as substantially improved power supply regulation gained from silicon rectifiers over
tube rectifiers were soon offset by cost-saving measures.

The traditional filter choke was an early casualty of cost reduction. Good quality 1940's and 1950's amplifiers used a full-wave
rectifer and a two stage choke input filter, however this progressively degenerated to the point where many popular amplifiers of
the 1970's had a voltage doubler power supply with no filter choke at all, relying on the combined effects of larger electrolytic
filter capacitors, loop negative feedback, and push-pull hum cancellation to produce an acceptable product.

Very few amplifiers included regulated power supplies for their Screen-Grids, because of increased manufacturing cost.
Although the RCA Receiving Tube Manuals published schematics incorporating Screen-Grid regulation, the most common
configuration was that the Screen-Grids were fed directly from the B+ supply - ie at high-voltage, often without any Grid-stopper
resistor. One variant was to use a dropping resistor and filter capacitor, from the B+ to supply the Screen-Grids, but this
arrangement results in poor Screen-Grid voltage regulation, with attendant drop in performance.

In other words, in an attempt to hold-down manufacturing costs over time, some tube amplifier manufacturers actually took the
audio industry backwards in terms of performance evolution.

So, apart from the highly acclaimed triode connected Williamson (D.T.N. Williamson 1947 and 1949), followed by the
magnificient tetrode connected U.S. McIntosh (F.H. McIntosh and G.J.Gow 1949) and later (but much inferior) U.K.Quad II
(1953) amplifier, with their "unity coupling" power output stage, and the original  Ultra-Linear (D. Hafler and H. I. Keroes of Acro
- 1951) design; there is little to show for 60 odd years' of progressive global technological evolution in tube audio.

Note: History shows the term "ultra-linear" was developed by Hafler and Keroes to define their specific configuration based upon
the original 1936 British Patent of A.D.Blumlien, which was fairly generic in respect to transformer ratios. It appears that his
innovative design was neither refined nor exploited commercially during the 14 year life of his patent, noting the British Quad II
amplifier of 1953 bypassed this opportunity, however some researchers suggest it had previously been used in Australia as far
back as 1933 - a fact that if true would have invalidated his patent. Perhaps the military demands of the WWII years diverted Mr.
Blumlien's attention to his prolific innovative design activity covering a wide range of other technologies and the ultra-linear
concept was left to others to further develop and exploit - however audiophiles remain indebted to his contribution. The wheels of
technological progress sometimes turn slowly!!

It is relevant that all these designs relied heavily for their final performance upon extremely high quality output transformers - in
the case of the McIntosh, bifilar windings (primary and secondary windings were wound together with no insulation between
them or between layers, requiring very high quality winding wire and winding techniques) and fully potted construction were
featured (a remarkable engineering achievement) - so manufacturing expense increased substantially in any event.

Despite the current raves for single-ended push-pull concepts, commercial attempts to exploit that particular technology
inevitably failed in preference to convention. One approach by the Dutch Philips group in the 1960's, used an output-
transformerless (OTL) single-ended push-pull amplifier connected to an 800 ohm Philips loudspeaker, requiring the consumer to
purchase a complete system from the one supplier - not a popular concept for modular hi-fi component buyers (particularly
those who already owned a fine set of loudspeakers), thus relegating this new technology to the mass consumer market -
thereby destroying its appeal to the audiophile. This technology faded into obscurity along with demise of the "radiogram" all in
one system.

Inevitably, all attempts to depart from proven simple audio circuit design principles resulted in increased cost, reduced reliability,
increased downtime and service costs, and consumer anger.

The realities of global markets and a long way to a competent service shop resulted in manufacturers being forced by
circumstances to limit their experimentation - or experiment to discover that alternatives to conventional design simplicity were
not commercially viable products. Most manufacturers were limited to sourcing components from a small pool of suppliers so
manufacturing costs were similar across the industry. Designs had to be both simple and cost-competitive.
 

Top Cap Tubes:

Another factor that produced suppression of innovation was the swing away from tubes for audio applications that incorporated
top caps for their plate, or anode, connection.

Users often found themselves "zapped" when changing a tube, by inadvertently touching the cap lead or terminal - particularly if
the amplifier was switched on - a most unpleasant experience.

Long Plate leads also present problems with induction to and from from nearby components, stray RF pickup, output stage
instability, transformer mechanical construction and chassis layout.

Although widely used in professional broadcast and public address applications during the 1940's and 1950's, top cap style
tubes - such as the 6146/QE05-40, 6DQ6A, 6CM5/PL36, 5B/254M, and the great 807, have not been popular for hi-fi or guitar
amplifier applications - the largest commercial market segments for tube use in applications greater than about 5 W RMS
output. Thus this style of tube, which offers considerably higher power outputs than no-top-cap standard octal socket styles, or
all glass 9 and 12 pin tube types (eg 7868), has been little used after 1955 in hi-fi and guitar amplifier designs (although still
extensively used in television receiver applications until the 1970's).

This pragmatic design philosophy forced tube manufacturers to develop tubes that produced more power from a conventional
(usually octal based)  tube having no top cap - in a valiant effort to put amplifier performance back to where it had already been.
Result - the EL34/6CA7 and KT88, both practically limited by the dielectric strength of the octal base and socket to about 600
VDC B+ supply - but both needing high Grid #2 ratings to offset the limited plate voltage as a means to retain adequately high
power output.

It is of interest that the KT88 is identical to the TT21 transmitting tube, which has a rated Plate Voltage of 1.25 kV applied to the
top cap connection. In the KT88, the Plate connection is relocated to the octal base. This modification results in a maximum
rated Plate Voltage of 600 VDC for the KT88.

For 250 to 300 VDC supplies, there are also the EL84/6BQ5/7189, and the 6V6GT, 6AQ5/6HG5/6005 and 6CZ5/6973 families.

However, in all these types, analysis of manufacturers' data shows the proportionately high Screen Grid voltage needed to obtain
maximum power output results in substantially higher harmonic and intermodulation distortion than seen in conventional RF
beam power tubes combining high plate voltage with relatively low Grid #2 voltage for the same audio output power- eg typically
4 to 5% instead of 1 to 2% THD without negative feedback.

The suitability of the EL34/6CA7 and EL84/6BQ5 to ultra-linear connection offsets this disadvantage somewhat, albeit at reduced
power output, but the original 6L6 family are not so fortunate being practically limited by their lower Grid #2 rating.

The original GEC KT88 thus became the only tube to offer a reasonable solution, providing up to 100 W RMS per pair, however
they were expensive, of widely varying quality, required substantial free-air space for ventilation, supporting componentry and
circuitry of professional broadcast standard, and were really a little large for an octal socket to support. Being heavy, the KT88
is not suited to inverted mounting (eg guitar amplifiers) without supporting straps to prevent them falling out of their sockets.
However from the outset (about 1960), transistor amplifiers were easily able to match this performance (on paper) in a
substantially cheaper, smaller, lighter and more reliable package, so the KT88 was soon displaced in the mass market.

In some industries that were high consumers of vacuum tubes, particularly in guitar amplifiers, there is also clear evidence that
tube designs were enhanced to cater for limitations in the final product. That old favourite, the 6L6, has been upgraded over and
over again, even though superior top cap versions (eg 807 and 1614) were available from the outset - albeit at significantly
higher cost.

Manufacturing cost, profit margins, market share and sales revenue were each in their own right, powerful design engineering
drivers and inhibitors.

Standard domestic quality driver tubes such as 6SN7GT, 12AT7, 12AU7A etc triodes and their popular pentode cousins, 6SJ7,
6AU6, 6U8 and EF86, have hardly changed throughout the 60 years since they were first released. Later improved "premium
quality" versions rarely found their way into commercial audio amplifiers, primarily because they cost more, offered no detectable
audible benefit to the listener, had electro-mechanical characteristics that provided in practice properties or performance only
marginally different to the standard tube - if at all (eg rattles and microphonics in "premium" tubes), and frequently could not be
replaced in the country of use - after all who wants a product that cannot be repaired or likely to be out of action for many
months whilst waiting for an expensive imported tube to arrive? Not only that, but the replacement cost of a premium quality
tube was often many times the cost of the equivalent standard type.
 

Summary:

So a review of commercial circuits shows that for the whole of that 60 year period between 1940 and 2001, only a few basic
types of tubes were used in all the audio amplifiers ever produced in the whole world.

The result is that:

1.    there is very little literature about Screen Grids

2.    there are are few examples of innovative design variants
3.    audio amplifier design standards reflected the need for simple tubes that could be overloaded and abused by users
4.    audio amateurs - ie hobbyists and project builders - have had to remain within a very rigid published design framework
5.    published manufacturer's tube data invariably fails to provide information about the effect of Screen Grid voltage upon
       Plate Current
6.    there is little published manufacturer's data available for non-popular tube types
7.    there is little practical knowledge available to facilitate experimentation with non-popular tube types
8.    a self-destructing commercial approach manifested that inhibited innovation in the tube based audio equipment
       industry, paving the way for their displacement by semi-conductors

This page attempts to quantify some of the major principles and possibilities regarding improving vacuum tube
technologies in the area of Screen Grids.

I do not claim it to have any technical expertise or validity whatsoever and am happy to be challenged in the interests of mutual
learning. If you can add any information that will benefit the audio enthusiast please email it to me.

2.    THE SCREEN GRID (GRID # 2) - PRIMARY FUNCTIONS

The Screen Grid is an extra element added to the basic three element configuration of triode tubes to form a four element
configuration tube called a tetrode.

Fig. 1:    Graphical Representation Of A Tetrode Vacuum Tube

The Screen Grid is also incorporated into multi-electrode tubes such as pentodes, heptodes and octodes.

The Screen Grid is assigned the functional title Grid #2, to indicate it is the second Grid from the Cathode.

The Screen Grid usually comprises a formed coil of wound turns of round wire, mounted physically concentrically between the
Control Grid (Grid #1) and the Plate (Anode) in multi-electrode vacuum tubes and acts as an electrostatic shield between them.
 

The primary functions of the Screen-Grid are:

to act as a screen. To perform this function, the Screen Grid is positioned between, thus separates or "screens", input
circuits (Grid 1) from output circuits (Plate or Anode) within the tube, thus the name "Screen-Grid".
to accelerate and attract electrons to the Plate
to provide a means for additional control over electron flow
to reduce Grid 1 to Plate capacitance and thereby improve frequency response, linearity and operating frequency
to reduce Plate to Grid 1 feedback
increase tube efficiency
increase gain
increase output voltage
 reduce distortion (compared to a Triode)
improve linearity

The most important attribute of the Screen-Grid - but one that is not clearly explained in Tube Manuals - is that it
becomes the primary ANODE in the tube - ie when the Screen-Grid is energised to a positive potential to the Cathode,
the Plate becomes simply a passive collector of those electrons emitted from the directly or indirectly heated Cathode
that have been attracted by, and passed through, the Screen-Grid to it.

The RCA 1937 Receiving Tube Manual tells us this way at Page 9:

"The Screen is operated at a positive voltage and, therefore, attracts electrons from the Cathode. But because of the
comparatively large space between the wires of the Screen, most of the electrons drawn to the Screen pass through it
to the Plate. Hence the Screen acts as an electrostatic force pulling electrons form the Cathode to the Plate."

Thus when a Screen-Grid is present, IT is the ANODE and the Plate becomes a secondary or pseudo-anode only.

Understanding this fundamental design feature is crucial to understanding the significance of maximum Screen Grid
rated voltages and their relationship to Plate voltages in all cases for triode, tetrode, pentode, beam power tube or
ultra-linear configurations.

The first tetrodes were introduced to enable stable amplification at radio frequencies - ie to shield the anode from the grid.  As
the screen was at ground potential for signal frequencies whist slightly increasing the input and output capacities the grid plate
capacity was reduced to the point that neutralization was completely unnecessary.

However if the anode was operated with a voltage of less than the screen, dynatron oscillations could occur - which led to the
introduction of the suppressor grid , the function of it being to repel secondary emitted electrons back to the anode rather than
allowing them to be attracted back to the screen when its potential was less than the anode.

In output valves the evolution of the tetrode / pentode was needed for completely different reasons.  As efforts were made to
increase the amplification by altering the tube geometry the resistance of the tube to DC went up, so the addition of the screen
grid enabled the anode current to be relatively independent of the anode voltage.  This fact did however dramatically increase
the AC resistance, which is not always a desirable result for an audio amplifier .

In some types of tetrodes and beam power tubes the Screen-Grids are positioned in-line with the Control-Grid (ie in-line with or
behind the electron beam) and this configuration is described as "aligned Screen-Grids" - eg 6L6GC.

However in many types of pentodes and beam power tubes the Screen-Grid is not aligned and this configuration is described as
"unaligned Screen-Grids" or "non-aligned Screen-Grids" - eg EL34. (Some users report that the early Philips EL34's had aligned
Screen Grids, whereas later production ex other manufacturers were unaligned, resulting in increased tube failure and
decreased performance - food for thought.)

In some electron tube designs, the coil is wound round and in others it is wound flat. Generally speaking, tubes having round -
ie cylindrical - Plates would have Screen Grids wound on a round former, and tubes having rectangular Plates would have
Screen Grids wound on a flat, or rectangular, former. This results in an arrangement whereby the Screen Grid is generally
parallel with the conducting (electron collecting) portion of the internal Plate surface and the conducting (electron emitting)
surface of the Cathode.

The usual arrangement is for the turns on the Screen Grid to be evenly spaced, however sometimes variable spacing is used
for particular effect - such as in "remote cutoff" tubes.

Of importance to this paper is the physical spacing, pre-determined during manufacture, between the Cathode (negative
terminal of the tube) and Screen Grid, and Screen Grid to Plate (positive terminal of the tube) - ie the relative positioning of the
Screen-Grid between the cathode and anode of the device.

Philips introduced the penthode - (as that is how they spelt it) - but Mazda played with a critical distance tetrode or, as they
referred to it, a "kinkless tetrode" , which works as per the description of a beam tetrode which, as a result of aligned grids and
beam forming plates, simply concentrated the cloud of electrons which behaved as a suppressor grid. (this contribution thanks to
Denis Cook)

The spacing between the grid and screen determines the ability of the screen in determining the characteristics of the tube. The
screen anode spacing is also of great importance in that it determines the location of the space in which the electron stream is
moving slowest and is therefore most concentrated so to act best as a virtual suppressor.
It is relevant to the concepts presented in this paper that in the case of directly heated filament tubes - ie where the
Filament is the Cathode - for practical manufacturing and cost control reasons the diameter of the Filament wire is
usually the same as that of the Screen-grid. However the length of the Screen-grid wire is usually substantially more
than that of the Filament. This means that the electron collecting surface of the Screen-grid is substantially greater
than that of the electron emitting surface of the Filament/Cathode. Consequently the Screen-grid has the capability to
collect significant numbers of electrons emitted by the lesser surface area Filament.
 

RCA Transmitting Tube Handbook TT-4 at pages 7 and 8 explains the function of the Screen Grid in this way:

"When a tetrode is used as an amplifier, the Screen Grid is operated at a fixed positive potential (usually somewhat lower than
the Plate voltage), and is bypassed to the Cathode through a capacitor having very low impedance at the operating frequency.

This capacitor diverts signal frequency alternating currents from the Screen Grid to ground, and effectively short-circuits the
capacitive feedback path between Plate and Control Grid.

The Screen Grid acts as an electrostatic shield between the Control Grid and the Plate, and reduces the Grid-Plate capacitance
to such a small value that internal feedback is usually negligible over the range of frequencies for which the tube is designed.

Because the Screen Grid is operated at a positive potential with respect to the Cathode, it collects a substantial number of
electrons and, therefore, reduces the Plate current which can flow at a given Plate voltage.

The addition of a Screen Grid thus increases the internal resistance, or Plate resistance of a tube. However, it also gives the
Grid No. 1 a greater degree of control over the Plate resistance, and thus increases the voltage amplification factor.

The voltage at which the Screen Grid is operated has a substantial effect on the Plate Current of a tetrode.

This characteristic makes it practicable to control the gain of a tetrode by variation of the DC Screen Grid potential, or to
modulate the tube output economically by the application of a signal voltage to the Screen Grid - as well as to the Control Grid.

It is usually necessary, therefore, to remove ripple and other fluctuations from the Screen Grid supply voltage to prevent
undesired modulation of the tube output."     (End quote)
 

SVETLANA say: "Regular tetrodes are rarely used for audio applications because of an effect called "tetrode kink", caused by
secondary emission. Most of it is due to electrons bouncing off the plate, some from the screen."
 
 

3.    RELATIONSHIP BETWEEN PLATE CURRENT AND SCREEN-

GRID CURRENT
 

It is important to understanding Screen-Grid function that one more aspect be considered for the case of POWER TUBES.

Compared with a Triode, the addition of the Screen-Grid to a Tetrode, Pentode or Beam Power Tube, dramatically changes the
electronic behaviour of the Power Tube.

The characteristics of a TRIODE POWER TUBE are illustrated in the following graph for the 6BQ5/EL84 Power Pentode
connected as a Triode and having a Plate and Screen-Grid Voltage of 300 VDC.
 Graph Courtesy of Philips Miniwatt Electronics Handbook (1960).

In a POWER TRIODE the Plate Current is directly proportional to Plate Voltage

In a POWER TRIODE the Plate Current is directly proportional to the Grid #1 (Control-Grid) Voltage.

In normal POWER TRIODE amplifier applications, the Plate Voltage is fixed by the B+ supply, hence the Plate Current (and
therefore power output) will vary in direct proportion to changes in the Control Grid Voltage (ie input drive AC signal). This has
the effect that small changes in signal voltage produce large changes in Plate Current.

It also has the effect that in push-pull TRIODE POWER TUBE applications, BOTH tubes must be accurately matched by
selection, test and Control-Grid Bias Voltage adjustment, to ensure both tubes amplify equally in the push-pull pair. Minor
performance differentials between tubes will produce marked results in the amplifier output signal - a good case for single-ended
TRIODE operation.

In an POWER TRIODE, careful examination of the Plate Current curves shows most POWER TRIODE amplifiers suffer from
non-linearity between low and high signal input AC drive voltages. As Grid #1 voltage increases it causes the Plate Current to
increase very rapidly, causing the Plate Voltage to decrease (by AC and DC losses in the output transformer windings and
rectifier circuit), resulting in loss of peak power at the crest of the signal voltage - ie transient signals are diminished in
magnitude by the output stage.

The above graph clearly shows that for any given value of Plate Voltage, the negative swing in signal voltage applied to Grid #1
will produce a different change in Plate Current to that produced by an equal swing in the positive excursion. This is why Class
A amplifiers must use a value of Grid #1 voltage that is sufficiently high (less negative) to enable reasonably equal positive and
negative Plate Current swings, whilst keeping within permissible Plate Dissipation limits.

It will be also observed that during the negative swing of the signal voltage, the more the Control Grid (Grid #1) swings
negatively, the less linearly Plate Current follows changes in Grid #1 voltage.

Such requirement inherently introduces some challenges in TRIODE POWER AMPLIFIER design and component selection -
particularly in Class A designs.

On the other hand, the addition of the Screen-Grid to create a Tetrode, Pentode or Beam Power Tube, dramatically changes
the electronic behaviour of the Power Tube insofaras Plate Current is no longer dependent upon Plate Voltage.
 
 

***********************************************************************************
The characteristics of a TETRODE, PENTODE OR BEAM POWER TUBE are illustrated in the following graph for the
6BQ5/EL84 Power Pentode connected as a Pentode and having a Plate and Screen-Grid Voltage of 300 VDC.

Please note this is the same tube and same applied voltages as shown in the above graph - just configured differently.

Graph Courtesy of Philips Miniwatt Electronics Handbook (1960).

The Plate Current curves for a Tetrode, Pentode or Beam Power Tube show that Plate Voltage can fluctuate markedly but does
not affect Plate Current at all. So long as the Plate Current responds to the AC signal drive voltage applied to Grid #1, then the
tube will produce a linear response to that signal.

The one requirement for this condition to be realised is that the Screen-Grid voltage be relatively constant, hence amplifier
designs using a common B+ supply to both Plate and Screen-Grid inherently lose some of the aformentioned attributes of
Screen-Grids.

This characteristic of Tetrodes, Pentodes or Beam Power Tubes offers tremendous options and benefits to the amplifier
designer.

Because POWER OUTPUT is calculated as the square of the output voltage divided by the load impedance, provided Screen
Grid voltage remains constant and Grid #1 Voltage is adjusted for correct zero signal idle current dissipation, power output may
be increased dramatically by the simple device of increasing the Plate Voltage - ie within practical limits, the driver stages can
be the same for more or less any configuration of Tetrode, Pentode or Beam Power Tube output stage.

That is to say, provided negative loop feedback is not used from the loudspeaker, any front-end can be matched to any power
stage - a most beneficial situation for the home constructor.

It will also be observed from Plate Current curves, that Tetrode, Pentode or Beam Power Tubes are generally more linear
between minimum and maximum AC drive signal conditions - particularly at the low-signal voltage end of the scale. Most high-
fidelity audio amplifiers are operated at low volume in the home hence tube behaviour at the lower end of the Grid #1 voltage
range is a critical issue - because the sound so produced is what the discerning listener hears.

On the other hand, public address amplifiers, guitar amplifiers and broadcast transmitters tend to be used at or near their
maximum output, so more interest is in the neahviour of tubes under full output conditions (where other challenges face us).
Finally, it should be noted that when a Tetrode, Pentode or Beam Power Tube is configured by wiring to "Triode Connection",
then it will behave as a Triode, with all the shortcomings (and benefits) of Triode operation.
 

*******************************************************************************************************************************************************

To consider how the Screen-Grid affects Plate Current as described above, let us examine implications of the above statement
from RCA that: "Because the Screen Grid is operated at a positive potential with respect to the Cathode, it collects a
substantial number of electrons and, therefore, reduces the Plate current which can flow at a given Plate voltage. The
addition of a Screen Grid thus increases the internal resistance, or Plate resistance of a tube." (end quote).

This is a little researched subject because in practical audio applications this relationship has been more or less of little concern
to audio amplifier designers.

However it does matter.

Tetrode and Pentode connected amplifiers always supply the Screen-Grids either from the Plate Supply (B+) or a separate
Screen-Grid supply, hence Screen-Grid Current is not usually a major consideration to amplifier designers - simply because
Screen-Grid current needs are easily met by the power supply.

There is a direct relationship between Plate Current and Screen-Grid Current which we must be aware of if we want to build
better amplifiers.

The following graph, courtesy of Philips Miniwatt, illustrates this very clearly.
  
 

Fig:2 - Philips Miniwatt 6KG6/PL509 Power Pentode for Video Applications

This rare original manufacturer's graph clearly shows us that under fixed conditions of constant Plate Voltage and Screen-Grid
Voltage, both Plate Current and Screen-Grid Current increase or decrease in response to change in Grid #1 Voltage.

That is to say Grid #2 Current is just as much affected by a change in Grid #1 Voltage as is Plate Current.

Thus Plate Current and Screen-Grid Current are a direct function of Grid #1 Voltage. This characteristic is typical for all audio
tetrodes, pentodes and beam power tubes.

However, the relationship between Plate Current and Screen-Grid Current is not linear.
In this case it can be seen that at -30 V Grid #1 bias, the Screen Grid Current is 1.7% of Plate Current, whereas at -20V bias is
1.9%, at -10 V bias is 2.5%, and at 0V is 3.3% of Plate Current.

These relativities could reasonably expected to be different with other values of Plate Voltage and Screen Grid Voltage

Although these differences may appear small, they tell us the tube is more efficient in Class A than in Class B, because the
Screen-Grid Current is a smaller portion of total current (Plate Current + Screen-Grid Current) in Class A (low Grid #1 Voltage)
than in Class B (high Grid #1 Voltage).

This phenomenon will be exacerbated by further changes in Grid #1 bias Voltage caused by the flow of Grid Current, such as in
Classes AB and B.

In other words a higher proportion of the electron stream reaches the Plate in a Class A amplifier than in a Class B amplifier.

This means that in an amplifier having characteristics that produce a variable Grid #1 Voltage there will be some offset to the
reduced power output resulting from reduced B+ supply voltage by the increased tube efficiency resultant from the change in
Grid #1 Voltage.

For example, where an amplifier has a single common power transformer supplying the whole of its power needs, increased total
current under peak signal conditions will cause reduced Grid #1 Voltage (from poor power supply regulation).

Note however, that the Philips Miniwatt 6KG6/PL509 video pentode shown above is not typical of audio tetrode, pentode and
beam power  tubes - a point demonstrated by reference to 'typical operating conditions" published in Manufacturers' Tube
Handbooks. These show that in a typical beam power tube, the Screen Grid current at maximum signal power is around 20% of
Plate current. This ratio of currents appears to be largely independent of Plate voltage.

It would therefore be reasonable to assume that up to 20% of prospective signal power is lost in the Screen Grid circuit
in a conventional amplifier.

(Note: Two notable exceptions are the 807 and 814 beam power tubes that incorporate advanced design technologies to
increase tube efficiency and reduce distortion, however in the overall sheme of things this technology appears to have been
limited to these two tube types - if you are aware of others please let me know)

The functional relationship between Plate and Screen-Grid is further illustrated by Radiotronics Magazine #90 of September
1938, which provides data for a pair of type 6L6 tubes operating as a beam power tube in push-pull Class A1, for an operating
condition having a common Plate and Screen Grid DC supply voltage.

This shows:
 

Plate and Screen Volts 100 150 200 250 290 (max)
Zero Signal Plate Current mA 32.5 55 85 120 150
Max. Signal Plate Current mA 37.5 65 100 140 175
Zero Signal Screen Current mA 2.5 4.5 7 10 12.5
Max. Signal Screen Current mA 4.2 7.4 11.4 16 20
Grid #1 Bias  V -6.25 -9.5 -12.75 -16 -18.5
Load resistance 
8000 6500 5600 5000 4600
P/P ohms
Power Output 
1.6 4.2 8.4 14.5 21.2
Watts RMS

Notice how much power output changes when the Plate and Screen Grid voltages drop from 290 to 250 - a likely situation with
a tube rectifer power supply - see rectifier forward voltage drop characteristics in manufacturer's tube handbook data.

A significant improvement to power supply regulation can be made by the simple change to full-wave silicon diode bridge
rectifier, and preferably the inclusion of at least one filter choke, which leaves only the power transformer regulation to deal with.

Notice also how the ratio of Screen Grid current to Plate current changes between zero and maximum signal and between
different operating voltages. This translates into non-linearity.
 
Plate Current - Control Capabilities of the Screen-Grid:

Let us also examine implications of the statements by RCA that :

"The voltage at which the Screen Grid is operated has a substantial effect on the Plate current of a tetrode."  (RCA
Manual TT4)

"As long as the Plate voltage is higher than the Screen voltage, Plate Current in a Screen-Grid tube depends to a great
degree on the Screen voltage and very little on the Plate voltage" (RCA Manual RC14)

Beyond the above basic design criteria, little discussion is offered in manufacturers' tube handbooks regarding the effects of
Screen-Grid voltage on Plate current.

The approach generally taken is to promote the application and use of vacuum tubes by publishing "typical" operating conditions
for vacuum tubes, including recommended Grid #2 operating voltages.

In the case of audio tubes there are copious examples provided that cover likely popular uses - often taken by designers 
verbatim or as recommended by tube manufacturers, without researching alternatives - resulting in copycat, "more-of-the-same"
designs. After all, why go the the trouble and expense of researching something that someone else had already pre-determined
and/or recommended - particularly if that "someone" has the expertise of a tube manufacturer?

Curves are nearly always provided for Control Grid (Grid #1) modulation characteristics, but no so for Grid #2.

Hence little published data is available to demonstrate the effects upon Plate Current from varying Grid #2 voltage.

A complicating, and perhaps confusing, factor is that tube manufacturers often recommend for "typical" applications, the same
Screen Grid voltage for a very wide range of Plate Voltages - particularly evident in high-voltage transmitting tubes.

There has also been no explanation as to why - except in the case of a small group of audio tubes - tube manufacturers
typically recommend Screen-Grid operating voltages that are mostly only around only half their maximum Screen-Grid rated
voltage - irrespective of applied Plate Voltage.

However one remarkable graph was published way back in 1957 that provides us with a deep insight into Screen Grid
behaviour, and is reproduced here for your information.

Fig: 3 - Philips Miniwatt 6CM5/EL36 Power Pentode for Audio and Video Applications

This remarkable rare original manufacturer's graph, shows very clearly the influence that Grid #2 voltage (Vg2) has over Plate
Current (Ia).
In this case, Grid #1 voltage (-1 VDC) has been selected to ensure it has negligible control over Plate Current, thus making Grid
#2 the controlling electrode. (In this tube Grid #1 voltage would normally be set at up to -29V to control Plate Current.)

This graph clearly shows that Plate Current is a direct function of Grid #2 Voltage. It is typical for all audio tetrodes,
pentodes and beam power tubes.

This graph also clearly shows that Plate Current is not a function of Plate Voltage in the useable range of Plate Currents - ie
Plate Current is linear and very dependent upon Grid #2 voltage in a reasonably linear relationship within the boundaries of
operation determined by the maximum plate dissipation rating.

It follows that the absolute limiting parameter of plate dissipation, although the product of Plate Voltage and Plate Current, is
directly determined by Grid #2 voltage - in other words, overheating or self-destruction of the tube may easily be achieved by
excessive Grid #2 voltage.

This graph shows very clearly why we should be concerned with the Screen Grid operating voltage and to take extra care that it
will be set at a value that will not only provide optimum performance but also extend tube life by ensuring tube dissipation is
within the prescribed limits.

Armed with the design knowledge provided by this graph, together with further analysis discussed below, we can make some
determining assumptions regarding the design of appropriate operating conditions for Screen-Grids.

4.    PRIMARY FUNCTIONS OF THE SCREEN-GRID

Thus, the primary functions of the Screen Grid in an Electron Tube are to:

a)    create an electrostatic shield between the Control Grid and the Plate
b)    minimise capacitance between the Control Grid (Grid 1) and the Plate
c)    control the electron flow in such a way as to make Plate current practically independent of Plate voltage over a certain
       range of circuit parameters
d)    control plate dissipation
e)    increase tube amplification and, in the case of a power tube, increase power output
f)     prevent feedback between the Control Grid and the Plate
g)    prevent unwanted oscillations of one type or another
h)    focus and accelerate the electron flow from cathode or filament to plate
i)     control electron flow to an extent more than that available from a single grid tube

In RF applications the screen grid may also be used to modulate the tube.

A further consideration is the mutual characteristic of a tube.

This term describes the inter-relationship between Control Grid voltage and Screen-Grid voltage and their combined mutually
interactive effect upon Plate Current.

To maintain Plate Current at a constant value, it is necessary to increase (ie make more negative) Grid #1 voltage to offset an
increase (make more positive) in Grid #2 voltage. The converse effect applies - ie decreasing Control Grid voltage (more
positive) requires a reduction (more negative) change to Screen Grid voltage.

The mutual characteristic is important to optimising operating conditions within the tube's maximum Plate and Screen Grid
dissipation ratings.

The following rare graph, which defines mutual characteristic for the ITT-Standard Type 4X150A Beam Power Tetrode,
courtesy of ITT-Standard publication MSE/123 published 1963,  illustrates this phenomenon.

This graph clearly illustrates the inter-relationship between Control Grid, Screen-Grid and Plate with the variables being Grid #1
and Grid #2 voltages.

It also shows that to limit Plate Current to safe or permissible values, it is essential to reduce Control-Grid voltage as the
Screen-Grid voltage moves closer towards the Plate voltage.
 
 
 ITT-Standard Type 4X150A Beam Power Tetrode

THEORETICAL APPROACH

For those who are theoretically minded, there is an excellent article on Screen Grid behaviour at http://www.burle.com/cgi-
bin/byteserver.pl/pdf/tp122.pdf
 

5.    SCREEN GRID (GRID # 2) - DC SUPPLY

The Screen Grid therefore, when connected for tetrode, pentode or beam power tube operation, should always be supplied by a
suitable low voltage direct current supply, having a low-impedance path to ground - ie effectively AC earthed.

The screen-grid supply should be regulated - or have good regulation properties - and be independent to the plate B+ supply.

The screen-grid supply should be capable of supplying transient peak current sufficient to supply the screen-grids with adequate
power to support transient signals - without incurring voltage drop at the screen-grids.

Voltage drop translates into significantly reduced gain in the tube, which translates into reduced transient peak power. Power
decreases at the rate of the square of the voltage reduction divided by the load.
A full-wave silicon diode bridge rectifier circuit with choke input to filter and as much capacitance as is practicable - ie at least
1,000 uF but preferably 5,000 to 10,000 uF - is desirable to ensure the screen grid voltage remains practically constant -
regardless of AC signal level and consequent DC Screen current.

Better still, a double section filter, comprising choke input to filter, followed by a second choke and capacitor, will ensure a high
quality DC supply.

Of course, a fully engineered regulated power supply is best to accommodate wide fluctuations in Screen Grid current..
 

DC Supply - Essential Requirements:

RCA Transmitting Tube Handbook TT-4 states:

At p7: "it is usually necessary, therefore, to remove ripple and other fluctuations from the screen-grid supply to prevent
undesired modulation of the tube output".
At p17: "Hum caused by the presence of ripple in dc plate, screen-grid (Grid No.2), or bias (Grid No. 1) supply voltages,
or by the use of ac filament or heater voltages, is also cancelled or substantially reduced in a push-pull stage."
At p30: "Plate and screen-grid supply circuits for single-ended Class A power amplifiers must be well filtered to minimise
hum and undesired coupling with other stages in the equipment."
At p30: "serious distortion and inadequate power output may result on large input signals unless the plate and screen-
grid supply voltages are well regulated and the bias is extremely stable."
At p31: "For optimum performance, plate supply regulation for Class B and AB2 amplifiers should be within 5%, and
screen-grid supply and grid-bias supply regulation should be within 3%."
At p73: :A bleeder resistor drawing about 10% of the total output current should be permanently connected across the
output of the supply. Although this resistor reduces the maximum useful output current slightly, it prevents the output
voltage from rising excessively when the external load is reduced, and thus improves regulation and provides a
substantial measure of protection for the filter capacitors. It also discharges the filter capacitors when the equipment is
switched off, and thus minimises shock hazards. Good regulation is desirable, even when substantially constant output
voltage under varying load conditions is not a primary requirement."
At p75: "In power supplies for cw transmitters, a ripple of not more than 5% is usually satisfactory. Power supplies for
variable-frequency oscillators and phone transmitters generally should have ripple of 0.25% or less. Power supply ripple in
high-gain speech amplifiers and receivers should not exceed 0.1% to prevent hum modulation of output signals. The most
economical method obtaining ripple voltages below 1% is by the use of double-section filters."

Note that if the Screen Grid supply is obtained from the Plate supply, both Plate and Screen Grid voltages will drop
simultaneously with high input signals, resulting in reduced power output, increased distortion and non-linearity - ie
reduced transient response in reproduction.
 

Effect of Plate and Screen Supply Regulation

When all element voltages change at the same time due to poor power supply regulation the change in performance will be very
audible.

The Radiotron Designers Handbook 3rd Edition (1940) at page 295 says:

"With a triode valve, the rise in average Plate current at full output (due to rectification) causes a decrease in the effective Plate
voltage, due to the resistance of the B supply. The result is a comparatively slight reduction in power output, since the drop in
Plate voltage opposes the rise in current.

With a Pentode or Beam Power Tetrode valve, however, the effect is much more pronounced. If the Plate and Screen operate at
the same voltage from a common supply, the drop in Plate voltage due to the resistance of the B supply also causes a similar
drop in the Screen voltage. This drop in Screen voltage results in a complete change in valve characteristics, the zero bias then
being lower than with full voltage. The cut-off grid voltage is then lower, and a lower grid bias is required for optimum operation,
possibly also accompanied by an increase in the optimum load resistance. The combined result is therefore to reduce the
maximum power output and to reduce the grid input voltage required for full output.

It is obvious that a Class A amplifier is less affected by poor regulation in the B supply than is a Class AB1 or other amplifier
drawing considerably more current at full output than at no output." (end quote)

Unfortunately a change in tube characteristics means a change in sound quality so the amplifier will not have constant tonal
characteristics throughout its dynamic power range - a very important attribute.

The amplifier will be non-linear when processing normal audio signals of say 20 db dynamic range.
Importantly, this non-linear quality will apply to all power stages relying upon cathode bias, which is one reason why guitar
amplifiers, which rely heavily upon accurate dynamic signal performance for their "sound" - ie transient response - prefer fixed
bias.
 

STC BRIMAR, in their Valve and Teletube Manual #8 (1959) state:

"The source resistance of the Screen voltage supply should be kept as low as practicable, and for most applications a potential
divider network, or other voltage source having good regulation, is preferred to a series resistor.

This is particularly applicable to pentodes having aligned Grids, and to unaligned Tetrodes, where the Screen current is subject
to relatively wide variation with operating conditions and between individual valves. In the case of Pentodes with unaligned
Grids, the variation is smaller and series resistors may be used.

Where variable Grid bias is applied to control gain, the use of a high-impedance supply to the Screen will result in a lengthening
of the Grid base.

At low anode voltages the Screen current tends to increase greatly, and care is required to avoid exceeding the Screen
dissipation. The Anode voltage should not be removed while the Screen is energised." (End quote)
 

EIMAC, in their Care and Feeding of Power Tubes website, present a different and comprehensive view of Screen-Grid current
flow and express concerns regarding secondary emission in tetrodes and pentodes.

Of particular importance is the concept of reverse current flow in the Screen-Grid circuit caused by secondary
emission - requiring not only a low impedance power supply for AC signal circuits but also a low resistance power
supply for the reverse current flow in the DC B+ circuit.

EIMAC recommend a current-bleeding resistor in the Screen-Grid supply circuit.

However these conditions are less severe with pentodes, due to the control over secondary emission by their Suppressor-Grid
(Grid 3)

Unfortunately Eimac's copyright restrictions prevent me from reproducing it here for your convenience - you will have to look it
up yourself at the above referenced link.

6.    SCREEN GRID (GRID # 2) - OPERATING VOLTAGE

RCA Receiving Tube Handbook RC-19 explains at Page 7:

"The Screen Grid is operated at a positive voltage and, therefore, attracts electrons from the cathode. However, because of the
comparatively large space between wires of the Screen Grid, most of the electrons drawn to the Screen Grid pass through
it. Hence the Screen Grid supplies an electrostatic force pulling electrons from the Cathode to the Plate. At the same time, the
Screen Grid shields the electrons between Cathode and Screen Grid from the Plate so that the Plate exerts very little
electrostatic force on electrons near the Cathode.

So long as the Plate voltage is higher than the Screen Grid voltage, Plate current in a Screen Grid tube depends to a
great degree on the Screen Grid voltage and very little on the Plate voltage" (end quote)
 

Important Notice: STC BRIMAR, in their Valve and Teletube Manual #8 (1959) state:

"At low anode voltages the screen current tends to increase greatly, and care is required to avoid exceeding the screen
dissipation." (end quote)
 

RCA Transmitting Tube Handbook TT-4 at page 8 further explains:

"If the negative excursion of the output signal swings the Plate to a voltage less positive than that of the Screen-Grid, electrons
moving from the Screen-Grid to the Plate tend to reverse their direction and return to the Screen-Grid.

The resulting decrease in Plate current causes a corresponding rise in Plate voltage, which terminates the negative swing of the
output signal before it completes its full excursion. This effect, which tends to reduce the power output of a tetrode below that
obtainable from a triode having equivalent plate-input rating, is emphasised considerably when there is secondary emission from
the Plate.

The loss of a portion of the output energy which occurs in a tetrode under these conditions reduces the power-handling
capabilities of the tube, and causes serious distortion of the signal waveform.

The output of the tube, therefore, contains harmonics of the signal frequency and other spurious frequencies which may cause
considerable interference to communications service. Such distortion may also be highly objectionable to the ear or to the eye
when a tetrode is used as an audio or video amplifier.

Although this effect may be minimised by reducing the amplitude of the plate-voltage swing so that the plate voltage never
swings negative with respect to the Screen Grid voltage, this expedient imposes further limitations on the tube output."

"Dynatron Action":

"The abrupt rise in the plate-voltage of a tetrode caused by the reversal of electron flow tends to draw both primary and
secondary electrons back to the Plate. Collection of these electrons then makes the Plate less positive than the Screen Grid so
that the tube current tends to reverse again.

This interchange of electrons between Plate and Screen Grid, called Dynatron Action, may continue for several cycles, and is
equivalent to an oscillatory current. Although dynatron action forms the basis of certain tetrode oscillator circuits, it is highly
objectionable when a tube is used solely as an amplifier." (end quote)

RCA Transmitting Tube Handbook TT-4 at page 8 further explains that the dynatron action problem is intended to be
overcome by the addition of a Suppressor Grid (Grid #3) in Pentodes which, when connected to the Cathode, establishes a
negative electrostatic field between the Screen Grid and Plate, to effectively prevent both primary and secondary electrons from
flowing backwards to the Screen Grid.

A different and comprehensive view of secondary emission in tetrodes and pentodes is provided by Eimac at their website -
Care and Feeding of Power Tubes. Unfortunately Eimac's copyright restrictions prevent me from reproducing it here for your
convenience - you will have to look it up yourself.

Essentially, there is a condition whereby the electron flow between the Screen-Grid and Plate cannot be controlled by Grid 1 -
ie once electrons have passed through the Screen-Grid they are more or less free to do whatever they want. Some go on to the
Plate but others return to the Screen-Grid.

Under certain conditions a situation of "thermal runaway" may develop, resulting in excessive Screen-Grid dissipation and
potential fusing of the Screen-Grid wire caused by excessive current flow back through the Screen-Grid to AC ground.
 

PENTODES AND BEAM POWER TUBES:

However, notwithstanding the above propositions, RCA Receiving Tube Manual RC-19 also states at Page 8:

"In power output pentodes, the Suppressor Grid (Grid #3) makes possible higher power output with lower grid-driving voltage; in
radio-frequency amplifier pentodes the Suppressor Grid makes possible high voltage amplification at moderate values of plate
voltage. These desirable features result from the fact that the plate voltage swing can be made very large. In fact, the Plate
voltage may be as low, or lower than, the Screen Grid voltage without serious loss in signal gain capability."

In the case of Beam Power Tubes, RCA Receiving Tube Manual RC-19 further states at Page 8:

"When a Beam Power Tube (ie a tetrode) is designed without an actual Suppressor Grid (Grid #3), the electrodes are so spaced
that secondary emission from the Plate is suppressed by space-charge effects between Screen Grid and Plate. The space-
charge is produced by the slowing up of electrons travelling from a high-potential screen Grid to a lower potential Plate. In this
low-velocity region, the space-charge produced is sufficient to repel secondary electrons emitted from the Plate and to cause
them to return to the Plate.

A feature of the Beam Power Tube is its low Screen-Grid current. The Screen Grid and Control Grid wires are wound so that
each turn of the Screen Grid is shaded from the cathode by a Control Grid turn. This alignment of the screen Grid and Control
Grid causes the electrons to travel in sheets between the turns of the screen Grid so that very few of them strike the Screen
Grid. Because of the effective suppressor action provided by the space-charge and because of the low current drawn by the
Screen Grid, the Beam Power Tube has the advantages of high power output, high power sensitivity, and high efficiency.
 Fig. 4 - Beam Power Tube Construction and Operation (Courtesy RCA)

Fig. 4 shows the structure of a Beam Power Tube employing space-charge suppression and illustrates how the electrons are
confined to beams. The beam condition illustrated is that for a Plate potential less than the Screen Grid potential." (end quote)

The design shown is typical of the 807 tube. Interestingly, the 807 (together with the 814) has the lowest Screen-Grid current of
any of the popular output tubes - ie substantially more of the total electron flow reaches the Plate, resulting in a more efficient
tube.

However, as will be seen below, increased Screen-Grid Voltage rating can only be achieved by increasing the physical
separation distance between Grid #1 and Grid #2 in the tube - ie shifting the Screen-Grid closer to the Plate, and/or reducing
the number of turns in the Screen-Grid wire to inhibit electron attraction - in such a way as to ensure compliance with the tube's
published specifications - resulting in reduced control over electron flow and a change in the "sound" of the tube.

Thus a 6L6GC with its 500 VDC Screen Grid rating, will have different dynamic characteristics (linearity) and will sound different
to a 6L6G with its 270 VDC Screen Grid rating - because its construction is different.

Theoretically, the earlier 6L6G tube having lower ratings should perform better than the later 6L6GC tube with higher ratings,
because in the latter case, the Screen-Grid has less electronic control over electron flow in the tube.

The 6L6GC thus would reasonably be expected to demonstrate higher total distortion than the 6L6G when operated within the
limits of the design ratings for the 6L6G, even though power output from both types should be the same when operated under
these conditions.

For a very detailed explanation of Beam Power Tube design and construction, refer to the engineering paper BEAM POWER
TUBES by RCA tube guru Otto Schade. Read with care, this paper provides the reader with an excellent insight into the design
rationale and theory of Beam Power Tube design. This paper is part of a set published in RCA Electron Tubes Volume 1 (1935-
1941) and Volume 2 (1942-1948).

Note: The original McIntosh amplifier applied 420 VDC to both Plates and Screen Grids, the latter being well above the rated
270 VDC design-centre value. It so happens that the much upgraded 6L6GC - and the better 7581 - are capable of handling the
420 VDC on the Screens with much less distress, so the 6L6GC and 7581 are therefore recommended as superior replacement
tubes for the McIntosh. In this application, the hi-fi version of the 6L6 - the 7027A - is also suitable, however the pin
connections are different and some rewiring of the socket connections may be necessary.

Further commentary on the screen-grid operating conditions of this amplifier was presented by Hugh Lockhart in 1956 - see 
http://www.tubebooks.org/Books/lockhart.pdf

Bruce DePalma, one of the few true Gurus of modern hi-fi amplifier design, presents an interesting and vital commentary on
Screen-Grids and other related issues in his Design Paper - "Analog Audio Power Amplifier Design"

Bruce developed designs that enable both Ultra-linear and low Screen-Grid voltage technologies to be successfully integrated -
eg Acrosound 6146 100 W RMS Hi-fi Amplifier.

7.    SCREEN GRID (GRID #2) - OPTIMUM DC OPERATING

VOLTAGE
To extend tube life and minimise distortion, it is recommended that the Screen Grid Voltage be as low as practicable - refer to
manufacturer's tube data sheets for recommended screen grid voltages.

The following JETEC USA design specifications explicitely limit Screen-grid voltage to pre-determined criteria.
 
 JETEC Specifications for Screen-Grid Operating Voltages
Courtesy of SYLVANIA 1959 Receiving Tubes Handbook

Note:    This JETEC design specification for operating conditions is also provided by RCA.
 

Notwithstanding the above JETEC design specifications - determined from extensive practical and theoretical research, design
type tested performance criteria and endorsed by leading manufacturers'- numerous examples of  commercial Guitar amplifiers
and Public Address (PA) amplifiers demonstrate typical design with a common Plate and Screen supply (as a cost saving
measure) having B+ supply voltages well above the above specified maxima.

However this operating configuration does not promote either long tube life or high-fi standard performance - in fact some tube
guitar amp designers deliberately configure the output stage to ensure desired distortion characteristics under sustained overload
conditions. But it can also be a recipe for overheating, unreliability, short tube life, instability, parasitic oscillations and/or
dynatron action in the output stage because the output tubes are running with the Plate Voltage less than the Screen Grid
Voltage (because of DC voltage drop in the primary of the output transformer).

This is particularly true of low-cost output transformers having high DC resistance windings - not to mention low primary
inductance and high leakage inductance which also facilitate parasitics.

Inaudible HF oscillations at full power output can easily damage loudpeakers - particularly tweeters having a "system power" or
"music power" rating. RC filters across the primary windings are typically used by commercial designers to roll-off HF response
in the output stage.

Remember, the purpose of the Screen-Grid is to accelerate and focus electrons towards the Plate. Excessive Screen-Grid
voltage attracts excessive electrons, increasing Screen-Grid temperature, current draw, and temperature rise - yes it does
matter!!

An important clue to Screen-Grid behaviour is found in the Tube Data Sheets for RF Transmitting power tubes. Here it will be
seen it is common - for a particular tube type - that the Screen-Grid voltage is expressed at a constant value , irrespective of
Plate voltage. Screen Grid voltage is always specified at a level substantially less than the Plate voltage.

What this practice suggests is that for a particular tube type, there will be an optimum value of Screen-Grid voltage
that will be sufficiently high to attract and accelerate electrons towards the Plate - irrespective of Plate voltage - beyond
which no significant advantage is gained.

The following examples illustrate this principle. Consider this sample of well known beam power tubes suitable for both RF and
AF applications:

Type 4-65A.
 The recommended Screen Grid operating voltage is only 250 VDC for all Plate voltages between 600 and 2500 VDC.
 
Type 4-250A.

The recommended Screen Grid operating voltage is only 300 VDC for all Plate voltages between 1500 and 3000 VDC.
 
Type 4X500A.

The recommended Screen Grid operating voltage is only 500 VDC for all Plate voltages between 2500 and 5000 VDC.
 
Type 807.

The recommended Screen Grid operating voltage is only 300 VDC for all Plate voltages between 400 and 750 VDC.

Furthermore, this engineering practice suggests that because the physical distance from the Cathode (or Filament) to the
Screen Grid is fixed during manufacture, once adequate Screen Grid voltage is applied to do the job then no further significant
benefit will be gained by increasing the Screen Grid voltage.

This statement needs to be considered alongside the reality that if Screen Grid voltage is increased then Plate current will
increase disproportionately, requiring a corresponding increase in Control Grid (Grid #1) voltage (ie more negative) to
compensate and keep plate dissipation within acceptable limits - thus reducing gain and operating capability to fully drive the
tube to maximum prospective power output for the available DC supply voltage.

Excessive Screen Grid voltage reduces its capacity to control electron flow in the tube and therefore affects gain, power output
linearity between zero and maximum signal, and increases distortion.

Another way of expressing this is to say that as far as the Cathode is concerned, the Screen Grid is the Anode. The rate of
electron flow will therefore be controlled by the Anode (Screen-Grid) voltage. What happens to the electrons after they pass
through the Screen-Grid and continue their journey to the Plate is of no concern to the Cathode.

It follows that the critical design element for a Tetrode, Pentode or Beam Power Tube is the Screen-Grid voltage,
because this is the effective Anode voltage.

As a rule of thumb, the screen grid supply voltage should NEVER be more than the manufacturer's rating. Higher applied
Screen-Grid voltage is likely to cause self-oscillation, parasitic oscillation, dynatron action or thermal runaway - any of which can
easily destroy a tube and associated components. MINIMAL Screen-Grid voltage will provide better performance including
cleaner, crisper sound with less distortion.

Tube Data handbooks typically recommend Screen Grid operating voltages at only half, or even less than half, the rated
maximum for a given tube type, warning us of the great control the Screen Grid has in determining tube performance.

In the case of pentodes having a separate Suppressor Grid, it is also relevant that the Suppressor Grid is usually either
connected directly to the Cathode inside the tube itself, or externally wired to the Cathode. Because the Suppressor Grid is
thereby at Cathode potential, it follows that excessive Screen-Grid voltage is likely to cause difficulties through interaction with
the Suppressor Grid.

It is also of importance to recall that the Screen-Grids of miniature amplifying tetrodes and pentodes as used in RF stages of a
receiver, or pre-amplifier stages of an audio amplifier, generally draw just a few milliamperes. Consequently, the actual Screen
current compared with the diameter of the Screen-Grid wire (hence its design-centre current rating) provides an inhernet safety
margin of headroom in terms of Screen-Grid current rating of the wire.

This design attribute enables the Screen-Grid to be bypassed directly to ground (Cathode), effectively creating an AC short-
circuit across the tube, with no apparent detrimental effect upon the Screen-Grid wire. One reason for this is the usually very
high value of Screen-Grid supply resistor - often 0.5 MegOhm or more - which limits Screen-Current to safe values. However
the same cannot be said for power tubes and further comments are made below on this subject.
 

Plate and Screen Dissipation of Tetrodes and Pentodes

The following rare graph, courtesy of ITT-Standard publication MSE/123 published in 1963,  illustrates how Screen-Grid current
changes with applied DC voltage to either Screen-Grid or Control Grid.

It clearly shows at 1 kV typical Plate Voltage and 0 VDC Control-Grid Voltage, that when we increase Screen-Grid Voltage
(above a critical value), the Screen-Grid current, and therefore Screen Dissipation, may increase dramatically.
Regrettably, corresponding data is not available for more negative Control-Grid voltages typical to audio applications

ITT-Standard Type 4X150A Beam Power Tetrode

Radiotronics Magazine No. 80 of October 1937 says:

"The power dissipated in the Screen circuit is added to the power in the Plate to obtain the total B supply input power. With full
signal input, the power delivered to the Plate circuit is the product of the full signal Plate supply voltage and the full-signal DC
Plate current. The power dissipated by the Plate in heat is the difference between the power supplied to the Plate circuit and
the power supplied to the load.

Screen dissipation increases with load resistance. In order to visualise this relation, assume that the sum of the Screen and
Plate current is independent of Plate voltage for zero Control Grid bias, or for a negative value of it. A decrease in Plate voltage
causes a certain decrease in Plate current; it is assumed that the Screen Current rises by an equal amount. Hence, when the
Screen Grid valve operates with a load which intersects the zero-bias characteristics below the knee, the Screen current rises to
high values during low-Plate voltage excursions of the output voltage. This action produces a rise in the DC value of Screen
current with signal. Therefore, the Screen dissipation with full signal input may be several times the zero-signal value. To reduce
Screen dissipation, the load should always be chosen so that it passes through the knee of the zero-bias characteristic.

Increasing the applied signal voltage to a value higher than that for which the load is designed also increases Screen
dissipation. For this reason, it may be advisable to use a value of load which is slightly less than the optimum value. This
precaution has another advantage, which is especially important at high audio frequencies. The impedance of a loudspeaker
increases with frequency. When the load is adjusted for the proper value at 400 Hz, the load is usually too high at 2000 Hz;
thus a Screen dissipation limit may be exceeded at 2000 Hz even though operation is normal at 400 Hz. The use of a load
which passes through the zero bias characteristic somewhat above the knee is desirable for these reasons." (end quote)

Note: The conditions described above are very likely in lead guitar amplifiers where the signal is of a single frequency nature.
 

METHOD 1:    AN EMPIRICAL OPTIMISING APPROACH:

The physical spacing between the cathode and anode in a vacuum tube is the gap across which the electrons must travel, and
is the gap across which applied voltage is measured and present (Plate Voltage).

Hence it can be stated with certainty that the DC voltage gradient across the cathode to anode gap is essentially linear.

Note: For those technically competent, early texts (Spangenberg, Beck, Argimbeau, Chaffee, Reicht, etc.) clearly show this
voltage tensor as having an exponential-shape, albeit not strongly, which starts at zero, then goes negative, then goes positive
to cross through zero at the "virtual cathode" point, and then climbs (always lagging the linear DC voltage gradient) toward the
maximum applied DC voltage. (Thanks to Earles L. Mc Caul for this contribution)

A simple example of this is seen in a vacuum tube rectifier, which comprises only a Cathode and an Anode - with a vacuum
gap between them.

It is relevant to note that a triode tube is just a rectifier with a Control Grid inserted between the Cathode and Anode to regulate
the electron flow through the tube - and hence through the circuit.

Examination of the physical construction of a vacuum tube, demonstrates that the control grids (Grid #1, Grid #2 and Grid #3
etc) are fixed in precise physical relationship to each other, to the anode, and to the cathode.

Further examination reveals that the relationship between manufacturers' Rated Plate Voltage and Rated Screen Grid
Voltage is directly proportional to the physical distance between each of them and to their common Cathode.

Given that the Rated Screen Grid Voltage is a maximum value and directly physically correlates with Rated Plate Voltage, which
is also a maximum value, it follows that when the actual applied Plate Voltage is less than the Rated Maximum - to maintain
linearity, or equal distribution of the applied DC voltage gradient across the tube, the applied Screen Grid Voltage MUST be
directly proportional to the linear relationship between Cathode to Screen Grid, and Screen Grid to Anode, within the tube.

If the Screen Grid Voltage exceeds the value indicated from the above method - as is common design practice - it can be
predicted with certainty that the velocity of electrons between Cathode and Screen Grid will increase, resulting in increased
Screen Grid Current, more secondary electrons produced from the Plate, increased distortion and greater propensity for the tube
to oscillate.

More importantly, there will be a mismatch between the "natural" Screen Grid Voltage - derived from the voltage gradient
created by its physical relationship in the electron stream gap - and the applied Screen Grid Voltage.

Thus this approach is suggested to determine the preferred Screen Grid voltage.

It assumes a linear relationship between Plate and Screen Grid voltages, by the formula:

optimum screen grid voltage =          actual plate voltage           X     maximum rated screen grid voltage
                                                        maximum rated plate voltage

Of course, plate and screen voltages are measured to the cathode or filament, as applicable.

This design approach ensures the Screen Grid voltage is optimised and will avoid unwanted secondary emissions and over-
excitation of the tube.

Note: Where the applied Plate Voltage exceeds the Rated Plate Voltage - such as in guitar amplifiers - it would seem prudent to
also proportionately increase the applied Screen Grid Voltage to maintain voltage gradient equilibrium. Note however there are
risks with this form of tube abuse and premature failure is a likely outcome.
 

METHOD 2:    A LOGICAL APPROACH:

RCA Receiving Tube Handbook RC-19 states at Page 8:

"In the case of Screen-Grid tubes, the proximity of the positive Screen-Grid to the Plate offers a strong attraction to secondary
electrons, and particularly so if the Plate voltage swings lower than the Screen-Grid voltage. This effect lowers the Plate current
and limits the useful Plate voltage swing for Tetrodes." (end quote)

Thus, another method is to adopt a policy that to optimise performance whilst maximising tube life, the Screen-Grid voltage
must never exceed the Plate voltage at full negative swing signal.

This is to ensure that the Plate will never swing negative in relation to the Screen-Grid thus causing the Screen to
replace the Plate as the PRIMARY ANODE during that portion of the signal cycle where the Plate is more negative than
the Screen Grid.

In other words, it is essential that the primary electron stream continue on past the Screen-grid to be collected at the
Plate - otherwise the Screen-grid will conduct too much current and melt.

Note also that the Screen-Grids are normally connected to AC earth via the screen bypass capacitor. Thus if the Screen-Grids
become the primary anode the signal will be short-circuited to ground, with disastrous consequences for distortion, linearity and
tube life (very short).

Therefore, returning to the primary proposition of this article - ie that Screen-Grid DC voltage must always be less than its Plate
voltage, an approximate value for optimising the DC Screen-Grid voltage may be determined by calculating the maximum Plate
to Plate AC signal voltage across the full output transformer primary winding.

This approximation, ignoring the effects of power factor in the AC circuit, may be determined by calculating the square root of
the value resultant from multiplying the output power in watts RMS by the primary load impedance.  (rms watts = output voltage
squared divided by load resistance)

eg Power output is 100 W RMS from a primary load impedance of 5,000 ohms.

Step 1:     100 x 5000 = 500,000.

Step 2:    Determine the square root of 500,000 = 707.
               Thus AC signal voltage is 707 V RMS plate to plate.
Step 3:    Half of that is 354 Volts. (half swings positive, half swings negative)
Step 4:    Hence to determine maximum permissible safe Screen Grid DC voltage subtract 354 from the actual Plate to
Cathode/Filament voltage.

For example, if the Plate voltage is 600 VDC subtract 354 V AC = 246 VDC absolute maximum applied to the Screen-Grids.

Obviously a lower voltage is desirable to ensure the electron flow continues past the Screen-Grid and on to the Plate, which is
their intended destination.

Remember too, that this calculation is based upon the tube manufacturer's rated output power - ie not actual, which may be
more if:

a)    the grid bias (Grid #1) is not set accurately, or

b)    the tubes have higher conductance than specified, or
c)    the amplifier is driven into overload, or
d)    a very high amplitude transient signal is amplified, or
e)    a different load impedance is used than that recommended by the manufacturer, or
f)    the reflected load impedance is different to the theoretical due to a variance between the stated and the actual loudspeaker
impedance, or
g)    the signal frequency coincides with the loudspeaker resonance frequency (primary load may increase up to six times the
nominal value). This situation is very likely with single note instruments such as an electronic organ or bass guitar, where no
signal averaging occurs.

To be sure, a margin of say 10% might reasonably be applied, so the calculated DC Screen-Grid voltage should be reduced by
at least a further 10% - more to accommodate transients.

It is interesting to note also that although RCA state in Transmitting Tube Manual TT-4 at page 9: " Beam Power Tubes may
also employ Suppressor Grids rather than space-charge effects to prevent the reversal of electron flow when the Plate swings
negative with respect to the Screen Grid." - a study of tube specifications reveals that RF Beam Power Tubes always have a
rated Screen Grid voltage substantially lower than the rated Plate voltage, thereby rendering the foregoing statement by RCA as
somewhat theoretical for both Pentodes and Beam Power Tubes.

It will be seen that when the above suggested formula is used to determine the negative AC signal voltage swing the resultant
calculated Screen Grid DC voltage will usually be above the manufacturer's Screen Grid DC Voltage Rating.

Using this method it will be observed that when the tube manufacturer's maximum rated Screen Grid DC voltage is used, the
AC signal voltage during its negative swing will still always be above the Screen Grid DC voltage, thus preventing adverse
effects.

eg compare these tubes of similar 125W rating:

Type                    Construction                     Max Screen DC Volts        Max. Plate DC Volts

4-125A                Beam Power Tetrode          400                                      3,000

4E27A/5-125B    Beam Power Pentode          750                                      4,000
803                      Pentode                               600                                      2,000
813                      Beam Power Pentode         1,100                                    2,250
 
 

IMPORTANT NOTE:

All the above discussion assumes the signal voltage at the Screen Grid to be simple sine wave waveform - of course in practice
it is not. This gives us yet another reason to further reduce Grid #2 voltage to ensure it always remains negative to the Plate.
 

8.    RESISTANCE IN THE GRID #2 SUPPLY - GRID STOPPER

       RESISTORS AND SCREEN GRID SUPPLY DROPPING
RESISTOR
 

RCA Transmitting Tube Handbook RC-19 states at page 60:

"The positive voltage for the Screen Grid (Grid #2) of Screen-Grid tubes may be obtained from a tap on a voltage divider, from a
potentiometer, or from a series resistor connected to a high-voltage source, depending on the particular type of tube and its
application. The Screen-Grid voltage for Tetrodes should be obtained from a voltage divider or a potentiometer, rather than
through a series resistor from a high-voltage source because of the characteristic Screen-Grid current variations of Tetrodes.

When Pentodes or Beam Power Tubes are operated under conditions where a large shift of Plate and Screen-Grid currents
does not take place with the application of the signal, the Screen-Grid voltage may be obtained through a series resistor from a
high-voltage source. This method of supply is possible because of the high uniformity of the Screen-Grid current characteristics
in Pentodes and Beam Power Tubes. Because the Screen-Grid voltage rises with increase in bias and resulting decrease in
Screen-Grid current, the cut-off characteristic of a Pentode is extended by this method of supply.

This method is sometimes used to increase the range of signals which can be handled by a Pentode. When used in resistance-
coupled amplifier circuits employing Pentodes in combination with the cathode-biasing method, it minimises the need for circuit
adjustments.

When power Pentodes and Beam Power Tubes are operated under conditions such that there is a large change in Plate and
Screen-Grid currents with the application of signal, the series resistor method of obtaining Screen-Grid voltage should not be
used. A change in Screen-Grid current appears as a change in the voltage drop across the series resistor in the Screen-Grid
circuit; the result is a change in the power output and an increase in distortion. The Screen-Grid voltage should be obtained
from a point in the Plate voltage supply filter system having the correct voltage, or from a separate source.

It is important to note that the Plate voltage of Tetrodes, Pentodes and Beam Power Tubes should be applied before or
simultaneously with the Screen-Grid voltage. Otherwise, with voltage on the Screen-Grid only, the Screen-Grid current may rise
high enough to cause excessive Screen-Grid dissipation." (end quote)
 

RCA Transmitting Tube Handbook TT-4 also states at p62:

"The danger of excessive screen-grid voltages is present principally when screen-grid voltage is obtained from the plate supply
through a series dropping resistor. In this type of supply circuit, sufficient resistance is connected between the screen-grid and
the plate supply to assure that the screen-grid voltage and dissipation at the values of screen-grid current, bias and driving
voltage required for full output are within the maximum ratings for the tube. Any condition which reduces the current through the
screen-grid dropping resistor to a very low value, therefore, may cause the screen-grid voltage to rise to an excessive value."
(end quote)

These sentiments are also expressed by Philips and STC Brimar.

It is therefore preferable that the dropping resistor should be part of a voltage divider network to further stabilise the supply and
to provide a direct current circuit to ground. All resistors in the voltage divider must be suitably rated.

When a dropping resistor is used from the B+ supply, a suitably large (ie large enough to offer a low-impedance path for the
frequency range being amplified) bypass electrolytic capacitor is essential to provide a return circuit to bypass AC signal voltage
to ground.  Note also that the power losses in such a resistor can be high, so a suitably rated wire-wound resistor is essential to
cope with the heat losses - eg typically 10 to 20W continuous power dissipation rating. Note also that this resistor may become
very hot after a while, so it must be located away from heat sensitive components such as electrolytic capacitors. The higher the
resistor's power dissipation rating the lower will be its temperature rise (Noting that present-day IEC standards permit a
substantially higher temperature rise than in days of old). A useful approach is to halve the resistor manufacturer's rated
dissipation.

In all cases, non-inductive grid stopper resistors (eg 500 to 1,000 ohms) must be fitted as close as is practicable to the socket
pin (read VERY close) to provide stable operation, minimise RF signal pickup, minimise inductance in the wiring, and prevent
parasitic-oscillation in the tube. Note that carbon film resistors may self-ignite if the insulating coating is not of fire retardant
material - be warned!! Composition carbon resistors may be a better practicable choice.

Philips Miniwatt put it this way in their "Miniwatt Electronics Handbook" (Australia 1960):

"The maximum value of peak Grid #2 dissipation is given to avoid the risk of impairing valve life by overheating the Grid #2
during long periods of excitation, which sometimes occurs with music or speech. In most cases, insertion of a non-decoupled
series resistor of 500 to 1000 ohms in the Grid #2 lead will reduce the actual value of peak dissipation to a large extent and not
seriously affect the output power.

During normal excitation with music or speech there will in general be no danger of exceeding the maximum value of Grid #2
dissipation when the valve is operated according to the published operating conditions.

In applications with a sustained sine wave input voltage" (bass guitar and electronic organ amplifier builders please note) "there
is a great risk of exceeding the maximum value of Grid #2 dissipation, so that in general full excitation is not allowed.

In order to prevent the maximum permissible Grid #2 dissipation from being exceeded it is necessary to ensure that the Plate is
always correctly loaded.

Hence the Plate lead must not be disconnected, nor must the loudspeaker be switched out, without replacing it by an
equivalent resistor". (end quote)

In the case of tetrode and pentode operation, do not use excessive values of unbypassed (non-decoupled) Screen series
resistance, because DC supply to the screen grid is likely to fluctuate substantially with screen current - thereby introducing non-
linearity, as well as separating the screen from direct connection to AC ground.

Voltage drop from DC Screen Current is a particular challenge with parallel-push-pull operation. Care is also needed with
conventional Class AB or Class B operation of single paired tubes.

Notwithstanding that though a word of caution:

Philips Miniwatt warn us in this way in their "Miniwatt Electronics Handbook" (1960):

"If the circuit is designed for operation of the valve below the knee of its plate current plate voltage characteristics, the Grid #2
series resistor must have a minimum value of * ohms in order to avoid the occurrence of Barkhausen oscillations." (end quote)
 

Some Food for Thought:

The usually recommended value of Grid Stopper Resistor is around 100 to 500 ohms (although to save on cost many
commercial amps successfully connect directly, with no Grid Stopper at all).

This has been standard practice for more than 60 years.

Interestingly, all the early tube literature shows the Screen Grid connected directly to the B battery, with no bypass capacitor.
This tells us that conceptually, early designers regarded the Screen Grid as being at a DC potential, with either no regard for the
AC signal component, or a reality that the battery provided the necessary AC bypass return circuit path to earth.

However an empirical approach derived from the manufacturers' data described on my ultra-linear operation page suggests a
value of around at least one half the Plate to Plate load impedance presented by the output transformer.

The basis for this proposition is that in normal push-pull tetrode or pentode operation there is little or no resistance
between Grid #2 and the B+ supply, so therefore there will be no Screen to Screen load equating to the Plate to Plate
load.

This is because the centre-tap of the output transformer primary is connected to AC earth (ground) via the bypass/filter cap at
that point.

In other words, in the case of normal push-pull tetrode or pentode operation there will be an AC SHORT-CIRCUIT
between the Screen Grids - and between the Screen-grids and their respective Cathodes.

Although this "short-circuit" - ie no load operation - will obviously increase Screen current, the AC signal current in the Screen
Grids will not appear in the output because it is diverted to earth through the bypass capacitor.

However no-load operation of the Screen Grids will increase the number of electrons collected by the Screens - a phenomena
we do not want - because we want the electrons to be only attracted to, but then continue on through and past the Screen
Grids on their way to their respective Plates.
Thus electrons collected by the Screen Grids not only increase Screen Current, but also divert electrons from the Plates and
therefore reduce power output.

It is this effect that results in the standard rating of "Screen Dissipation" - expressed in Watts. "Screen Dissipation" is the result
of DC Screen Input Watts minus AC Screen Grid Signal Output Watts. Thus if AC Signal Output Watts is zero or close to zero,
because the AC output between push-pull Screen Grids is more or less short-circuited, then the DC Input Watts will be
maximised under all signal conditions.

What we want is for the Screen Grids to be at a DC potential sufficiently high enough to attract and accelerate electrons towards
the Plates but, to maximise power output, not to collect and divert them to earth through the B+ supply.

Clearly there will be a particular value of Screen Grid Stopper Resistor that will provide optimum balance between the
conventional "short-circuited" Screen Grid configuration and an arrangement whereby the Screen Grids are suitably loaded.

The optimum value will clearly be variable depending upon the particular circuit configuration and operating voltages.

However, as a rule of thumb, and noting the advice of Philips Miniwatt to instal a value of Grid #2 resistor of between 500 to
1000 ohms in each Grid #2 supply lead, we can assume that a value of 50% of the Plate to Plate primary load impedance is an
approximate ideal for the Screen to Screen loading.

This will result in a grid stopper resistor value of:

500 ohms per Screen Grid when the transformer primary load impedance is 2,000 ohms Plate to Plate
1000 ohms per Screen Grid when the primary transformer primary load impedanceis 4,000 ohms Plate to Plate
2000 ohms per Screen Grid when the primary transformer primary load impedanceis 8,000 ohms Plate to Plate

For other values of Plate to Plate load, calculate on the basis that each Screen Grid resistor should be 25% of the transformer
Plate to Plate primary load impedance

In all cases, pursuant to Philips Miniwatt advice, the Screen Grid resistor is "non-decoupled" - ie is unbypassed.

This resistor must be installed directly to the Grid #2 pin of the tube socket and be preferably non-inductive.

The Screen resistors must have sufficient heat rating to operate safely and reliably without distress.

When multiple pairs of output tubes are used in parallel push-pull configurations, the Screen currents can attain reasonably high
values - eg 4 x 6CA7/EL34 = 100 mA. Ensure the Screen Grid resistors can handle this current without excessive heating,
noting the resistors will conduct heat from the tube pin/socket in addition to internal heat losses and temperature rise.

In multiple tube operation, to accommodate variations between individual tubes and to minimise the risk of self-oscillation, each
Screen-Grid must be supplied from its own individual grid-stopper resistor. This method also enables each grid-stopper resistor
to be mounted directly to each individual tube socket.
 

9.    "ULTRA-LINEAR" OPERATION

An alternative to normal tetrode, pentode or beam power tube configurations is the ultra-linear circuit, that avoids the need for a
separate screen grid supply.

"ULTRA-LINEAR" is a term, when applied to audio amplifiers, that describes the output stage configuration whereby the screen
grids (Grid 2) of tetrodes or pentodes are fed from a tapping on the primary of the output transformer, instead of from a separate
DC supply.

Ultra-linear is also known as distributed load operation.

Taking note of the above information regarding Screen-Grids, full details are provided in my separate ULTRA-LINEAR page.

10.    GUITAR AMPLIFIERS

Using the above knowledge about the behaviour of Screen-Grids, the following design rules can be applied to guitar amplifiers:
a)    BRIGHT, CLEAN SOUND (Minimum Distortion) - eg Lead Guitar, Country, Steel
Screen-Grids to be supplied from a reasonably constant voltage power supply. In simplest form this can be just a large
filter cap - say 100 uF after a filter choke (not a dropping resistor) from the Plate supply

Silicon Rectifier - full-wave bridge or voltage doubler

Plates to be supplied from a transient-current capable power supply. Use high-quality industrial grade (high-ripple) large
filter caps to plate supply - at least 100 uF per each pair of output tubes (double that for bass guitar amps)

Power Transformer continuous current rating to be double the theoretical output stage Cathode current (to improve
regulation)

Screen-Grids preferably supplied from a separate low-voltage (ie 40% of Plate Voltage) supply. Where Plates and
Screen-Grids are supplied from a common pwer supply then the filter caps should be as large as will fit into the chassis -
preferably 200 uF minimum - the bigger the better!!

Tube heaters preferably supplied from a separate transformer (to prevent voltage drop during peak power output)

Beam power tubes. Tubes must have aligned Screen-Grids (eg 6V6GT, 5881 (USA), 6L6GC/7027, 6550, KT66, KT88)

Output Transformer must have low DC resistance (to prevent voltage drop on peak power signals)

Output Transformer for lead-guitar amps to have low inductance (to rolloff low frequencies and prevent overloading of the
loudspeaker - especially in an open-backed cabinet)

ALL interstage coupling capacitors selected to provide -3db rolloff at 50 Hz

No negative feedback loop from loudspeaker

Grid #1 (control Grid) supplied from a full-wave rectifier

Grid #1 stopper resistors to be as low a value as is practicable - eg 100 k Ohms max per tube

Class AB2 or Class B operation

b)    SMOOTH, NATURAL SOUND - eg Jazz, Rythm, Folk, Bass

All the above but larger interstage coupling capacitors

Ultra-linear output stage and parallel-push-pull tubes configuration is essential for bass guitar (to reduce output
impedance)

Highest practicable loudspeaker impedance - eg series connected multiple loudpeakers (to reduce Output Transformer
turns ratio)

Silicon Rectifier - full-wave, full-wave bridge or voltage doubler.

Indirectly heated tube rectifier - eg 5AR4, 5V4G, 5Z4 etc - can be used but good filtering and regulation is needed (Not
recommended for bass)

Class AB1 or AB2 operation

c)    DISTORTED SOUND - eg Grunge, Heavy Metal, Blues

Directly heated tube rectifier (to ensure high voltage drop on transient peaks) - eg 5AS4,  5U4G/GB, 5Y3GT, 5Z3 etc

Small filter caps - eg 8 or 16 uF (to provide poor regulation)

High DC resistance filter choke (to provide voltage drop on transient peaks)

Plates and Screen-Grids supplied from common power supply

Plates and Screen-Grids at the same DC voltage

 Output Transformer to have high DC resistance (to reduce Plate voltage on transient peaks)

Pentode output tubes - eg EL34, EL84

Cathode bias

Class A operation

d)    BASS GUITAR

 

Screen-Grids to be supplied from a reasonably constant voltage power supply. In simplest form this can be just a large
filter cap - say 100 uF after a filter choke (not a dropping resistor) from the Plate supply

Silicon Rectifier - full-wave bridge or voltage doubler

Plates to be supplied from a transient-current capable power supply. Use high-quality industrial grade (high-ripple) large
filter caps to plate supply - at least 100 uF per each pair of output tubes (double that for bass guitar amps)

Power Transformer continuous current rating to be double the theoretical output stage Cathode current (to improve
regulation)

Screen-Grids preferably supplied from a separate low-voltage (ie 40% of Plate Voltage) supply. Where Plates and
Screen-Grids are supplied from a common pwer supply then the filter caps should be as large as will fit into the chassis -
preferably 200 uF minimum per pair of tubes - the bigger the better!!

Tube heaters preferably supplied from a separate transformer (to prevent voltage drop during peak power output)

Beam power tubes. Tubes must have aligned Screen-Grids (eg 6V6GT, 5881 (USA), 6L6GC/7027, 6550, KT66, KT88)

Output Transformer must have low DC resistance (to prevent voltage drop on peak power signals)

Output Tranformer for bass guitar amps to have grain-oriented silicon steel laminations and high inductance (to ensure
high power at low frequencies). Preferably interleaved windings for good frequency response, stability and low leakage
capacitance.

ALL interstage coupling capacitors selected to provide -3db rolloff at 20 Hz

No negative feedback loop from loudspeaker

Grid #1 (Control Grid) supplied from a full-wave rectifier - preferably from a separate transformer

Grid #1 stopper resistors to be as low a value as is practicable - eg 100 k Ohms max per tube

Class AB2 or Class B operation

 
 
 

Please let me know if you can add to this body of new knowledge and I will add it to this commentary.

Of course the smart thing to do here is to use tubes that are already triodes - instead of messing about with compromises - but
then none of the triodes are beam tubes and thus do not offer the benefits of  beam tube technology.

Also most of us have a junkbox stock of perfectly good tetrodes, pentodes or beam power tubes just waiting to be used - so it is
a tough call.
 

electron's
 
 
 
 
 
 

ULTRA-LINEAR
DISTRIBUTED LOAD
AUDIO AMPLIFIERS
 
 
 
 
 

*                *                *

ULTRA-LINEAR OPERATION, also known as DISTRIBUTED LOAD OPERATION, is a term when applied to single-ended or push-
pull vacuum tube audio amplifiers, that describes the particular output stage configuration whereby the Screen Grids (Grid 2) of
Tetrodes, Pentodes or Beam Power Tubes are fed from a tapping in the output transformer (single-ended) or each half of the primary
of the output transformer (push-pull) - typically 43% turns or 18.5% impedance when measured from the centre-tap, instead of from a
DC supply either independent of, or common to, the anodes.

The sonic properties of the ultra-linear output stage configuration - albeit single-ended or push-pull design - are midway between
triode and tetrode/beam power tube "tone".

Some audio-engineers describe the sound of triodes as "smooth, sweet, mellow, natural" and tetrode/beam power tubes as "clean,
bright, sharp, punchy".

Technically speaking, the ultra-linear configuration delivers the same power output as for pentode operation of the same tube
under the same operating conditions and typically about twice the power output of triode operation of the same tube under the
same operating conditions with the same applied DC voltages - but with substantially less harmonic distortion or intermodulation
distortion (see comparative performance graph below).

The ultra-linear configuration also offers improved overload characteristics, resulting in more effective power output - ie what the
listener actually hears at full power levels.

Output impedance is similar to triodes, allowing minimal or zero negative loop feedback to be used.

The ultra-linear amplifier concept is beautifully described by David Hafler and Herbert Keroes in their 1952 US Patent Application
2710312.

The following extract is included for convenience:

 The following illustrations show the conventional Ultra-Linear configuration in three typical arrangements.
 

The first shows the Ultra-Linear output stage with Cathode Bias having a common bypassed Cathode Resistor.
Superior results will be achieved by the use of a separate bypassed Cathode Bias Resistor to each tube.

This next illustration shows Fixed Bias Ultra-Linear operation, with each power tube having an individually adjustable bias supply for
improved balance in the output stage - essential for large power tubes:
Because the Screen-Grids are supplied with DC from taps on the output transformer, the ultra-linear configuration therefore avoids the
need for a separate Screen-Grid supply.
 

The third illustration shows the Acrosound Ultra-Linear design, also by Hafler and Keroes, which features a separate and lesser
voltage supply to the Screen-Grids for improved performance and cooler operation.

The ultra-linear output configuration is also suitable to single-ended output stages, however only push-pull amplification is discussed in
this paper.
 

How It All Began

One authority, F Langford Smith of Radiotron Designers Handbook fame, attributes the original design concept to a pair of Australia
inventors, R. Lackey and R.R. Chilton of the Australian Radio College, however the documented evidence has been lost. (If any reader
has that information please email it to me)

Consequently, in the absence of proof of the Australian design, the ultra-linear design concept is attributed by documented evidence
to "a British subject" Alan.D.Blumlein in 1936 - see British Patent 496,883 June 5, 1937. (Application No. 15620/37). See also expired
US Patent 449,4776. Also US Patent 2218902 (22 October 1940).

Blumlien's Patent specified an optimum feedback tapping ratio being between 25 and 50% of plate output voltage to the Screen Grid,
however he did not use either of the terms "Ultra-Linear" or "Distributed Load" in his Patent.

The much later detailed work of D. Hafler and H. I. Keroes of Acro USA (1951) demonstrated the optimum feedback voltage ratio to
the Screen Grid for a range of defined tube types.

To our benefit, they published a paper on the topic in the November 1951 edition of "Audio Engineering" magazine.

Specifically, the 6L6, 807and KT66 families prefer a Screen-grid load of 18.5% Plate load impedance. (43% turns or 43% Plate
Voltage).

Interestingly, they also claim the 6V6 family prefers a Screen-grid load of 5% Plate load impedance. (22.5% turns or 22.5% Plate
Voltage).

This concept is fully described in US Patent 2710312, dated June 7, 1955. History shows Hafler and Keros successfully patented in
the USA a further development of the general concept that had already been patented by Blumlein in the UK, had run most of its full
term, then expired (maybe not quite, because the H and K application was lodged May 20, 1952). Their claims included a level of
impedance loading to the Screen Grids of between 5% and 26% Plate Impedance - a much lower range to that of Blumlien.

This ratio was later verified by other respected designers, including GEC/MOV, GE USA and RCA

See also the thoughts of Norman Crowhurst.

Of particular interest is the Australian work of F Langford-Smith (of Radiotron Designers Handbook fame) in 1955, which investigates
the Ultra-linear concept in detail for a range of tube types.

See AWV Radiotronics Magazine May and July 1955

It is worth noting that the original Blumlein Patent would have remained in force until 1953, so not much would have - or could have -
happened regarding practical implementation of this technology in commercial products by other than Blumlien until after that date.

However, long before Blumlein's patent expired, he was killed in the year of 1942 in an aircraft crash whilst testing a new British radar
design. Thus we can only speculate as to what might have been had he survived. It would be reasonably safe to assume that hi-fi
audio would have been allocated a very low priority during the years of WWII during which Britain was struggling to survive.

Further reading on the life and times of this remarkable man is available at http://en.wikipedia.org/wiki/Alan_Blumlein

Since the famous "Williamson" amplifier design appeared some years after that date, it is reasonable to assume the "ultra-linear"
concept was either safely locked away, or rejected by the gurus of the time in favour of triode mode. In fact, Williamson not only
rejected the term but openly condemned the concept altogether - see Williamson and Walker - Wireless World 1952.

Thus the term "Ultra -Linear" may be rightfully attributed to Hafler and Keroes - an opinion supported by the "Radiotron Designers
Handbook 4th Edition. This is the term used in their 1955 US Patent 2710312 (Application date 20 May, 1952) and is supported in
that Patent by technical justification for use of the term.

In contrast to the above documented evidence, some researchers claim the Ultra-Linear concept was previously used in Australia as
far back as 1933.

The question is - "How many ways is it possible to configure a 4 electrode device in an electronic circuit?"

Of course the answer is that someone, somewhere, at sometime will figure it out.

The famous Williamson triode-connected beam power tube amplifier was later modified by Hafler and Keroes to incorporate the ultra-
linear design concept. Details are provided in their paper "Ultra-linear operation of the Williamson Amplifier" by Hafler and Keroes,
published in the June 1952 edition of "Audio Engineering" magazine. It seems then that they had the last word, by improving
Williamson's design despite his dismissal of the concept.. Numerous examples of this particular design are now available on the
internet.
 

The Ultra-linear configuration in audio amplifiers is also known as "DISTRIBUTED LOAD" operation.

It appears this term was intoduced by Willamson and Walker of the UK, who objected to the term "ultra-linear".

Of course this term is also technically correct, because the taps on the output transformer primary do in fact present a push-pull
reactive load to the Screen Grids, enabling them to contribute to useable power output. This is explained in detail in the Hafler and
Keroes Patent.

However importantly, Mullard describe the "distributed load" configuration as a system of negative feedback.

The term "distributed load" was adopted by Mullard as corporate policy. Their use of this term was published in the May and June
1955 editions of "Wireless World" in their article "Design for a 20 Watt High Quality Audio Amplifier".

It is interesting to note that this design appeared from Mullard, notwithstanding Williamson's previous condemnation of the concept.

I am unable to verify the specific designer of the 5/20 amplifier but later credits go to Mullard Applications Research Laboratories
Engineers Mssrs R. S. Babbs, D. H. W. Busby, P. F. Dalloso, C. Hardcastle, J. C. Latham and W. A. Ferguson. (No mention of
Williamson and Walker). It is assumed the commercial pragmatists won the argument.

Mullard continued to use this term through until at least the 1960's when their compendium of audio amplifier and pre-amplifier circuits
"Mullard Circuits for Audio Amplifiers" was published as a second edition reprint in 1962.

The following extract from that publication provides an overview of Mullard's design concept.
For an excellent technical introduction to the ultra-linear design concept see http://www.aikenamps.com/UL.pdf

More technical detail, single-ended ultra-linear circuits and non-power tube ultra-linear applications is also available from Glass Audio
- "Tube CAD Journal"  Vol. 2 No. 1 - January 2000

For further views and technical details see also http://www.vintageradio.me.uk/amplifier/10watt.htm

Important Note: Performance described as "ultra-linear" is available only under very specific operating conditions - and in
the case of popular commonly used audio tubes, usually where the Screen load is 18.5% impedance or 43% turns of the
Plate load impedance - measured from the power supply source (B+) terminal.

More detailed information is provided below.

In other cases, the term "distributed load operation" may be more appropriate.
 

Tube Operating Conditions:

The attached graphs, courtesy of GE USA, show the marked difference in operating conditions for pentode, triode and ultra-linear
operation of the 6550 Beam Power Tube.

6550 Beam Power Tube Operating Conditions - Pentode Connection

  
 

6550 Beam Power Tube Operating Conditions - Triode Connection

 
 

6550 Beam Power Tube Operating Conditions - Ultra-Linear/Distributed Load Connection

 
 

Benefits:

The ultra-linear configuration ensures constant stage specific voltage feedback between Plate and Screen grid in the output stage,
thus reducing output impedance and distortion, whilst improving linearity and frequency response under variably reactive loudspeaker
load conditions.

An important advantage is that being single stage-specific, ultra-linear operation avoids the problems of time-delay and phase-shift
commonly associated with cascaded stage amplifiers using negative feedback from the loudspeaker for the purposes of reducing
distortion and increasing loudspeaker damping.

Ultra-linear output stages provide automatic constant ratio stage-circuit specific feedback, free from the adverse effects of conventional
multi-stage feedback systems.

Ultra-linear output stage power output is dependent upon the proportion of Plate load that is applied to the Screen-Grids - typically in
the range 75% to 100% that of tetrode/pentode connection at the same plate voltage - but is still twice triode connection for the same
tube type under the same operating conditions.

Ultra-linear tone or "sound" approximates that of triodes.

Ultra-linear operation is very forgiving of circuit design and generally provides an acceptable quality sound from average quality output
transformers. Note however that the esteemed Norman Crowhurst, in a November 1959 "Audio" Magazine article entitled "Puzzled
About Amplifiers", indicates that a high-quality output transformer is essential for ultra-linear operation, to ensure high-fidelity
performance over the entire audio frequency range.

To this end, Herbert Keroes of Acrosound developed a special transformer for ultra-linear configuration output stages. The main
purpose of this transformer is to enable the use of transmitting tubes, where the Screen-grids must operate at a significantly reduced
DC voltage to that on the Plates. Full specifications are provided in US Patent 2791646 (May 7 1957).

A set of high-fidelity amplifier designs was published in the Acrosound Ultra-Linear Transformer Catalogue, published a little later.

The flagship of the range - the TO350 - offers 100W from a pair of 6146 tubes. This is an exceptional amplifier by any standard.
 

Design Concept:

A good example of a typical conventional ultra-linear circuit is shown in the GEC KT88 100W Amplifier.

A more highly developed design is shown in the GEC KT88 400W Amplifier.

The conventional ultra-linear configuration is arranged such that the screen grids have maximum DC voltage applied to them (in
practice slightly higher than their corresponding plate/anode voltage because of voltage drop in the primary winding) - thus maximising
power output and efficiency, whilst simultaneously receiving an AC negative feedback signal from the output transformer - thus
minimising output impedance and distortion.

The magnitude of the AC feedback signal is directly proportional to the percentage turns ration of screen tap to full winding.

The screen tap may be positioned anywhere from 0% (pentode connection) to 100% (triode connection), however performance and
tonal qualities change relative to the ratio of the screen tap - see GEC graph below.

In practice, research by the inventors of this method of amplifier configuration suggests that the ideal for most tube types is in the
range 40 to 50% turns of each half primary, measured from the primary centre tap.

Further research by Mullard UK for the EL34/6CA7 and EL84/6BQ5, GEC/MOV Valve Co. for the KT88, GE USA for the 6550, and
RCA for the 6973 (6CZ5 hi-fi), 7027A (6L6GC), 7591, 7868 (6L6 family), also recommends 43% turns (or 43% of plate signal voltage),
or 18.5% impedance.

An example of EL34/6CA7 operation with cathode bias is shown in the following graph - courtesy Amperex Electronic Corporation
 
 
 

Screen-Grid Operation:

One very important feature of the ultra-linear configuration - different to normal Tetrode/Pentode operation where both Screen-Grids
are at nominal AC earth potential - is that when one Screen-Grid in a push-pull pair is driven positive, the opposite Screen-Grid is
driven negative by the turns ratio of the output transformer acting about the centre-tap of the transformer, which is at nominal AC
earth. This is not a problem because the opposite Plate is also being driven negatively anyway by the action of the push-pull
driver/phase splitter.

The opposite applies when the alternating signal reverses polarity.

In Class B operation, the output transformer operates as an auto-transformer, so the opposite Plate and Screen-Grid are still driven to
opposite polarity (together in a constant ratio to each other), even though they are not conducting.
 

        *               *               *               *               *               *               *               *               *               *               *

Balanced Amplifier

It should be noted that in the case of a push-pull amplifier, the DC negative bias voltage (even if it is at 0 VDC) applied to the Control
Grid #1 is located at the centreline axis of the balanced input signal. This means the inputs are floating and the centre-axis is
earthed. That is to say, the input Grids are being driven in a push-pull manner about a common centre-point or axis - eg as seen with
a centre-tapped push-pull driver transformer.

Thus in a conventional push-pull Tetrode or Pentode amplifier, any alternations of the balanced input signal to the Control Grids will
proportionately increase or decrease current flow in both tubes of the push-pull pair in response to the input alternating waveform
shape.
Now in each tube of a push-pull pair, the negative terminal (Cathode) of each tube is AC earthed - even in Cathode bias.

The load on each tube is connected between the Plate and the transformer centre-tap, so the negative terminal of the load - which is
at the output transformer centre-tap - is also effectively AC earthed.

But most importantly to ultra-linear operation, in a push-pull amplifier output stage, Screen-Grid behaviour will be similar to that of
Grid #1 - ie the Screen-grids will behave as a balanced amplifier - balanced about the centre-tap of the output transformer which, as
previously noted, is effectively AC earthed - ie variations in Screen-Grid voltage will produce proportionate variations in electron flow
in both output tubes simultaneously.

In conventional Tetrode and Pentode operation, when the DC applied voltage to the Screen-grids is constant, the AC voltage
appearing at the balanced Screen-Grid terminals will the product of the electron flow within each tube and will always be a voltage
determined by the natural AC voltage gradient applied internally across each tube.

Also in a conventional Tetrode/Pentode output stage, the AC signal voltage appearing at the Screen-Grids is diverted to AC earth at
the Screen-Grid terminals via the filter capacitor and is lost as heat. Hence the output voltage appearing at the Screen-Grid terminals
is of no consequence to the sound produced by the amplifier - ie it is not reproduced in the output transformer or loudspeaker.

However, in the case of the Screen-Grids in ultra-linear push-pull amplifier configuration this is not so, because the Screen-Grids are
located at about 43% of the lineal distance between Cathode and Plate. Thus the AC signal voltage appearing at each Screen-Grid as
a result of linear voltage gradient between the Cathode and Plate within each tube will thereby be about 43% of the Cathode to Plate
signal voltage.

The Screen-Grids are connected to the load so they will contribute to the sound produced by the amplifier.
 

Class A and Class B Ultra-Linear Operation

In some transmitters, it has been the practice to drive or control the output power tube by means of the Screen-Grid, rather than the
Control Grid.

This method offers some benefits to RF situations but is relevant to the explanation of ultra-linear audio amplifiers.

Because the Screen-Grid is located much further into the physical Cathode to Plate distance - ie typically nearly centrally between
them - it follows that a substantially higher AC signal voltage must be applied to the Screen-Grid if that element is to control the
electron flow in the tube.

Operation will be the same as for Grid #1 but at a higher AC voltage.

It follows that if the DC Screen-Grid voltage controls electron flow within the tube and it is varied by means of a superimposed AC
voltage, then the Plate Current will vary in response - as for conventional Grid #1 operation.

But what if the Screen-Grid voltage is applied in 180 degrees directly opposite phase to the electron stream within the tube - as is the
case for ultra-linear operation?

Obviously the electron stream permitted by the AC signal as applied to Grid #1 is offset by the opposing signal voltage applied to Grid
#2 via the tapping on the output transformer - because they are both in the same circuit at the same time.

Consequently behaviour of the tube in response to the controlling signal voltages applied to BOTH Grid #1 and Grid #2
simultaneously, will be different to that of either a Triode or Tetrode.

This is what the above GE graphs demonstrate - ie a change in the operating characteristics of the basic tube, verifying the claim of
Hafler and Keroes that they had produced a "virtual" tube, intermediate between a triode and a tetrode.

The situation is however very different between Class A and Class B operation.

In the case of Class A ultra-linear operation, the Screen-Grid of one tube will be driven AC positive but the Screen-Grid of the
opposite tube will be driven to an equal, but opposite polarity, voltage by the output transformer balanced output about the AC earthed
centre-tap. This will have the effect on the second tube of reducing the effective Screen-Grid control voltage, thereby reducing voltage
gain and therefore power output - regardless of the shape of the input waveform to that tube.

But also note that Grid #1 of both tubes is also controlling current in them.

Hence in Class A ultra-linear operation, it is necessary to consider the effects upon Cathode current in BOTH output tubes by  BOTH
Control Grids and BOTH Screen-Grids.

However in the case of Class B ultra-linear operation, although the second tube will be similarly driven to reduce its gain, there is no
signal in it at that moment in time because its Control Grid #1 is at cutoff bias during the relevant half-cycle of signal input so zero or
near zero current is flowing in that tube.
Thus a Class A ultra-linear amplifier will have completely different behaviour to a Class B ultra-linear amplifier.
 

Distributed Load Operation

On the other hand, that portion of the output transformer primary winding between the Screen-Grid tapping and the centre-tap is not
subject to cancelling out by out of phase signal in the power circuit.

This portion of the winding - usually 18-19% Plate to Plate load impedance - therefore imposes a load directly onto the Screen-Grids.

Thus a portion of the power output is delivered through this Screen to Screen winding.

However, although the winding and its corresponding load impedance is presented to the Screen-Grids, they are incapable of
delivering much power because the Screen-Grids will be aligned in the electron stream such that the small diameter grid wires are not
directly in the electron stream - so because of space-charge effects cannot attract electrons.

Furthermore, the Screen-Grids are charged to opposite polarity to the electron stream during half of each signal alternation cycle,
which further detracts from their electron collecting capability.

For the record, in their US Patent 2,710,312 Hafler and Keroes state: "It should be noted that power is transferred to the Screen only
over part of the signal cycle, i.e. when the absolute value of Plate potential falls below the absolute value of Screen potential. This
transfer has the effect of linearizing the Plate characteristics."
 

Power Output

Ultra-linear operation typically (for the popular audio tube types) delivers 100% power output compared with the same tubes in tetrode
or pentode connection at the same plate voltage and bias system (GEC) - but sometimes less for other tube types.

For example, Mullard quote the power output for the  EL34 tube as being the same for pentode and ultra-linear connection with 20%
turns screen-taps, but for minimum distortion the screen-taps increase to 43% turns, which provides a power reduction of 15% (but
distortion is halved).

Note:    Worthy of note is the KT88, which GEC (MO Valve Co./Genalex) claim produces the same power in ultra-linear connection as
in Pentode connection at 43% turns. This may be due to the applied Screen-Grid voltage used in ultra-linear operation being twice the
recommended value as that for Tetrode operation. Importantly, the electrode structure of the original GEC-MOV KT88 is
internally physically identical to the GEC TT21 and TT22 RF transmitting beam power tubes, so the GEC-MOV made KT88 is
therefore capable of operating at its rated 600 VDC maximum Screen-Grid DC operating voltage continuously without
distress.

The TT21 or TT22, which have a top anode cap and intended for professional broadcast use, may be used in lieu of the KT88 as a
direct electrical substitute. Grid to plate capacitance is reduced, which should produce superior audio performance over the KT88.

One of the challenges to the home constructor is that manufacturers' tube manuals and data sheets often quote "typical operation" for
fixed bias pentode connection but cathode-bias for ultra-linear connection.
 

Ultra-linear Circuit Characteristics:

The following graph, courtesy of GEC and AWV Radiotronics Magazine (May 1959), shows the comparative characteristics of the
KT88 in triode, pentode and ultra-linear connections. These comparative relationships between the different connection configurations
should be typical for most (but not all) tetrodes, pentodes and beam power tubes.
Note:

the relationship between output impedance and ratio of screen tap turns.
the relationship between output power and ratio of screen taps.
the relationship between THD distortion and ratio of screen taps.
the effect of load impedance upon output power.
the power output for ultra-linear operation is greater than that for pentode or triode operation.

This graph shows that for ultra-linear operation the original GEC/MOV KT88 is unsurpassed!!
 

WARNING: 6550 V KT88

Although having identical general specifications the 6550 Beam Power Tube - its USA cousin, has substantially lower
electrical ratings for pentode, triode and ultra-linear connections - so be warned!!
Although the 6550 and KT88 are often specified in tube handbooks and amplifier schematics as equivalent substitutes for each
other it may be wise to check the ratings first before making a change from KT88 to 6550.
Physically, some brands of 6550 tubes have a parallel glass envelope smaller in diameter than the KT88 so there may be
inadequate space to instal the KT88 or to provide adequate natural air cooling without major modification. The reverse change
is electrically not a problem.
Although the KT88 is derived from the TT21 and TT22 transmitting tube stable, the origin of the 6550 is not clear.
Some users have reported detailed examination of certain Russian made KT88 and 6550 tubes suggests they are physically
identical but branded for use as either KT88 or 6550 thus it is wise to check the manufacturer's specifications before installation
in any high voltage circuit.

WARNING:

The GEC/MOV KT88 octal socket centre base spiggot is made from a brittle plastic moulding material and tends to break off if
the tube is not inserted and removed very carefully from its socket. This tube is big and heavy in relation to its octal base so is
unsuited to horizontal or inverted mounting unless retained by a suitable device.
The GEC/MOV KT88 was typically made with one or two base pins omitted (presumably to save on cost) that has the effect of
 increasing stress on the remaining pins whilst providing a less secure installation.
Do not use the KT88 in inverted mounting without a retaining device.

EL34/6CA7 Ultra-linear Circuit Characteristics:

The following graph by Mullard UK shows performance of the EL34/6CA7 valve.

Note the comparative "linear" performance for power out v distortion over a wide range of loads and similar operating conditions.
 

Further information is provided by the following table ex Mullard UK, that shows comparative performance between the Mullard EL34
and EL84 valves in various output stage configurations.

Recent Research

I am indebted to Rudolf  Moers, a distinguished Electrical and Electronics Engineer located in the Netherlands, who has made
available for us his wonderful recent scientific investigation into the design theory and practice of Ultra-linear audio amplification.

These papers are posted with permission from Linear Audio www.linearaudio.net and their author Rudolf  Moers.

Part 1 - Power Point presentation

Part 2 - Power Point presentation

Parts 1 and 2 of above Power Point presentation - combined in pdf format

Paper - The Ultra-Linear Power Amplifier: An adventure between triode and pentode - pdf format
 

The engineering design methods developed by Mr Moers may be used to determine theoretical plate/screen load ratios for ultra-linear
operation of power tubes.
 

Application:

In practice, for tube types other than KT88, the real loss of useable output power from the ultra-linear connection is actually
significantly less than any power differential measured with resistive loads might suggest (ie a 1.5 db reduction in loudness produced
by the loudspeaker), because the ultra-linear connection produces a higher coupling efficiency between the amplifier and loudspeaker
than tetrodes or pentodes - ie is more triode like - thus approximating an equivalent "loudness" to pentode connection.

This phenomenom is particularly true of low frequency reproduction, suggesting that ultra-linear connection is superior for double bass
violin, bass guitar and general hi-fidelity reproduction down to about 40 Hz - which is the lowest musical frequency normally
reproduced in popular music.
The 6L6 family of tubes can be used for ultra-linear connection, but only safely and reliably at reduced voltages not exceeding rated
screen-grid voltage. More suitable 6L6 style types include 5881 and 7027A - see manufacturers' data sheets for typical circuit values.

Ultra-linear operation typically delivers about two to three times the power output compared with the same tubes in triode connection
at the same plate voltage.

Most importantly, inter-modulation distortion is substantially lower with ultra-linear connection compared with pentode or triode
connection for the same tubes.

Note: In the case of hi-fi systems, percussive instruments such as the bass drum, tympany, harpsicord, guitar and the piano present
typically short-duration/transient low fundamental frequency signals rich in harmonics. Thus they are less demanding to reproduce than
the electronic organ, pipe organ or bass guitar, which produce an essentially long-duration signal, approximate to a sine wave, in the
lower register.

Triode connection is still the preferred option for seriously loud organ music because of the lower frequencies to be reproduced, the
need for "boom" free bass around the loudspeaker resonant frequency and, most importantly, a need for consistent gain (flat
response) throughout the musical scale to ensure all notes are reproduced with equal loudness when they are recorded that way.
However triodes deliver substantially higher intermodulation distortion than ultra-linear operation so adequate power headroom is
essential to ensure clear mid to upper range reproduction simultaneously with sustained lower register signals.

A major advantage of ultra-linear connection over tetrodes and pentodes is the improvement in tone for recorded music and public
address reproduction. The tone is more triode like, being smooth and mellow (but "clearer" than triodes), compared with the harsh tone
of tetrodes and muddy tone (read "less definition") of pentodes.
 

Guitar Amplifiers:

Ultra-linear operation is not usually suitable for lead guitar amplifiers because it lacks the "crispness", "harshness" or "bite" in the
sound commonly provided by beam power tetrodes and pentodes, however some jazz and country music guitarists may find it
preferable where a smooth, natural mellow tone is sought.

However, ultra-linear operation is superior to tetrode/pentode connection for bass guitar applications because low frequency power
delivered to the loudspeaker is substantially greater - ie it is "louder" - due to lower output impedance and improved coupling to the
loudspeaker.

It also has a "deeper" tone, suggesting improved sub-harmonic performance.

 

Summary:

In summary, ultra-linear connection offers:

useable power output approaching that of tetrodes and pentodes but with substantially reduced harmonic and intermodulation
distortion and improved tonal characteristics, thus making it the preferred configuration for high-fidelity applications.
a "warmth" of tone, not evident with the "clinically clean" sound of unity coupling, where the load is distributed between plate
and cathode.
simplicity in circuit design and construction
elimination of need for a separate power supply to the screen-grids
employment of a simple output transformer
constant-voltage negative feedback
high stability operation over a range of reactive and capacitive loads
strong useable low-frequency power output - ie as heard through the loudspeaker
good dynamic range performance and transient response

Ultra-linear operation also enables lower quality loudspeakers to be used for satisfactory results.

Note that in high power applications - ie more than 100W RMS, transmitting TRIODE tubes such as 805, 809, 810, 811, 812, 833,
845, 8000, 8005, etc may be a more economical and practical solution than trying ultra-linear configuration with tetrodes/pentodes
because of simplified wiring, output transformer and power supply requirements.
 

TUBES FOR ULTRA-LINEAR OPERATION

The "Ultra-linear" configuration avoids the conventional conflict between plate and screen voltages by creating a voltage divider
network through the output transformer primary to AC earth (transformer centre-tap), ensuring the screen voltage tracks and thus
always remains, both below and proportional to the plate signal voltage. By this configuration, the screen is intended to be prevented
from exceeding its power dissipation rating.
This statement is subject to the Screen Grid always being operated within its Rated DC Screen Grid voltage and at a DC
voltage less than that of the Plate.

Applied plate (B+) voltage for ultra-linear connection should never exceed Grid 2 rated voltage, so standard ultra-linear configuration is
only suited to tubes designed for audio applications having a Grid 2 rating approximating either the plate voltage rating (or actual
applied plate voltage if less than rated maximum).

Important:    Reference to tube data sheets will show that few output tubes have ratings remotely matching this requirement.

Important:    The use of poor quality output transformers having a high DC resistance in the primary windings may establish a situation
whereby the actual DC Plate voltage drop across the transformer primary winding is high, causing the DC Screen-Grid voltage to be
higher than the Plate at high signal levels.

Unfortunately, only a small number of tube types are thereby suitable for ultra-linear operation, because in ultra-linear mode the
screen grid is operating at or above the plate voltage - a dangerous operating region for any tube.

Only a few tube types were recommended by their manufacturers as being suitable for ultra-linear connection, the most notable being
EL34/6CA7, EL84/6BQ5, KT88, 6550, 7027 and 8417 - see manufacturers' data sheets for typical circuit values.

In a typical output tube, the Screen Grid is the ANODE, or positive electrode. It is designed to accelerate electron flow from Cathode
to Plate, but is structured in such a way that most electrons pass through it and on to the Plate for collection.

Excessive screen grid voltage attracts excessive electrons to it, thereby resulting in excess grid current, excessive grid power
input/dissipation, overheating and melting. The fused screen grid wires may short-circuit B+ to earth, damaging the output transformer
and/or power supply components.

Unfortunately, most audio tetrodes, pentodes and beam power tubes are designed such that the Screen Grid may be operated only up
to a maximum DC voltage that is well below the Plate voltage - typically 150 to 300 volts.

Plate current is very much controlled by the Screen Grid, thus when the Screen Grid is made ineffective (ie over-active) by application
of excessive voltage, the Plate Current is likely to exceed the tube ratings and also melt the plate.

Another way of saying this is that if the Screen Grid voltage is excessive, the capability of Grid #1 (Control Grid) to control electron
flow in the tube is diminished - or lost altogether.

Operation of typical power tubes (having a screen grid voltage rating substantially lower than the rated plate voltage) in ultra-linear
connection is likely to result in loss of control over electron flow by the screen, resulting in thermal runaway or dynatron action -
resulting in self-destruction of the tube. Fire is a constant risk.

For example, tubes designed for RF power service typically have a plate voltage rating many times higher than their corresponding
screen grid voltage rating.  This class of tube (eg 807, 2E26, 6146, 4CX series) has the screen grid physically positioned close to the
control grid (Grid 1) and tend to self-oscillate or suffer thermal runaway when the screen grid voltage is higher than their rated screen-
grid voltage, which is always the case with conventional ultra-linear service, thus rendering them unsuitable for ultra-linear service.

WARNING: When using tubes fitted with a plate top cap (anode cap) in ultra-linear configuration, consider also the risk of self-
oscillation and/or parasitic oscillations from the combination of  long leads from the plate top caps and screen grids to the output
transformer - particulary significant when using multiple tubes to obtain higher power. Some of this lead length may be avoided by
chassis layout however in the case of top cap style tubes the screen connection is always under the chassis, thus ultimately requiring
connection in some way from top to bottom of the tube through the output transformer.
 

High Power Ouput:

Where more power is needed it is preferable to use a larger tube that is known to be more suited for ultra-linear circuitry, such as the
KT88, KT90 or 813.

Another useful option is to run multiple pairs of tubes in parallel push-pull, such as the arrangement used in the GEC 400W KT88
Amplifier. It goes without saying that normal precautions against instability and parasitic oscillations are essential in layout, lead dress,
use of grid stopper resistors wired directly to the pins, and keeping inputs away from outputs. The output transformer must be of high
quality with low leakage capacitance and  low leakage inductance between windings. Tubes should be mounted close together to
minimise inter-connecting lead length. Grid #1 circuit resistance must be held within the manufacturer's ratings. A low-impedance
driver, such as a cathode-follower or transformer is recommended.

For high power applications where high plate voltages are needed, success may be achieved by adding a separate winding for each
screen grid to the output transformer, to enable the screens to be AC coupled to the plates thus providing ultra-linear operation, but
separated from the DC plate supply, thus enabling the screens to be supplied within their rated voltage from an independent supply.

The ACROSOUND 100W Ultra-Linear Amplifier using 6146 tubes and the TO-350 Transformer is an example of this excellent and
innovative design configuration.
One of the benefits of this configuration is that the Screen-Grid supply can be independent to the Plate supply and therefore better
regulation can be incorporated into the Screen Grid circuit. Noting the Screen-Grid is the ANODE in a Tetrode or Pentode tube, a
regulated supply will deliver improved transient response and a "brilliance" to the reproduction not available with a common B+ power
supply as is the case in conventional Ultra-Linear operation.

Bruce DePalma, one of the few Gurus of modern hi-fi amplifier design, presents an interesting and comprehensive commentary on the
core design philosophies supporting this approach in his excellent Design Paper - "Analog Audio Power Amplifier Design"

Bruce has developed designs that enable both Ultra-Linear and low Screen-Grid voltage technologies to be successfully integrated in
such a way that extremely hi-fi performance results.
 

BIAS

Grid 1 bias for ultra-linear operation is normally higher than that for tetrode/pentode connection, so output stage sensitivity is reduced.
Higher output voltage from driver stages is therefore needed.

Grid-stopper resistors to Grid #1 and Grid #2 are still required for ultra-linear operation.

Important: The technique of using a silicon rectifier diode on each Screen Grid in series with the Grid-stopper resistor, as described in
my "OPTIMISED ULTRA-LINEAR ©" page, is very helpful in ultra-linear connection. AC signal output voltage from the Screen-Grids
is prevented from conducting through to the output transformer, which means that theultra-linearoutput stage operates as a negative
feedback system only - ie Plate voltage is fed back to the Screen Grids via the transformer taps but not the other way around. This
method ensures all the electron flow goes to the Plates, with all the advantages described previously.

Note:    This configuration has opposite polarity as when zener diodes are installed to reduce screen-grid voltage, so therefore the
benefits described are not relevant to the zener diode technique.
 

Important - When using a high B+ voltage:

To ensure Plate Dissipation remains within manufacturer's rating at both zero and maximum signal, it may be essential to use Class B
operation - thus introducing further complexity into the circuit design and perhaps offsetting much of the benefit offered from the ultra-
linear configuration. It may be more prudent to use a tube having a higher rating.
 

RECOMMENDED TUBES

The following tubes are known to be suitable for conventional ultra-linear operation having nominally equal Plate and Screen-grid DC
supply volts:

TUBE TYPE            MAX. ULTRA-LINEAR CONNECTION PLATE TO CATHODE VOLTS

6AQ5A                            285
6BM8                              300
6BQ5/EL84                      300
6BW6                              285
6CA7                               500
6CM5/PL36                      250
6CZ5                                285
6L6GB                             270
6L6GC                             400
6V6GT                             285
5881                                400
6146A, B, W etc               250    (Top cap style)
6550                                450    (Note: GE USA "design maximum" rating. Some brands may not tolerate this voltage)
6973                                300
7027A                              450
7189A                              400
7581                                450
7591                                400    (50% turns recommended by RCA)
7868                                400    (50% turns recommended by RCA)
807/5B-255M                    300   (Top cap style)
8417                                500
EL34                                425
EL84/6BQ5                       300
KT66                                400
8417                                500
KT88                                 600     (Note: GEC UK "design maximum" rating. Some brands may not tolerate this voltage)
KT90                                 600    ("Absolute Maximum" rating)
813                                  1100   (This tube is an excellent option for serious audiophiles but has a top cap and requires a centre-
tapped 10 V AC/DC filament supply - 5A per tube. External wiring must be screened to prevent RF induction and parasitic oscillations.
Adequate ventilation is essential)
 

In my humble opinion, the most suitable candidates for ultra-linear connection are:

6CM5/PL36    (250 VDC)

6AQ5/6V6GT (275 VDC)
6CZ5/6973      (285 VDC)
5881/6L6GC/7027A/7591/7868    (400 VDC)
KT88/KT90   (600 VDC) (may need to reduce to 450/500 VDC with non GEC manufacture)
813              (1100 VDC)

All are well proven fine quality beam power tubes - each famous in its own right.
 

Other options are the EL34 (425 VDC) or 6CA7 (500 VDC) or EL84/6BQ5 (300 VDC) pentodes, however in my opinion sound from
these tubes is not as clean as those above. All of this family of tubes "sounds" similar because of their generally identical electrode
construction.

If you have had successful experiences with other tube types that are useful for ultra-linear connection please email your comments.
 

For further information regarding Ultral-linear operation of vacuum tubes see my OPTIMISED ULTRA-LINEAR © OPERATION page
 
 
 

© NOTICE:
INTELLECTUAL PROPERTY COPYRIGHT © D.R.GRIMWOOD 2002 - ALL RIGHTS RESERVED.
 
 

BACK TO HOME PAGE

 

This page last modified 16 April 2012

This page is located at http://www.oestex.com/tubes/ul.htm

 electron's

OPTIMISED ELECTRON STREAM © TECHNOLOGY

in

AUDIO and RF MODULATOR

TETRODE, PENTODE AND BEAM POWER TUBES

Readers, please note this page is presented for your education, information and guidance only - it is not intended to be a
technical/scientific treatise. The concepts and ideas presented herein are just as much subjective as objective.

It attempts to assist those experienced audiophile home constructors who wish to explore further tube amplifier development options
and are prepared "to go beyond the square" - to challenge the paradigm of "the status-quo".

This paper refers only to the characteristics and performance of push-pull tube audio amplifiers without negative feedback.

For reasons detailed elsewhere in my website I have no interest whatsoever in single-ended amplifiers - however the concepts
described herein are as equally applicable to single-ended amplifiers as to push-pull configurations.

For reasons detailed elsewhere in my website I have no interest whatsoever in the use of trans-stage negative feedback thus the
concepts described herein are intended to optimise electron tube performance without the use of negative feedback.

For full ratings and applications of specific tube types in which you are interested please refer to the manufacturer's catalogue.

Copyright in all quoted works remains with their owner, author and publisher, as applicable.

Please note that no warranty is expressed or implied - see footnote notice.

Excepting where rights in Intellectual Property previously exist, all rights in the applied engineering concepts expressed in
this paper remains exclusively with the author.

The term OPTIMISED ELECTRON STREAM © TECHNOLOGY applies to all part or whole uses of the concept.

The whole or part thereof of this paper and/or the designs and design concepts expressed therein may be reproduced for
personal use only without limitation - but must not under any circumstances be applied or used for commercial gain or
reward without the express written permission of the author, being the intellectual copyright owner.

All rights reserved.

 

OPTIMISED ELECTRON STREAM © TECHNOLOGY

1.    INTRODUCTION

It is recommended that to best understand the concepts presented here, the reader first carefully study my explanatory papers.

To directly access please click on each of the links below:

SCREEN-GRIDS

ULTRA-LINEAR

OPTIMISED ULTRA-LINEAR © OPERATION

TRIODE OPERATION OF TETRODES AND PENTODES

POWER SUPPLIES

HOW TO DESIGN AND CONSTRUCT A HI-FI TUBE AMPLIFIER

 

It is intended that the set of OPTIMISED ELECTRON STREAM © TECHNOLOGY concepts presented below represent - together as a
set, an extension of design thinking to the above linked presentations elsewhere on my website, and a major breakthrough in electron
power tube technology available to the home audio constructor.

Historically we have been restricted by convention to a small set of engineering design principles that have imposed a barrier to further
development, reinforced by commercial manufacturers staying with the tried and true "safe" design configurations.

However the availability of low-cost high-performance transistor amplifiers and pre-amplifiers, together with advances in recording
formats - currently DVD standards - have provided stimulus to tube hi-fi enthusiasts to improve the performance of their existing
equipment.

Many of the old constraints are longer with us - hiss, noise, wow, flutter and rumble etc. do not present on CD or DVD - so enable us
to open the frequency range and dynamic range window a little more.

The OBJECT of OPTIMISED ELECTRON STREAM © TECHNOLOGY is to OPTIMISE performance in a high-fidelity audio
amplifier by simultaneously loading ALL the internal elements of a power electron tube with a load directly proportional to
their ACTUAL applied voltages - AC and DC - as distributed between the positive and negative terminals of the tube - ie
anode and cathode - AND in direct proportion to their physical electrode spacing (dielectric strength).

Hence OPTIMISED ELECTRON STREAM © TECHNOLOGY may also be described as OPTIMISED DISTRIBUTED LOAD ©
TECHNOLOGY

OES  © TECHNOLOGY offers superior performance because it is simple and because the tube is always in equilibrium.

The conceptual model is illustrated in the following circuit diagram for Beam Power Tubes, Tetrodes and Pentodes:
To prepare a foundation for the concepts presented on this page, please read my SCREEN-GRIDS page before reading this paper,
because the function and behaviour of the Screen-grid is fundamental to OPTIMISED ELECTRON STREAM © TECHNOLOGY.

OPTIMISED ELECTRON STREAM © TECHNOLOGY: SCOPE

This paper introduces eight new applied engineering OPTIMISED ELECTRON STREAM © TECHNOLOGY concepts for refining and
optimising the electron flow in a vacuum tube.

Optimised Screen-Grid Voltage and Load in Tetrodes, Pentodes and Beam Power Tubes

Optimised Suppressor Grid Voltage and Load in Pentodes

Silicon Diode Feed in Triodes, Tetrodes, Pentodes and Beam Power Tubes

AC Circuit Bypass Capacitors - Plate Circuit, Screen-grids, Suppressor-grids, Control-grids, Cathode Bypass

RATIO of Screen-grid and Suppressor Grid and Cathode Bypass to Plate Bypass Capacitors

Decoupling of the Power Amplifier from the Power Supply by means of a series connected silicon diode

Decoupling each amplifier stage from the next following by means of a series connected silicon diode

Symmetrical balanced AC signal drive system whereby the central axis of the AC signal OUTPUT from the Driver stage
circuit and the AC signal INPUT of the Driven stage circuit are at the same common reference voltage potential.

Subject to individual circuit design, all or any of the above OPTIMISED ELECTRON STREAM © TECHNOLOGY concepts may be
used individually, or in combination, or all together - each offering a specific benefit to enhancing electron flow within the tube or
circuit.
 

2.    OPTIMISED ELECTRON STREAM © TECHNOLOGY:

       OPTIMISED SCREEN-GRID VOLTAGE AND LOAD

My OPTIMISED ULTRA LINEAR ©  page describes a design method for optimising the DC supply voltage and load impedance to
the Screen Grids in Tetrodes, Pentodes and Beam Power Tubes for "Ultra-linear", or "Distributed Load" operation.

In order to determine what optimum transformer primary tap ratio to use it is essential to physically determine the actual Screen Grid to
Anode physical (linear distance) space ratio, because for general audio class electron tubes the ratio of screen grid to plate voltage
will be optimally equal to the physical distance ratio of cathode to screen-grid v cathode to plate.

The same principle for optimisation of tube performance may be applied to conventional operation of Tetrodes, Pentodes and Beam
Power Tubes in "Pentode' operation - ie, where the Screen-Grids are supplied from a separate DC source to the Plates.

To optimise the Screen-Grid voltage, first carefully and accurately measure the physical gap spacing between the Cathode, Screen-
Grid and Plate.

In most cases this can be easily done without having to destroy a valuable tube. Just measure the spacing between Cathode and
Anode and of the grid support pins at the top of the tube. In the case of beam power tubes this may be difficult because there is
usually a box-like assembly covering the Plate structure. But a bit of ingenuity should solve the problem

If it becomes necessary to destroy a tube to examine its internals safety precautions must be observed:

1.    Wrap the tube in a strong cloth

2.    Place the tube in a vyce and squeeze the glass bottle slowly until it implodes
or
3.    Gently hit the glass with a hammer at a point not directly over the electrode assembly
4.    Carefully remove the broken glass and dispose of safely
5.    Carefully cut away the electrodes until the elements can be measured

The "gap" spacing ratio can then be calculated for the particular tube type.

To determine the Optimised Screen-Grid voltage, multiply the Plate voltage by the gap ratio.

It is obvious the Screen-Grid voltage will be substantially less than the Plate voltage - typically in the range 30 to 50 % but usually
40% - so a separate stable supply is essential.

The designer has the option of using:

a voltage divider network from the Plate Supply (not recommended for hi-fi)

a stabilised Screen-Grid DC supply derived from the same AC source (power transformer) as the Plate supply (not
recommended for hi-fi)

a separate Screen-Grid DC supply derived from a separate power transformer and full-wave rectifier to the Plate supply (tube
rectifier not recommended for hi-fi)

a separate regulated Screen-Grid DC supply derived from a separate power transformer and full-wave rectifier to the Plate
supply (tube rectifier not recommended for hi-fi)

What is most important is for the Screen-Grid DC supply voltage to be reasonably constant between no-signal load and full-signal
load conditions (ie well regulated).

This can be effectively achieved by using a separate power transformer of generous rating - ie at least twice the maximum signal DC
Screen-Grid current, a full-wave silicon bridge rectifier and humungous filter capacitors in capacitor input to filter configuration -
preferably with a low DC resistance filter choke installed too.

For the more theoretical designer some thought could be given to the "Virtual Cathode" concept, which suggests that the portion of 
electron stream devoted to the negative bias applied to Grid #1 creates a more negative "Virtual Cathode" in the region of the physical
Grid element. This concept however is difficult to assess because so long as the tube is conducting some of the electron stream is
continuous between real Cathode and Plate. Maybe only the outer portions of the electron stream are affected by the negative bias
and not the entire thickness of stream.

The perfectionist - or those having the necessary equipment - looking for the ultimate optimisation could determine the optimum
operating voltage for the Screen-Grids by means of a Signal Generator, Distortion Meter and variable voltage power supply to the
Screen-Grids.

The optimised AC load impedance for the Screen-grid will be the ratio of DC Screen-grid volts to Plate volts x Plate load impedance.
If the Screen-grid load is provided by a fixed resistor the ratio for a Screen-grid - or set of Screen-grids in one side of a push-pull set,
will be the DC voltage ratio x one quarter Plate to Plate load impedance

Note however that if the load is provided by a transformer as in ultra-linear or equivalent mode, the Screen to Screen load impedance
will be directly proportional, but the turns ratio will be the square of the DC voltage ratio.

eg If the DC ratio is 40% then the turns ratio for the Screen-grid tap on the output transformer will be 0.4 x 0.4 = 0.16 or 16%

A technical paper describing benefits of reduced screen-grid voltage is presented at http://www.tubebooks.org/Books/lockhart.pdf

 

Recent Research

I am indebted to Rudolf  Moers, a distinguished Electrical and Electronics Engineer located in the Netherlands, who has made
available for us his wonderful recent scientific investigation into the design theory and practice of Ultra-linear audio amplification.

These papers are posted with permission from Linear Audio www.linearaudio.net and their author Rudolf  Moers.

Part 1 - Power Point presentation

Part 2 - Power Point presentation

Parts 1 and 2 combined presentation - pdf format

Paper - The Ultra-Linear Power Amplifier: An adventure between triode and pentode - pdf format

The engineering design methods developed by Mr Moers may be used to determine theoretical plate/screen load ratios for ultra-linear
operation of power tubes.
 

3.    OPTIMISED ELECTRON STREAM © TECHNOLOGY:

       OPTIMISED SUPPRESSOR-GRID VOLTAGE AND LOAD IN PENTODES
 

The PENTODE electron tube has an extra grid (Grid #3) installed to suppress the effects of secondary emission, as described in my
SCREEN-GRIDS page.

Although few true Pentodes are available for audio power amplifier output stage applications, this section relates to them because they
are suitable for use in OPTIMISED ELECTRON STREAM © TECHNOLOGY applications.

Manufacturers' tube data sheets generally describe Pentodes as having higher measured distortion than Beam Power Tubes so most
audio amplifier designs have focused on the latter types.

In the case of recycled or salvaged tubes from yesteryear, it is evident that by WWII, RF design engineers preferred Beam Power
Tubes because of their capability to operate at higher RF frequencies, hence the true Pentode enjoyed only a short claim to fame.

One exception to this is the mighty 803 Pentode, which found widespread use in submarine and destroyer RF applications because of
its ruggedness, stability, reliability and long-life. Of interest is that the 803 pentode has aligned grids, each having a ceramic coating to
limit electron attraction, thereby improving predictability and stability of operation, as well as efficiency.

The 837 Pentode tube is also renown as a very reliable oscillator tube for RF transmitters. The 837 Pentode has a Plate structure
similar to that of the famous TT21 and KT88 Beam Power Tubes.

However Pentodes do offer user benefits over Tetrodes and have a place in high-fidelity audio and RF amplifiers - particularly for
home constructors who might have a box of useful Pentodes just waiting to be used in a suitable design.
 

It is a well established tube engineering principle that the current in a tube can be regulated using ANY of the grid elements.

For example, in domestic radio receivers, the use of multi-grid tubes such as 6A7, 6A8 and 6L7, each having 5 grids, is a standard
application of radio engineering design.

However for audio engineering purposes, three grids appears to be the practical limit - beyond which no appreciable benefit is to be
realised. In fact. the beam Power Tube utilises only two grids to control current through the tube.

Using the same applied engineering technology as developed for OPTIMISED ULTRA LINEAR © operation, it is possible and
practicable in power tubes fitted with a SUPPRESSOR-GRID having an independent base pin connection, to apply a DC voltage and
AC load impedance to the SUPPRESSOR-GRID having the same ratio to Plate voltage and load respectively, as the internal electrode
physical gap ratio.
 

IMPORTANT NOTE 1: It is the case in many of the popular PENTODES that the SUPPRESSOR-GRID is internally connected during
manufacture to the CATHODE, so it is not possible to externally access the Suppresser Grid. In this class of tube the manufacturer
controls the behaviour of the Suppressor-grid and the user cannot do anything to change that.

IMPORTANT NOTE 2: Tube identification protocols describe and illustrate by base pinout diagrams Beam Power Tubes as
"Pentodes". These tubes are essentially Tetrodes that have a beam forming electrode used to both confine the electron stream to a
pre-determined width within the tube, and to focus the electron flow onto a particular area of the Plate. They usually have aligned
Grids to arrange the electron flow in multiple sheets.

Therefore it is essential to physically examine tubes of interest to determine their actual mechanical construction.

DO NOT RELY ON DESCRIPTIONS OR BASE DIAGRAMS PUBLISHED IN TUBE MANUFACTURERS' CATALOGUES, MANUALS
OR DATA SHEETS.
 

However those tubes having a separate base pin connection for the SUPPRESSOR-GRID offer the designer the option of applying a
suitable DC voltage consistent with the proportional voltage divider effect of the internal electrode gap.

The rules and requirements for an adequately regulated DC power source are the same as described above for the Screen-Grid.

For example, in the Type 803 - 125W dissipation power transmitting pentode tube - the Suppressor Grid is set at nominally 68% of the
Cathode to Anode gap. Therefore the OPTIMISED ELECTRON STREAM © value for the Suppressor Grid DC voltage will be the
applied Plate Voltage x 68% - but only up to a value 158% of that permitted by the manufacturer's rated Screen-Grid Voltage
(because in the case of Tube Type 803, the maximum permissible Screen-Grid voltage is nominally 43% of the applied Plate Voltage).

In other words, the maximum applied DC Screen-Grid Voltage and/or Suppressor Grid Voltage for OPTIMISED ELECTRON
STREAM © operating conditions will be determined by either the maximum rated Screen Grid or Suppressor Grid voltage -
whichever is the lesser.
 
 

IMPORTANT:

The SUPPRESSOR GRID (GRID #3) regulates the flow of electrons in the tube in the same way as is the case for Control-grid #1
and Screen-Grid #2.

In conventional "Pentode" configuration, the Suppressor-grid is directly connected to the Cathode either internally by the tube
manufacturer or by the user.

Having left the Plate as surplus, randomly travelling electrons, they find their way to the Suppressor-grid, thence diverted to the
Cathode to be absorbed back into the electron stream. This arrangement obviously creates a short-circuit in respect of those electrons
attracted to the Suppressor-grid.

Thus there is an effective internal or external circuit (as applicable) created between the Suppressor-grid and the Cathode, that diverts
some of the electrons back to the Cathode. This current is lost to the output power and therefore reduces efficiency in the output
power stage.
In the case of the relationship between the Screen-grid and Plate, most experts suggest that the Plate sees the Screen-grid as the
"Cathode", thus if this is so then the DC potential between Plate and Suppressor Grid will be again determined by the linear distance
between them - unless the Suppressor-grid is purposefully connected to the Cathode.

It follows that in an OPTIMISED ELECTRON STREAM © technology amplifier where the Screen-grid DC voltage is around 50% of the
Plate voltage, if the applied DC voltage is increased from 0 volts (Cathode potential) to a positive voltage greater than the Screen-grid
voltage, then the free electrons deflected from the Plate will still be attracted to the Suppressor-grid because it is still negative to the
Plate.

It also follows that because the Suppressor-grid is now positive to the Screen-grid, Plate current will increase - albeit slightly.

Since both Screen-grid and Suppressor grids DC voltages will be fixed, it becomes obvious that to control the electron flow within
permissible limits, the negative bias voltage applied to Grid #1 Control Grid will need to be made MORE NEGATIVE.

It is also obvious that to limit the DC current flow in Grid #3, and to prevent an AC signal short-circuit at the Suppressor-Grids,  it is
essential to load the Suppressor-grids by installing a Grid-stopper resistor of around 75% of equivalent Plate to Plate push-pull load
impedance.

eg for a Plate to Plate load of 8,000 Ohms, the transformer will present a load of 2,000 Ohms to each tube in the push-pull pair.
2,000 Ohms x 75% is 1500 Ohms. This is still a relatively small value so should not present significant voltage drop or regulation
issues.

This is an important difference between convention and the OPTIMISED ELECTRON STREAM © configuration.

In a conventional pentode circuit with grounded Cathodes, the Suppressor-grid is directly connected to the Cathode so the
Suppressor-grid is fixed at 0 VDC. However in the case of  OPTIMISED ELECTRON STREAM © configuration, the Suppressor-grid
will be fixed at a relatively high DC voltage set to enable the AC voltage to be aligned with the physical gap in the tube. The circuit
design then needs to accommodate this to protect the Suppressor-grid from self-destruction through over-current.

The optimised AC load impedance for the Suppressor-grid will be the ratio of DC Suppressor-grid volts to Plate volts x Plate load
impedance.

If the Suppressor-grid load is provided by a fixed resistor the ratio for a Suppressor-grid - or set of Suppressor-grids in one side of a
push-pull set, will be the DC voltage ratio x one quarter Plate to Plate load impedance
 

When used in an OPTIMISED ULTRA LINEAR © configuration as shown above, it will also will be loaded by the tapped primary
winding on the output transformer. The AC load impedance will be determined by the ratio of the physical gap between the Cathode
and Suppressor-grid compared to the Cathode to Plate, as applicable - nominally 75% of the Plate to Plate load.

The Suppressor-grid to Suppressor-grid load impedance will be directly proportional, but the turns ratio will be the square of the DC
voltage ratio.

eg If the DC ratio is 75% then the turns ratio for the Suppressor-grid tap on the output transformer will be 0.75 x 0.75 = 0.5625 or
56.25%

Consequently, proportionate audio POWER is drawn off from the Screen-grid and Suppressor-grid by the output transformer.
 
 

PENTODE tubes suitable for high-fidelity audio power output applications using OPTIMISED ELECTRON STREAM © technology
include:

802
803
804
837

I have not listed the EL34/6CA7, its smaller brother the EL84/6BQ5 and cousin 6M5,  because the construction and effect of the
Suppressor-grid in these tubes is "nominal" and not in the same league as those of the transmitting tubes listed above. Being
manufactured from very fine wire, the Suppressor Grid in these tubes is not be capable of handling significant current or power. For
high-powered audio amplifiers superior options are available as shown above.

However the EL34/6CA7,  EL84/6BQ5 and 6M5 are suitable for effective use in OPTIMISED ELECTRON STREAM © amplifiers where
the Screen-grid is set at 40% of Plate voltage and Suppressor-grid is connected directly to the Cathode in the usual way, and in
OPTIMISED ULTRA LINEAR © amplifier designs.
 

3.1    OPTIMISED ELECTRON STREAM © TECHNOLOGY - SILICON DIODE FEED TO GRID 3

Consider also OPTIMISED ELECTRON STREAM © TECHNOLOGY - SILICON DIODE FEED (as descibed in Section 4 below) to the
Suppressor Grid.

In this case, a conventional pentode configured amplifier - where the pentode tube has a separate externally connected independent
Grid 3 pin (typically connected to the cathgode or ground) - may be modified by installing a silicon diode between the Suppressor-grid
pin on the tube socket and Cathode or ground - as applicable. The arrow must point towards ground - ie the marked terminal on the
diode to the Suppressor-grid pin.

This configuration allows AC current to flow to ground but blocks DC current into the tube.

If any instability or non-linearity occurs, a 1,000 Ohm resistor may be installed shunting the diode, to create a permanent DC current
path to the tube and anchor the Suppressor Grid to the Cathode or ground.

4.    OPTIMISED ELECTRON STREAM © TECHNOLOGY:

       SILICON DIODE FEED

Conventional historic tube amp design using tetrode, pentode, or beam power tubes, evolve about a push-pull output stage
configuration typical of the sketch below:

In the example cathode-bias is shown, however the same principles apply for a fixed-bias design.

The notable feature of this design is that everything is symmetrical and so is easy to draw, read and understand.

The configuration of each output tube is considered to be equally distributed about the central axis of the output stage, (being the B+
and centre-tap of the output transformer).

The core electronic design concept assumes (deems) the Plate current in each tube will be identical at all times and therefore any
distortion in the plate circuits is cancelled out via the push-pull effect. For an explanation of this refer to the RCA Tube Manuals.

Screen-grids are not considered to be more than a power output/efficiency enhancing benefit derived from tetrodes v triodes, and as
shown in the example above, are not connected in any manner other than the most basic available - ie connected together and
supplied from the same power source as the Plates.

So in theory, all of the components operate in a synchronous manner such that the output stage is efficient and delivers its power at
low distortion.

Unfortunately things are not always as they appear.

The standard tolerance on components is usually around + or - 10%, however some components, including the tubes themselves, are
assigned much wider tolerances.

Modern resistors are typically + or - 5% however + or - 10% tolerance is still used.

Modern capacitors display tolerances that vary with construction material and style.

Electrolytic capacitors typically display -20% + 100% on nominal capacitance, thereby affecting gain across the frequency range.

Vacuum tubes offer a tolerance of + or - 20% on transconductance when new. Depending upon circuit parameters, performance
usually varies downwards with use. If one power tube draws grid current early on then performance of the pair will suffer.

Transformers vary dramatically. Even if the primary centre-tap is exactly in the centre of the total number of turns - ie the number of
turns either side of the centre-tap is exactly equal - the transformer may still not deliver an equal power transformation between
primary and secondary if the magnetic properties of the two halves of the primary are not identical. Also, if the secondaries are not
equally distributed about the primary the induction into each secondary may vary between halves. Factors such as core design and
winding design will influence the end result.

And what about the symmetry of the driver stage? This too is subject to tolerance variables and may deliver an unsymmetrical signal
drive voltage to the output tubes, resulting in uneven power output from each half.

So what appears to be a perfectly symmetrical design is in fact a widely varying practical configuration.

One obvious variant is Plate-current.

Many of the popular tube types do not deliver a linear response between zero signal conditions and maximum signal conditions. Whilst
most guitar amp users would be familiar with the concept of "re-biasing" after changing an output tube, it is not the norm in hi-fi
equipment. However despite good intent, guitar amp (and hi-fi amp) users are blissfully unaware that their beautifully balanced output
stage at zero signal is nowhere near balanced at maximum signal.

In lower-cost/lower quality transformers it will be found by measurement that the DC resistance of one half of the primary is somewhat
different to that of the other half - resulting in uneven Plate voltage as Plate-current increases.

Imbalance in a push-pull output stage results in loss of power and increased distortion.

This condition is more likely in high-gain tubes like the EL34/6CA7, EL36/6CM5, KT88 and 6550.
 

SCREEN-GRIDS

Ideally, the output from the Screen-grids is also identical and therefore cancels out at the output transformer centre-tap.

But note from the above design that any voltage appearing at the centre-tap of the output transformer - whether it be derived from a
difference in Plate characteristics, transformer characteristics, untransformed signal, reflected back emf from the loudspeaker, or simply
hum as a ripple voltage - will appear on one or the other Screen-gid as either direct injection or feedback.

Obviously Screen-grid current will normally be a portion of (ideally balanced) Cathode-current, but if different brand tubes - or tubes of
differing design from the same manufacturer - are used in the output stage pair it is likely that the Screen-grids will behave differently
- resulting in not only different tube characteristics across the range but also differences in the Screen-grid performances. Imbalance
becomes apparent as distortion.

It follows then that a useful object of amplifier design would be to eliminate the Screen-grid as a variable from the system.

This proposition is very well explained by Renato D. Tancinco of the Phillipines, in his 1961 US Patent 3153766

In this patent,Tancinco presents his design, which aims to eliminate crossover distortion in tetrodes and pentodes operating in Class B
mode - EL34/6CA7 users take note.

Unfortunately this design only works in Class B mode because it cancels out part or whole of the opposite polarity alternating signal in
the output stage, however it does offer significant advantage as the patent itself explains.

As noted in my Screen-Grids paper linked above, reference to tube handbooks shows that in a typical beam power tube, the Screen
Grid current at maximum signal power is around 20% of Plate current. This ratio of currents appears to be largely independent of Plate
voltage. It would therefore be reasonable to assume that up to 20% of prospective signal power is lost in the Screen Grid circuit in a
conventional amplifier. (Power = volts x amps. Power supply voltage to the screen-grids can be up to 100% of plate voltage)

(Note: Two notable exceptions are the 807 and 814 beam power tubes that incorporate advanced design technologies to increase tube
efficiency and reduce distortion, however in the overall scheme of things this technology appears to have been limited to these two
tube types - if you are aware of others please let me know)

Two conventional options present to overcome this nominal ratio of 20% plate Current:

The first is ultra-linear connection, where all the electrons collected by the Screen Grids are fed into the output transformer, but in the
process modify the output stage characteristics.

The second is to increase the value of the Screen Grid-stopper resistor to a value sufficiently high to resist the flow of electrons to the
Screen Grid.

However the Grid-stopper resistor must be non-inductive to prevent oscillation. It must also be capable of handling the Screen current
passing through it. One way of doing this is to use parallel carbon composition resistors (not film type - to prevent fire) of sufficient
number to obtain the required heat dissipation rating to do the job without excessive temperature rise in the resistors.

A further problem here is regulation of the Screen Grid supply. Obviously a Grid-stopper resistor of say 5,000 - 10,000 ohms will
present a significant voltage drop when Screen Grid current flows - if it does.

The loss of regulation may be a price we have to pay to obtain a high standard of performance.

Another and previously unpublished option to creating an operating environment where the Screen Grids will be at a DC potential
sufficiently high enough to attract and accelerate electrons towards the Plates but - to maximise power output - not to collect and divert
them to earth through the B+ supply, is the humble silicon diode semi-conductor rectifier.

By inserting a standard half-wave silicon rectifier diode in series with the Grid Stopper resistor, an electronic control circuit is created
whereby the Screen Grid will be able to be energised at DC potential attracting and accelerating electrons towards the Plate - still
electrostatically controlling current flow in the normal way - but blocking the flow of AC current from the Screen Grid back to the DC
source - ie "one way traffic"

This works because the current flow in the tube is always from the Cathode to the Anode (Plate). The diode, being a semi-
conductor, blocks current flow in the reverse direction, thus enabling DC current to feed it in the conventional manner but blocks AC
current from passing back through it to a load.

Thus then there is no circuit formed between the Screen-grid and the load so no current can flow in the usual direction.

This causes the tube to behave very much like a triode, with Plate-current being more or less controlled exclusively by Grid #1.

The diode is connected between the B+ supply (line) and the Screen Grid (load) such that the arrow points towards the Screen Grid.
ie forward current is from the line to the load.

This arrangement offers huge benefits, because it prevents the Screen Grids from collecting electrons - thereby diverting all the signal
output to the Plates, increasing tube efficiency, reducing distortion and increasing frequency response, as well as eliminating the usual
effects on changes in Screen Grid voltage on Plate Current - therefore improving transient response.

One major benefit is that the diode is not in the signal path and therefore does not modify the sound.

A secondary benefit is that there are no bypass or power supply capacitors (eg paper, polyester or electrolytic) in the Screen-
Grid signal path, which is a further major improvement.

A further benefit is that the non-linearity described under Fig. 2 above will be less of a problem for us because the Screen-
Grid component of the signal (ie those electrons normally collected by the Screen-Grid)  is diverted to the Plate.

Thus by inserting the humble silicon semi-conductor diode in the output circuit, we can completely break the bonds of
traditional audio practice and take a great leap forward!!

This is not tube heresy, because the diode is not in the signal path - it merely prevents the signal from being affected by adverse
circuit parameters such as short-circuited Screen-Grids, fluctuations in Screen-Grid voltage and power supply filter capacitors.

The relatively low reverse resistance of the diode appears to adequately satisfy the need for a low impedance return path to AC earth,
so the Screen Grid is not actually isolated from AC earth - but there is sufficient impedance in the circuit to discourage electron flow
through it.

Note: Having regard to the EIMAC articles regarding secondary emission in tetrodes referenced above, it should be the case that the
use of a silicon diode in the Screen-Grid supply will not impede reverse current flow - provided a suitable bleeder resistor is used
between the diode and B+ source - as would be the case without the diode in the circuit.

Some audiophile experimenters have used single or series strings of zener diodes in this kind of circuitry to regulate the DC Screen
Grid voltage derived either directly from the Plate or from the B+ supply, however I have used single conventional 1A 1000 PIV silicon
rectifiers (from B+ supply only) with good results.

Important Note: When Zener Diodes are inserted in series with the Screen-Grid supply to both drop and control Screen-Grid
voltage, they are connected in reverse polarity to the normal rectifier style diode described here, hence their effect on
Screen-Grid behaviour and of "sound" is quite different. A real danger with the series Zener Diode configuration is that if the
diode breaks down and short-circuits then full supply voltage will be applied directly to the Screen-Grid. This may destroy
the tube in the process.

A further (and very effective) enhancement is to use a separate diode for each Screen Grid (or each set in parallel-push-pull) to
ensure there is no cross-modulation in the push-pull activity. This places back to back diodes between the Screen Grids, which
makes each half of the AC push-pull circuit independent to the other. Less signal averaging and less cross-talk occurs between each
half of the push-pull pair, so the sound is cleaner.

In stereophonic amplifiers using a common power supply, this system provides significantly greater channel separation.

Of course, silicon diodes can be retrofitted to an existing amplifier however the negative feedback loop should be re-calibrated to suit
the changed output circuit conditions.

It may be necessary to re-calibrate Grid #1 (Control Grid) bias to ensure Plate Current and Plate Dissipation are optimised
within the tube manufacturer's design centre ratings.

In already set-up hi-fi amplifiers, it should not be necessary to change operating conditions because the Plate current is already
determined by the Screen-grid voltage and not the Plate voltage - but a prudent owner will check in any case to be sure.

If any reader can shed further light on this breakthrough new concept please email your thoughts.

Important: Please note this modification is not suited to Class B guitar amplifier applications (this means most of the "big"
guitar amps) because the sound produced by the silicon diode to Screen-Grids configuration is cleaner and less distorted,  dynamic
range (transient response) is substantially improved and power output is substantially increased - all advantages for hi-fi but not so
good for a guitar amp.

This is because many guitarists, in order to attain or emulate a particular "sound" - and thus for them "normal" usage of their guitar
amplifier - operate the amplifier into the severe distortion range by simply driving to full output or more. Some operate in the sustained
overload range continuously, using reverb, tape echo, electronic echo or acoustic feedback as a musical effect.

In a typical commercial guitar amplifier - particularly those with tube rectifiers - the power supply will collapse and simply run out of
puff when overloaded, resulting in substantially reduced B+ supply voltage, lower power output and increased (severe) distortion as
the output signal goes into square-wave like response. But the use of silicon diodes changes the tube characteristics insofar as the
normal limit on Plate current as controlled by the Screen-Grid is removed, allowing Plate current to increase in proportion with the
signal up to a maximum "saturation" point where more drive in does not produce more power out.

Note also that to maximise output power and minimise distortion in a Class B amplifier, it is vitally essential to balance each half of the
push-pull pair of tubes to ensure DC current in the output transformer is reasonably equal. The more tubes in the output stage the
harder this is to achieve.

Unfortunately, most high-power (100 W RMS +) guitar amplifiers do not provide individual grid-bias adjustments for the output tubes.
Under such conditions, to achieve balanced DC Plate-current it may be necessary to set Grid 1 bias control voltage at maximum signal
- not at quiescent (zero signal) as is popularly expounded, and to mix and match the tubes either side of the output transformer to
attain reasonably equal balance in total Plate-current per side. (This may result in some unbalanced DC hum at zero signal).

In an amplifier having a single bias supply it may be also necessary to modify the circuit to install a means for balancing AC signal
drive voltage into the output tubes because, as well as being dependent upon primary Grid #1 DC bias voltage, the Plate Current will
also be dependent upon AC input volts to Grid #1. Note though that the downside to this is that when a tube is replaced, re-biasing is
essential.

For example, I conducted a test with a Marshall Model 1959 100W super-lead amplifier, with 4 x EL34 tubes running at 520 VDC B+. 
When silicon diodes were fitted to the Screen-Grids instead of the designed 1000 ohm grid-stopper resistor - one to each Screen-
Grid, power output increased from about 90W RMS in OEM configuration to about 160 W RMS with diodes. Screen-grid current was
negligible, suggesting the tubes behave like triodes, however Plate current increased to about 180 mA per tube, which increased net
plate dissipation at full output to about 55 Watts - a certain recipe for very short tube life considering the EL34 has a rated Plate-
dissipation of 25 W.

Plate current then in this situation is primarily controlled by Grid #1 alone.

In this case the amplifier simply gets louder and louder without the usual breakdown signs before clipping, enabling the amplifier to be
overdriven continuously to self-destruction. In this particular amplifier, the power transformer is rated at about 250 VA, and a quick
calculation will show this component will also have a very short life expectancy with silicon-diodes to the Screen-Grids - but the sound
is great!!
 

GENERIC CONCEPT CIRCUIT

A conceptual circuit using the Type 803 Power pentode Tube in an OPTIMISED ELECTRON STREAM © TECHNOLOGY design as
described above, is illustrated below:
This circuit is generic to all 3 Grid power pentodes.

For Tetrodes and Beam Power Tubes just delete the Grid #3 output transformer transformer tap, and its separate Grid #3 DC supply
- a Tetrode or Beam Power Tube will have only a Grid 2 in the circuit.
 
 

Note: Westinghouse declared the following in their June 1941 user instruction sheet supplied with each 803 tube:
Note the reference to "Tetrode" connection and its effects.

This design feature offers a range of options to the DIY designer/constructor.

Important Note: In the case of a Beam Power Tube, the beam forming plates in a Beam Power Tube are not normally connected to
an active circuit element - ie are usually internally connected within the tube to its own Cathode by the tube manufacturer. If available
as a separate connection, they should be externally connected to the Cathode in the usual way.

Note: If a voltage measurement is taken either side of the silicon diode in an ultra-linear configuration circuit - ie on the screen-grid
side and on the transformer side - obviously a reading will be evident. The reading on one side will be out of phase with the reading
on the other side. DC current through the diode can be measured by inserting a small resistor (10 ohms) in series with the diode on
the supply side and reading the current through the resistor.
 
 

 4.    OPTIMISED ELECTRON STREAM © TECHNOLOGY:

        PLATE, SCREEN-GRID (GRID #2), SUPPRESSOR-GRID (GRID #3)
        and CATHODE AC CIRCUIT BYPASS CAPACITOR

4.1     OPTIMISED ELECTRON STREAM © TECHNOLOGY:

          BYPASS CAPACITOR CONSTRUCTION AND INSTALLATION

4.1.1    Non-polarised Bypass Capacitor

Irrespective of choice of Screen-Grid and Suppressor-Grid DC supply method, to obtain the full benefit from the application of 
OPTIMISED ELECTRON STREAM © TECHNOLOGY in Tetrode, Pentode and Beam Power Tube wide-band amplifiers, it is
absolutely essential to bypass (or wholly replace) the final polarised electrolytic capacitor in ALL of the B+ Plate AND Screen Grid
AND Suppressor Grid supplies, with a suitable non-polarised high quality mica, polyester, polypropylene or oil-filled paper etc.
capacitor, having a suitable value (of say 10 to 100 uF for audio), to provide an efficient and stable AC bypass at all signal
frequencies and under all operating conditions.

The reason for the use of a non-polarised capacitor is simply that polarised electrolytic capacitors offer asymmetrical impedance to the
current flow - depending upon the direction of the current through the capacitor. Most importantly for high-fidelity reproduction,
polarised capacitors offer symmetrical characteristics to forward or reverse AC current. This is evidenced by the widespread use of
polarised capacitors in loudspeaker crossover networks (which are purely AC) - even in the cheapest commercial speaker systems.

The reason for the relatively large value of capacitor suggested is that it is in series with the load on the Plate and Grids, forming an
LC or RC (depending upon the output stage configuration) series network. If the value of C is low then peaks or resonance can occur
in the audio range - particularly in the mid to high frequency band where harmonics are present.

This capacitor serves to effectively AC short-circuit (or bypass) the DC power supply and thus eliminate the power supply and its
components from the AC signal path, and to ensure that any shortcomings in polarised electrolytic capacitor performance are
eliminated - but in such a way that the signal is not significantly aurally affected.

In the case of the Plate circuit of Triode, Tetrode, Pentode, Beam Power Tube and Ultra-linear wide-band amplifiers, the non-
polarised capacitor should be installed at the centre-tap of the output transformer.

In the case of amplifiers having separate Plate, Screen-grid and Suppressor-grid DC power supplies this requirement also
applies to each of the Screen-grid and Suppressor-grid DC circuits as applicable.

To ensure adequate low frequency response each of the separate circuits must have as large a value as is practicable.

A 400 VAC (560 VDCW) Motor Start Capacitor is ideal for this function and they are readily available in non-polarised polypropylene
construction. The voltage rating must be adequate to handle the sum of the B+ DC voltage plus the AC rms signal output voltage.

I would recommend a parallel connected non-polarised capacitor arrangement, to provide 100 to 200 uF total. For higher B+ voltages
the capacitors can be connected in series (800 VAC 1120 VDCW), noting that when seriesed, the effective capacitance is halved.

A further refinement is to apply the "Rule of Hundredth's", which says that instead of using a single capacitor - eg a large electrolytic, a
bank of capacitors is installed. Each capacitor is one hundredth the value of its adjoining capacitor. For example, instead of installing
one single 100 uF electrolytic, instal a 100 uF, 1 uF, 0.01 uF, 0.0001 uF etc. wired in parallel. Lead length of the smaller capacitors
must be kept a short as possible to prevent stray capacitance or inductance. It is also important to ensure all capacitors installed are
capable of coping with the applied voltages without stress.

Of course the bypass capacitor is installed on the line (source) side of the Screen-Grid-stopper resistor and/or diode, as the case may
be.

Note: This method may result in resonances in the audio range, particularly to harmonics, resulting in sibilant accentuation and "hissy"
voice. It is thus not suitable in all amplifiers.

Note 1: Please note that the irrespective of the value of the electrolytic filter caps - even to many thousands of uF - the value of the
non-polarised cap is still critical in relation to tone, or spectral balance, across the audio range. To optimise tone experimentation is
essential.
Note 2: Modern "fast" capacitors have a different tone to older oil-filled types. It may be necessary to use oil-filled paper caps to
obtain a smoother, less harsh tone. It all depends upon the circuit design and componentry used. There is no definitive answer.
Unfortunately the older oil-filled paper caps are physically larger so need more chassis space. They also do not usually come with
flying leads, which makes wiring more difficult because the terminal lugs are exposed. Ensure voltage rating is adequate.
 

4.1.2    Electrolytic Bypass Capacitors to B+ Supply

Please note that as explained in my Power Supplies page, notwithstanding the obvious benefits from the use of non-polarised
capacitors as described above, it remains essential to good transient response to instal large values of electrolytic capacitors to store
adequate power to satisfy the demands for transient peak signals.

The question is "how big a value"?

In the case of power requirements, full particulars are provided in my Power Supplies page.

But what about "sound"?

It has been demonstrated by early amplifier designs produced in the 1940's and 1950's that the frequency response - even in the
highest quality equipment, tended to roll-off at both high and low frequencies.

Frequency response usually deteriorated as power output increased. This was an attribute ("power response") not usually presented in
the glossy sales brochures.

Modern digital recording techniques and playback media have dramatically increased both low and high-frequency response for the
typical recording, challenging even the very best of tube audio amplifier designs - particularly when played at high-volume.

But all is not lost!!

Modern capacitors offer improved performance and reliability over their ancestors. They are also available in values unheard of in the
days of tube rectifiers. (eg 32 uF was "large" in the 1950's but now 100,000 uF is "medium.")

The thin film caps of today charge and discharge much faster, enabling larger values to be used in common applications where large
values were previously forbidden - such as interstage coupling caps.
Modern capacitors offer reduced unwanted side effects, such as inductance, leakage and resonance, so offer improved high-frequency
performance and audio clarity.

One effect of this is a reduction in the level of negative loop feedback needed to offset rolloff in frequency response of an amplifier - a
definite advantage.

In the case of low frequency performance we can retrospectively improve old designs by using higher values of filter bypass
capacitors.

Since the bypass capacitor forms the return AC circuit for each stage in the amplifier, it follows that the impedance of the bypass
(filter) capacitor at any given frequency will be a portion of the total impedance of the circuit being bypassed. Now since the
impedance of a capacitor varies in direct proportion with the value of capacitance at any given frequency, it follows that providing we
reduce the value of the capacitor's impedance to a value that has minimal effect upon the circuit, then we can attain improved
performance from the circuit.

Another way of explaining the concept is that traditional tube electronic design engineering principles assume that the B+ rail is at AZ
zero (or earth) potential - regardless of the applied DC voltage.

This is a practical approach for many applications, but ignores the reality that capacitors - particularly those of the electrolytic variety -
have their own characteristics, which are injected into the circuit and therefore MUST influence the sound we hear.

Since the capacitive reactance of a 25 uF capacitor at 30 Hz is nominally only 212 Ohms, it suggests that the influence of the
capacitor may be negligible. However if the inductance of the capacitor is only 0.1H then the inductive reactance in the capacitor is
3140 Ohms at 5 kHz - and more at higher frequencies.

So we can see that increasing capacitive reactance will decrease low frequency performance and decreasing capacitor induced
inductive reactance will increase high-frequency performance.

Since measured inductance and capacitance values from typical large capacitors - eg above 500 uF - do not appear to display the
characteristics experienced aurally there may be an alternative explanation. It may be that the bypass capacitor forms an RC or LC
series network with the plate resistor and/or output transformer, forming a bass boost circuit. This is easier to see with the output
transformer, where either one end (in the case of a single-ended output stage) or the centre-tap (in a push-pull output stage) forms
the terminal at which the output is taken off. but in the case of driver stage it would seem the series decoupling resistor causes the
formation of the RC network that creates the bass boost effect.

Either way it is desirable to tune the network to the preferred frequency - eg 40 Hz

The question is what is the magic number?? How big to we need to go??

To attain extremely good low-frequency response, from practical experimentation I would suggest the minimum value of the bypass
capacitor for any stage in an amplifier or pre-amplifier, including the output power stage, may be calculated by dividing the constant
15,000,000 (15 million) by the value of the plate load resistor (or cathode-follower cathode load resistor) for that stage.

This approach will produce values of 150 uF for a 100k plate resistor and 2,250 uF for a 6,600 Ohms Plate to Plate output stage.

If those values appear frightening, then try a lesser constant - say 10,000,000

If the B+ supply supplies more than one stage or, in the case of some phase-splitters having more than one plate resistor, then the
value of the resistor should be calculated as the average of all the resistors in the circuit.

If the B+ supply supplies a stereo pair of amplifiers then the value of the plate load resistor used to calculate should be half
of the single channel value - ie each bypass capacitor should be twice the size as for a single channel.

If the output stage uses more than one pair of tubes from a single transformer, then use the transformer actual nominal rated load
impedance - not the effective impedance as seen by each pair.

It will be noticed that these values of capacitance are substantially higher than convention, however this is what I have determined
form extensive critical listening tests.

The general object of this design approach is to introduce a circuit resonance induced tonal characteristic that is pleasing.

Too high a value of capacitor will deliver undefined bass, so some tweaking may be necessary.

It is desirable for all stages to have the same tonal characteristic, so this formula assists to achieve that. In other words, each stage
should ideally have proportionately the same tonal or frequency characteristic.

Note 1:  This "rule of thumb" formula deals only with tonal characteristics - power (energy) requirements for the output stage are not
covered by this. However it is easily seen that the values produced by this design approach will supply adequate power for most
applications.
Note 2: Regardless of the value of electrolytic capacitors used for bypassing and filtering, non-polarised bypass capacitors are still
essential for good high-frequency performance and low intermodulation distortion.

Note 3: In the case of guitar amplifiers the above formula is not applicable because the lowest frequency to be reproduced is about 80
Hz. In this case a constant of say 10,000,000 or less would be appropriate.

Note 4: In the case of bass guitar amplifiers, where the lowest frequency to be reproduced is around 40 Hz, then for outstanding
results the constant needs to be in the region of 20,000,000. This value of capacitance will also deliver adequate power to the output
stage.

Note 4: Large values of capacitor can be lethal, so discharge resistors should be installed as per instructions provided in my Power
Supplies page.

Note 5: In my experience, larger bypass capacitors have the effect of improving circuit stability and improving high-frequency audio
performance.
 

4.2     OPTIMISED ELECTRON STREAM © TECHNOLOGY:

  RATIO OF SCREEN-GRID AND SUPPRESSOR-GRID BYPASS TO PLATE BYPASS CAPACITORS

4.2.1     Introduction

Since the advent of Tetrodes and Pentodes, it has been standard practice to instal a small bypass capacitor from the Screen-grid to
ground (Cathode) in Tetrodes and Pentodes used in voltage amplifier stages.

This is shown as C2 in the figure below.

The purpose of the bypass capacitor is to reduce Grid to Plate capacitance, remove undesirable audio and high-frequency signals 
such as RF components from the output before the load, and to improve the decoupling and stability of the stage.

In voltage amplifier (driver) stages the value of this capacitor has historically been in the order of 0.1 to 0.5 uF, the latter value being
considered by Radio Engineers to be adequate for good quality audio purposes. A similar situation exists in RF power amplifier stages.

It should be noted though, that this value of capacitor is usually associated with small tubes in high-impedance circuits, such as the
EF86, 6AU6, 6U8, 7199 etc, where the value of Screen-grid supply resistor may be in the range 100k to 1 Meg Ohms with very low
Screen-grid current, .

An example is shown in the Mullard High-fidelity Pre-amplifier.

 

4.2.2    Theory

For the theoretically minded, the formula for calculating the value of the bypass capacitor in a voltage amplifier stage is given courtesy
of the Radiotron Designers Handbook, 3 rd Edition (1940).
 

4.2.3    High-fidelity Audio POWER AMPLIFIER Applications

Notwithstanding the above historic convention, in high-fidelity audio POWER AMPLIFIER output stages a very different situation
applies.
If we think for a moment, it can be easily seen that there is both a DC and an AC signal path from the negative terminal of the bypass
capacitor up through the capacitor to the Screen-grid external to the tube, thence from the Screen-grid to the Plate inside the tube.
The latter will be the case (even if we do not want it) because the Screen-grid is negative to the Plate.

As explained above, the portion of Screen-grid current in a power tube can be quite high - depending upon output stage configuration
and applied voltages to the Screen-grid and Plate respectively.

Since the internal impedance of the bypass capacitor will be relatively small and the Screen-grid is operating independently of Control
Grid #1, it follows that the magnitude of the current flowing from Screen-grid to Plate will depend more or less entirely upon the actual
applied DC voltage between Screen-grid and Plate and the value of the Plate load - which will be also seen by this secondary circuit.

That is "secondary" to the primary Cathode to Plate circuit.

Obviously, the higher the difference between actual applied DC voltage between Screen-grid and Plate, the more current will flow.

In Tetrode and Pentode and Beam Power Tube applications where the Plate and Screen-grid operate at the same DC voltage,
including Triode and Ultra-linear connections, in the conventional and very common configuration shown above, it is suggested by
most writers that the Plate will function as the primary anode so long as the Plate signal voltage does not drop below the DC Screen-
grid voltage. In this case, tube manuals show that about 10-20% of Cathode current is lost in the Screen-grid circuit.

Since the bypass capacitor C1 is common to both Plate and screen-grid circuits, in terms of frequency response and dynamic
response whatever happens in one will happen in the other.

But when the DC Screen-grid voltage is less than the Plate voltage - as in RF linear amplifiers, transmitter modulator amplifiers,
transmitting tube audio amplifiers, Public Address amplifiers or OPTIMISED ELECTRON STREAM © TECHNOLOGY applications as
shown in the following diagram - then current MUST flow between Screen-grid and Plate.

This is because a secondary DC and AC circuit is established between Screen-grid and Plate, with the Screen-grid forming the
negative terminal/element.

In high-voltage amplifier designs, the voltage between Screen-grid and Plate may be in the order of several kilovolts (kV).

Since the lower leg from Cathode to Screen-grid is through the bypass capacitor, and that is external to the tube, it follows that the
limiting factors to current flow in this secondary circuit will be the Plate load impedance for AC current (the DC resistance of the output
transformer is negligible) and the value of the Screen-grid resistor if used (either voltage dropping resistor or grid-stopper) for both AC
and DC current.

For explanatory purposes, the internal tube resistance between Screen-grid and Plate may be assumed to be zero.

It follows then that the Cathode to Screen to Plate secondary AC circuit is in parallel with the Cathode to Plate primary AC circuit.

As explained above, the value of Screen-grid current - and therefore its contribution to audio power output, can be significant.

Since the current in both circuits combine together in the negative to positive Screen-grid to Plate section of the circuit inside the tube
(and thence common return to the Cathode via the B+ filter capacitor at the output transformer centre-tap) it follows that any
difference in the audio signal between primary and secondary circuits will be apparent at the output transformer.

For example, if the Screen-grid bypass capacitor is too small, that portion of audio power output contributed to by the Screen-grid
secondary circuit as described herein, will not have the same low-frequency response as the primary Plate circuit and therefore low-
frequency power output will be proportionately reduced.

For example, in the Mullard High-fidelity Pre-amplifier circuit, typical of conventional design, the Screen-grid bypass capacitors C9 is
80 times the value of C8, and C17 is 160 times the value of C12.

The value of the Screen-grid bypass capacitor will also affect the operation of the Plate circuit B+ bypass capacitor, because the B+
Screen bypass cap and B+ Plate bypass cap are in series in the signal circuit.

Hence it may be deduced that:

Where a separate Screen-grid power supply is provided,  it is most important for full-power hi-fi reproduction at very low
frequencies, and for signal balance within the tube, to ensure the value of the final Screen-grid B+ bypass filter cap is not
less than the value of the Plate circuit B+ bypass filter cap.

In the above diagram the Plate bypass capacitor is shown as C1 and the Screen-grid bypass Capacitor as C2.
 

Note: Even though the installation of a voltage dropping resistor or grid-stopper resistor to the Screen-grid (R3 in the voltage amplifier
circuit diagram above) may reduce AC and DC current, the issue of AC impedance remains - hence the grid resistor may be
disregarded for this aspect of hi-fi design.

It will be readily seen that in a power output stage where the Plate and Screen-grid share a common AC circuit return at the power
supply, and regardless of the use of a grid-stopper or dropping resistor (R3 in the above diagram) or not, the above argument does
not apply because both Plate and Screen-grid will share a common B+ bypass capacitor.

BUT - where the screen-grid is supplied through a filter choke from the centre-tap of the output transformer at C1, and bypassed by
its own electrolytic capacitor C2, the configuration shown above is still is applicable.

However bear in mind that the regulation of Screen-grid voltage is also extremely important so a separate supply should remain an
essential design element.

The above comments also apply to the Suppressor-grid

 

4.2.4    Cathode Bypass Capacitor

In the case of Tetrode or Pentode amplifier output stages having Cathode Bias, it is usual to instal a Cathode-bypass Capacitor as
shown in the following diagramme:
It can be readily seen from the way the circuit is drawn that the Cathode-bypass Capacitor is in series with the B+ Plate and Screen-
grid supply.

It follows that if the Cathode-bypass capacitor is too small, that portion of audio frequency response determined by the Cathode circuit
will not have the same low-frequency response as the primary Plate circuit and therefore low-frequency power output will be
proportionately reduced.

From the above explanations it can be demonstrated that the value of C3 must be at least equal to the value of C1.

The more theoretically minded can calculate the actual values needed for equilibrium in the frequency response characteristics for
each part of the circuit.

Where a separate Screen-grid power supply is used, as shown below,

it can be readily seen that the Plate and Screen-grid bypass capacitors C1 and C2 are in AC parallel but the set of both is in series
with the cathode bypass capacitor C3.

Hence it is also essential that the nominal capacitance value of C3 is equal to the SUM of C1 + C2.
Where separate Cathode-bypass capacitors are used to each Cathode, the above rule still applies for each capacitor.

It is usual for Class A amplifiers to use a single common cathode resistor and no bypass capacitor per push-pull pair of tubes - this is
obviously more AC linear.
 

4.2.5 Power Factor

The maximum ideal values for C1 and C2 and C3 can be in the region of up to 5,000 uF per push-pull pair of  tubes - see
warning re risk of electrocution.

To maintainOPTIMISED ELECTRON STREAM © TECHNOLOGY design standards in the output stage circuit, some circuit tuning is
essential to optimise the "POWER FACTOR" in the output stage circuit.

In terms of "sound", this is a very critical issue.

This extra step is needed to tune resonances in the output stage circuit to match the characteristics of the output tubes, output
transformer and loudspeakers to the signal "sound" or "tone" and listening room acoustics. Too low a resonance may result in a "dull"
or "flat" bass sound. Too high a resonance may result in "bass boom" and loss of definition.

To optimise the output stage circuit and deliver maximum power from minimum output impedance with maximum loudspeaker damping
characteristics, changes in power factor produced by the inductive reactance in the output transformer will ideally be cancelled out by
the capacitive reactance in C1.

To optimise the value of C1 some degree of practical "trial and error" experimentation is needed - a frustrating experience but one
worth the effort.

Note C2 and C3 need to be suitably modified as described above and hereunder to maintain the correct ratio to C1 - ie the same
capacitance value.

Note also that to overcome non-linearity in the AC bypass circuit, C1 and C2 may need to be bypassed by a small stabilising capacitor
in the region of 0.5 uF - but since this small capacitor will directly affect the high-frequency response, it too must be chosen with care
regarding both size and material of construction.

IMPORTANT:

All of the Plate Circuit output stage power passes through the AC circuit formed by the output transformer and C1 - and
whatever else shunts C1.

Since L1 is directly connected to the Power Supply, it follows that the Power Supply is in series with L1 and shunts C1.

This is illustrated in the diagram below:

So to minimise the effects of the Power Supply rectification and filter circuitry on "sound", the value of L1 should be as
large as can be practicably sustained - noting that the DC resistance of L1 will directly reduce the available B+ voltage to the
Plates.

The larger L1 is, the more AC current will be forced through C1 and less through L1 and associated circuitry.

Another way of expressing this is to say that whatever AC power is lost into L1 and its associated source bypass
components - all of which shunt C1 - the less linear the output stage will be at low frequencies.

An inductance of about 10 Henries (minimum) is desirable for L1.

This also holds true for simple one stage filter choke systems, because the rectifier/filter is always shunting C1 - thus will
always affect its performance and effect on "sound".

IMPORTANT: To prevent instability in the power supply (which dramatically affects audio sound as heard through the
loudspeaker) it is desirable to ensure C2/L1 and C3/L2 are of equal value. Where only C2/L1 are used then it is desirable that
C1 and C2 are of equal value. It is thus obvious that a three stage filter enables a higher value of C1 than for a two stage
filter.

This point will be expanded in future revisions of this page.

 

4.2.6    Silicon Diode Feed

When silicon diodes are installed in the Screen-grid and Suppressor-grid circuit when using my OPTIMISED ELECTRON STREAM ©
TECHNOLOGY design, the above bypass requirements still apply because although the diode will block AC from passing from the
Cathode to the Screen-grid and Suppressor-grid internally, it cannot stop the current flow in the external secondary circuit described
herein.
 

4.2.7    Voltage Doubler Power Supply

Where the Screen-grid B+ supply is taken from the output of a voltage doubler rectifier, it may be the case that for convenience the
Screen-grid supply is taken from the mid-point of the two series-connected filter capacitors.

However if we analyse the effect of this arrangement having regard to the above, it is easy to see that the external Screen-grid
bypass current is passing from negative to positive through the lower filter capacitor.
It is also easy to see that return external Screen-grid bypass current is passing from positive to negative through the lower filter
capacitor.

Consequently, it is evident that the two currents will cancel out.

Bad move!!

So in other words, this configuration is not desirable for high-fidelity reproduction, even though it may appear to work satisfactorily in
public address amplifiers.

There is also an issue of the effect of differences in capacitor characteristics in the forward and reverse directions, as well as the
effects of distortion caused by the output transformer.

There is also the issue of excessive ripple in the DC supply.

One solution to both issues may be to instal a 60 mA filter choke in the line (supply) side of the Screen-grid B+, then add a second
filter capacitor to the Screen-grid B+ at the load side, to provide a separate path for forward and reverse currents.

The value of the second (final) bypass capacitor should be equal to or greater than the value of the seriesed pair.
 

4.2.8    Capacitive Voltage Dividers

Where the Screen-grid B+ supply is taken from the centre-point of a pair of series-connected electrolytic capacitors in the Plate B+
circuit, the same problems arise as explained above.
 

4.3     OPTIMISED ELECTRON STREAM © TECHNOLOGY:

          SCREEN-GRID AND SUPPRESSOR-GRID BYPASS CAPACITOR SIZE:

In the case of OPTIMISED ELECTRON STREAM © TECHNOLOGY where a separate Screen-grid and Suppressor-grid power
supply is provided, for full-power hi-fi reproduction at very low frequencies it is essential to ensure the value of the final
Screen-grid B+ bypass filter cap (and Suppressor-grid in the case of a Pentode) is NOT LESS THAN the total effective value
of the the sum of all Plate circuit B+ bypass filter capacitors directly connected to the centre-tap of the output transformer -
polarised and/or non-polarised.

Where Cathode-bias is employed, in addition to the above it is also essential to ensure the value of the Cathode-bypass
capacitor(s) is not less than the value of the Plate and screen-grid supply capacitors

Note: Where one or more filter chokes are used in the B+ supply, the value of capacitors before the choke(s) may be ignored for this
requirement.
 

4.4    NEGATIVE LOOP FEEDBACK

Where negative loop feedback is used from the loudspeaker terminals back to an early voltage amplifying stage, it can be seen from
the above that any defects in performance in the output stage will simply be transferred back to the input, with the result that the
amplifier will always be in a constant state of correcting itself.

Hence any use of negative loop feedback should be deferred until the output stage is optimised according to the requirements of my
OPTIMISED ELECTRON STREAM © TECHNOLOGY
 

Please refer to my SCREEN-GRIDS page for full particulars of this requirement.

5.    OPTIMISED ELECTRON STREAM © TECHNOLOGY:

       CONTROL GRID (GRID #1) AC CIRCUIT BYPASS CAPACITOR

In output stages of audio amplifiers it is common to use FIXED BIAS or BACK BIAS to apply and control the Grid #1 - Control Grid
bias voltage in the output tubes.

Designers always consider the requirements for direct current operating conditions but often ignore requirements for alternating current
conditions - ie signal voltage.

The Grid #1 Resistor to each output power tube forms part of the load for the preceding driver stage. Hence for maximum efficiency
and stability, the return path from the output of the driver stage circuit back to its Cathodes should be direct and have very low
impedance.
 

5.1    OPTIMISED ELECTRON STREAM © TECHNOLOGY Principles

Bias, by definition, requires a voltage potential to be present between the Control Grid and the Cathode.

CATHODE BIAS usually results in the Control Grid being at nominally 0 VDC and the Cathode at the required bias voltage being +
VDC.

FIXED BIAS usually results in the Control Grid being set at the required bias voltage being - VDC and the Cathode at nominally 0
VDC. In the case of FIXED BIAS it is essential to bridge the difference between the central axis of the AC signal input and the power
tube Cathodes, such that the central axis is at the same AC voltage as the power tube Cathodes.

It is essential for stable high fidelity operation to ensure that:

a)    the Grid #1 circuit has a reliable and predictable low-impedance return path for the AC signal voltage from the preceding stage
b)    the Grid #1 circuit has a reliable and predictable low-resistance return path for the DC bias voltage between Grid #1 and the
Cathode of the same tube.
c)     in a balanced driver stage for Classes AB, AB1, AB2 and B, where the power tubes are biased towards cut-off and driver stage
output signal voltages are symmetrically balanced about a virtual central axis having a nominal potential of 0 VAC, it is imperative to
ensure the junction of the two Control Grid resistors of the output tubes be AC grounded to 0 VAC.

Thus in an RC (resistor/capacitor) coupled amplifier, it is a fundamental requirement that the Grid #1 Resistor to each power
tube provides a return circuit path to the Cathode of  BOTH the preceding driver tube AND the power tube to which it is
connected.

This is usually, but not always, through the earth or ground terminal of both AC and DC applied circuits.

Fortunately, in most amplifiers using FIXED BIAS or BACK BIAS, the Cathodes of the output tubes are directly earthed (to the
chassis), thereby providing a convenient return circuit at 0 VAC potential.
 

5.2    Cathode Bias

In most amplifiers using CATHODE BIAS, the Cathodes of the driver tubes are earthed either through a bypassed or unbypassed
Cathode resistor, hence the Cathode terminal is indirectly AC earthed to complete the return circuit for the driver tube(s).

This principle applies particularly to CATHODE BIASING of power tubes, where the Cathode terminals may well be at say +50 VDC
but simultaneously at  0 VAC. . (An exception is where the output stage is in Class A and the common cathode resistor is unbypassed
to develop negative current feedback in the Cathode circuit.)

Typical Ultra-linear output stage configuration with CATHODE BIAS. Note the bypass capacitor on the common Cathode resistor,
thereby providing a low-impedance AC path to ground.
 
Typical Triode-connected Pentode output stage configuration with adjustable CATHODE BIAS.
Note the absence of a bypass capacitor on the common Cathode resistor.

This circuit features + or - adjustment of common bias voltage and also fine adjustment of balance between the tubes.
 

5.3    Fixed Bias Adjustment

It is common practice in FIXED BIAS amplifiers to incorporate an adjustable bias control circuit, incorporating an adjustable resistive
network to enable precision adjustment to the Control Grid voltage and/or Plate Current of the output power tubes.

Typically, this negative polarity DC voltage is sourced from a half-wave or full-wave tube or solid-state rectifier, filtered by a simple
resistor/capacitor network.
Typical Fixed Bias amplifier incorporating adjustable bias to the power tubes.

Note: In this design, the 20 MF (uF) capacitor serves to bypass the AC return circuit from the driver stage to ground, thereby providing
a direct return circuit path to the Cathodes of the driver tube. This capacitor is connected in such a way as to ensure the Control Grids
are equally earthed at all times through their respective Grid Resistors, regardless of applied DC bias voltage. It also ensures the
central axis of the balanced signal input is at the same potential as the Cathodes - an extremely important component.
 

5.4    Bypass Capacitor Material

In commercial amplifiers with FIXED BIAS, to minimise cost the bypass capacitor is nearly always of the polarised electrolytic variety,
but this means that the AC return circuit is not symmetrical. (Electrolytic capacitors have different characteristics in positive and
negative polarity circuits).

Consequently,to ensure symmetrical AC circuit configuration it is essential to instal a suitable non-polarised bypass capacitor
into the circuit at a point closest to the Grid #1 of the output tubes as is practicable.

Typically this will be at the junction of the two (or more) Grid Resistors.
If a driver transformer is used, and its centre-tapped secondary is not directly connected to the power tube Cathode circuit,
then instal the bypass capacitor between the centre tap of the secondary and earth.

It is therefore absolutely essential to bypass (or wholly replace) the final polarised electrolytic capacitor, with a suitable non-polarised
high quality mica, polyester, polypropylene, paper or oil-filled paper capacitor, having a suitable value (of say 1.0 uF or a 10 uF or
more motor start capacitor as above for audio), to provide an AC bypass at all signal frequencies and under all operating conditions.

This small capacitor serves to effectively short-circuit (shunt or bypass) the DC power supply and thus eliminate the power supply and
its components from the AC signal path of both the driver tube and power output tubes, to ensure that any shortcomings in
polarised electrolytic capacitor performance are compensated - but in such a way that the signal is not significantly aurally affected.

Importantly, it also provides an automatic safeguard against the adverse effects of poorly contacting bias potentiometers and/or
adjustable wire-wound resistors.

Normally, the value of the bias supply capacitor will not exceed 10 uF (to ensure fast charging to full bias voltage before the output
tubes heat up and commence to conduct Cathode Current) so an extra one or two uF will not significantly affect the charging circuit
performance.

If preferred, to achieve the same end result as described above, the final Control-grid power supply filter polarised
electrolytic capacitor (and the first as well if so inclined) can be wholly replaced with a polyester or paper or motor start
capacitor of say 8 uF value and having a suitable DC voltage rating.

Suitable capacitors may also be reclaimed from unwanted fluorescent lamp-holder assemblies.

For the ultra-fastidious, the "Rule of Hundredth's" may also be applied.

Note: There is no practical limit as to the value of the bypass capacitor, provided the bias supply is capable of charging it quickly to
ensure bias voltage is present when the power tubes warm up and commence conducting. I have successfully used values around
100,000 uF, shunted by suitable non-polarised polypropylene caps.

6.    OPTIMISED ELECTRON STREAM © TECHNOLOGY:

  DECOUPLING OF THE POWER AMPLIFIER FROM THE POWER SUPPLY
       BY MEANS OF A SERIES CONNECTED SILICON DIODE
 

The conventional B+ supply looks something like the following diagram:

It is explained above why Filter Capacitor C1 forms a vital part of the AC circuit in the output stage.

It is also explained why it is desirable to "optimise" the value of C1.

Further, it is explained that the Power Supply is connected in parallel with C1 and therefore also forms part of the AC circuit of the
output stage.
Consequently, it can be easily demonstrated that the "sound" of the amplifier will be directly affected - maybe adversely - by the
characteristics of the Power Supply.

Since the object of OPTIMISED ELECTRON STREAM © TECHNOLOGY is to "optimise" both the performance and sound of the
amplifier, it follows that if the Power Supply can be isolated, or decoupled, from the power stage then a positive benefit might result.

This object may be easily and economically realised by installing a series connected silicon diode rectifier into the B+
supply - having its arrow pointing towards the load - between the Power Supply output terminals and C1 - ie AFTER  the
Power Supply and BEFORE C1.

Note that C1 must always remain part of the Amplifier, because it forms a direct path in the AC output circuit.

So regardless of whether the Power Supply is of the simple centre-tapped full-wave or full-wave bridge rectifier or full-wave voltage
doubler rectifier, or it has a capacitor filter, choke filter or some other combination of filter - or even includes a voltage stabiliser or
regulator - to ensure the OPTIMISED VALUE OF C1 is not altered by shunting the Power Supply circuit, significant improvement to
performance can be made by installing a series silicon rectifier diode as described.

To ensure cool operation and reliability, the current rating of the rectifier should be generous in relation to the power requirements of
the amplifier - eg 3A or 6A. The rectifier can be of the same specification as that used in the Power Supply B+.

Where a separate Screen-grid Power Supply is used, then the same principles apply to C2.
Where an inductor - such as L1 above - is used in the Power Supply, it is preferable for the diode to be installed after the inductor - to
ensure the Power Supply is completely isolated from the amplifier.

This method makes it easier to adjust the tonal balance or "tone" of the amplifier power stage - because fewer components are now in
the signal path - therefore there are fewer interactive variables to deal with when changing component values.

The same principle apply to L1 and C2.

It will be seen for the above that in the case of simple filters having just one capacitor after the rectifier - and nothing else - the
requirement described is met. Unfortunately that simple configuration results in significant hum and ripple and poor quality direct
current (DC) - so is not recommended for high-fidelity amplifiers.
 

7.    OPTIMISED ELECTRON STREAM © TECHNOLOGY:

  DECOUPLING OF EACH STAGE OF THE VOLTAGE/POWER AMPLIFIER
       FROM ITS NEXT-FOLLOWING ADJOINING STAGE BY MEANS OF A
       SERIES CONNECTED SILICON DIODE

Having described how decoupling of the amplifier from its power source facilitates and enhances performance, it follows that the same
principles apply to driver stages.

In a conventional amplifier, the stages are connected in a "cascade" configuration.

That is to say, each stage "cascades" into the next - just as a stream "cascades" over its riffles as it runs down a slope.

Conventional theory says that up to three stages may be connected to the same B+ supply before instability becomes a problem.

However:

For optimum performance each stage should be a self-contained discrete circuit - wholly independent from those before and
those after.

The usual method of decoupling is to instal a series resistor into the B+ supply between stages. But this produces significant voltage
drop when all or any tubes draw current over the steady state condition.

This convention also allows signal from one stage being conducted to an adjoining stage when there is a signal voltage difference
between those stages.

So to produce effective decoupling between stages and to prevent each stage sharing the bypass capacitors and associated circuitry
installed in adjoining stages - before or after - it is essential to decouple effectively and completely.

To understand the OPTIMISED ELECTRON STREAM © TECHNOLOGY: SERIES DIODE DECOUPLING concept first we need to
consider the amplifier as a "system"

A "system" is defines as a series of "processes", each having an input and an output. The ideal system is self-adjusting or
self-correcting, by means of a feedback loop (s).

It follows that in such a system what happens in one part also happens in others.
In the case of an amplifier, the signal is present on the B+ rail at each individual stage. In some cases the signal is in phase with
other stages, and in other cases the signal is out of phase with other stages.

Therefore the B+ rail is common to all stages. Looking at it another way (ie from the B+ line), the B+ rail is the output from a simple
parallel mixer system - each stage being an input to the mixer.

Clearly then the number of stages and those in phase and those out of phase will affect the "sound" of the amplifier.

Conventional design principles regard the junction between the plate load resistor and the immediately adjacent filter cap as being at
AC "earth" - because the filter cap is regarded as being a very low (insignificant) impedance path for the AC in the circuit

However practical tests reveal a different story

I submit any given electrolytic cap has different characteristics when comparing its forward and reverse current flow. One way to
demonstate this is to replace an electrolytic in the B+ rail with a non-polarised motor start cap - the difference in tonal quality is
huge.

This proves that the junction of plate resistor and filter (signal bypass) cap is not at AC earth at all, but at some point above it - or at
best the AC signal circuit will be affected or influenced by the series connected bypass capacitor's inherent internal characteristics.

Now since we are talking a VOLTAGE amplifier we do not need to consider the CURRENT in the B+ rail but the VOLTAGE

Specifically, the TRANSIENT PEAK VOLTAGE

When a transient signal appears on the B+ rail at any point it will appear across the entire rail - not just at the stage in which it
originates

So if, for example, the phase-splitter B+ voltage momentarily sags, the signal from the previous stage - which does not suffer as
much sag because it is a low current stage - will flow to the phase-splitter stage and be mixed with it at the filter cap

Ohm's Law applies, so wherever there is a voltage difference you will see current flow - in whatever direction it chooses

The purpose of the series diode in the B+ rail is to prevent SIGNAL voltage from transferring forwards (positive feedback)  in the
circuit whenever a voltage difference in the B+ rail appears

It will not of course reduce negative feedback between stages through the B+ rail

Provided a suitable diode is installed to each and every stage in this way, this simple device prevents the AC in any single-stage
circuit from being shared by an adjoining upstream (line-side) stage - thus ensuring all of the signal current is passed through the
stage discrete bypass (usually "electrolytic") filter capacitor.

Thus each stage can be optimised in its own right.

 

8.    OPTIMISED ELECTRON STREAM © TECHNOLOGY:

  SYMMETRICAL BALANCED AC SIGNAL DRIVE SYSTEM.

This element of OPTIMISED ELECTRON STREAM © TECHNOLOGY describes the design principle wherein the AC signal drive
system circuit comprises a symmetrical balanced AC signal drive system whereby the AC signal OUTPUT from the Driving
stage circuit and the AC signal INPUT of the Driven stage circuit must both be directly connected to a common central axis
at reference voltage potential.

That is to say, in a push-pull amplifier stage, the mid-point central axis of the AC push-pull signal input voltage MUST be directly
connected to the mid-point central axis of the the AC push-pull signal output voltage - ie the INPUT voltage to the driven stage.
 

Transformer Coupling/Driver System

In a conventional transformer coupled circuit, the supply terminal of the single-ended driving side, or centre-tap of the push-pull
driving side is earthed via the B+ bypass cap.

In this case, the return circuit for the AC signal is through the transformer primary back to the driver cathode via that B+ bypass cap.

Being connected to earth it follows that the driver cathode bias resistor should be also connected to earth. In the diagram below that
would be via a centre-tap on the driver Triode AC filament transformer.

In some Class B designs using zero bias tubes, and in a cathode bias design, the centre-tap is usually connected to ground as
shown.
 
 
But in a fixed bias design, the centre-tap of the transformer secondary is usually connected to the bias supply.

Thus when the centre tap of the secondary is connected to the bias supply then the primary and secondary centre-taps do not share
a common connection - except through the bias supply earth point, which is at positive polarity in respect to the bias voltage.

This can be seen in the following design, where the single-ended driver is coupled to a push-pull output stage.

Note the cathode bias resistor and bypass capacitor in the driver stage are grounded but the centre-tap of the transformer
secondary is connected to the negative voltage DC bias supply.

But being a transformer none of that matters, because the primary and secondary are two completely independent and
isolated circuits.
 

RC Coupled Driver Stage

However when a tube RC coupled driver circuit is used with either plate or cathode output - then the two circuits must share a
common in and out AC signal axis - or else the sound will be affected.
This is explained below.
 

Fig 1: Cathode Biased Output Stage

In a cathode biased amplifier the requirement for driving and driven circuits to share a common connection is normal, conventional
design practice. A typical standard configuration is shown in Fig 1.

Notice how the central axis of the driving and driven circuits is common. In this case the central axis is at ground potential.

Since the circuitry in both halves of the push-pull circuit are symmetrical and exactly equal it follows that the AC signal voltages in
both halves of the push-pull circuit are symmetrical and exactly equal.

However this is not the case with fixed bias amplifiers.

 

Fig 2 and 3: Fixed Biased Output Stage

Fig 2 shows a conventional fixed bias circuit.

Fig 3 shows a conventional fixed bias circuit with the bias supply shown.

Notice how these circuits also appear to be balanced each side of the push-pull central axis.

However these designs are not balanced and are not symmetrical.

Further explanation is provided below.

 

Fig 4: OPTIMISED ELECTRON STREAM © TECHNOLOGY Fixed Bias Symmetrical Balanced Drive System.
Notice how this circuit appears to be the same as Fig's 2 and 3.

However there is a subtle difference.

Notice how the Cathode circuit of the driving stage is not connected to ground - as is conventional design practice - but to the bias
supply for the driven stage.

This is the profound difference.

Further explanation is provided below.

 

Fig 6: Conventional Fixed Bias System.

Fig 6 shows the AC signal path in a conventional fixed bias system.

Notice how the AC signal must pass through the bias supply components - particularly the bypass electrolytic filter capacitor (usually of
the electrolytic variety). In this case, the bias supply shown is simple, but in most designs the bias supply is complex and often
comprises chokes and other harmonic producing or refining components - in complex inter-relationships.

Moreover, it is common practice to use half-wave rectification in bias supplies, so the residual ripple voltage will appear as a series of
pulses in series with the grid driving circuit.

Consideration must also be given to the effects of spurious harmonic and transient spike signals injected into the bias circuit from the
bias supply mains transformer.

Since this part of the driving circuit is common to both halves of the push-pull drive, it follows that the signal will ALWAYS
be modified by the bias supply characteristics.
 

If that was not enough to contend with, it can also be seen that the bias supply negative DC voltage is permanently in the signal
circuit.

Furthermore, the bias voltage is offset negatively to the centre-line axis of the signal voltage, which is nominally at ground potential.
The greater the applied bias voltage the greater the offset.

Since the AC signal alternations in both halves of the push-pull circuit have both positive and negative polarity, it follows that the bias
supply voltage will enhance the signal in the negative alternation and offset the signal in the positive alternation.

This DC voltage and current source will try to support the flow of AC signal sourced energy to the grids when grid current flows in
Classes AB1 and AB2 or B - and also Class A if gas is present in the power tubes . However to get to the grids it must first pass
through the driving tube circuit. Since the interstage coupling capacitor will not pass DC it follows that the DC component of grid drive
power will be dissipated into the plate resistor of the driving tube.

But along the way it can increase the cathode bias voltage of the driving tube (which will attenuate the signal) whilst simultaneously
reducing the plate voltage (which will attenuate the signal and reduce AC output voltage)  - thereby affecting the capability of the
driving stage to respond to the signal input.

Since the plate load resistor of the driving tube provides an alternative return circuit (to ground) for the signal voltage it follows that the
conventional fixed bias configuration which adds DC currents into the signal circuit includes elements guaranteeing circuit instability.
When negative loop feedback is added to such a circuit - eg from the speaker secondary windings of the output transformer back to
the driving stage - then the result must be difficult to predict or control.

Finally, the bias supply final shunt load resistor is in series with the output tube grid resistor and forms an integral part of the grid to
cathode circuit of the output tubes. It is common to see relatively high values of bias supply final shunt load resistor compared with the
output tube grid resistor value - eg 47k to 100k. Sometimes complex bias adjusting networks are also in there. It follows that the
closer the (total) bias supply resistor is to the grid resistor value then the more effect the bias supply characteristics will have upon the
AC signal behaviour.
 

Fig 5:  OPTIMISED ELECTRON STREAM © TECHNOLOGY

Fig 5 shows the symmetrical balanced AC signal drive system whereby the AC signal OUTPUT from the Driving stage circuit
and the AC signal INPUT of the Driven stage circuit have a common central axis at reference voltage potential.

In this example, the bias is set at -40 VDC.

Notice how using this system as shown in Fig 4, the signal path circuits on both sides of the push-pull circuit are exactly equal and
balanced - in a similar manner to that of Fig 1: Cathode Bias Output Stage.

Moreover, the signal path is direct between both driving and driven stages and is completely independent of and therefore unaffected
by the bias supply.

Consequently neither actual bias voltage nor bias supply characteristics will affect the signal circuit.

Note that the grid to cathode circuits of the power tubes is the same for both Fig 5 and Fig 6 designs.

This system is very stable and provides enhanced frequency response and reduced distortion.

It is entirely suitable for a cathode-follower driver stage - ensure the base of the cathode follower load resistor is connected to the bias
supply as shown in Fig 5.

An added benefit for a cathode-follower driver system is that the cathode load resistor now spans the bias voltage as well as the B+
voltage. Consequently the cathode follower load resistor may be made substantially larger whilst still maintaining the same plate to
cathode voltage in the tube. This can offer increased output voltage to the driven stage.
 

Load Resistor

Generally speaking, the greater the load resistance/impedance to the driving stage the higher the output capability.

However there are practical limits to this objective since output tube grid circuit resistance must be held within manufacturers'
published specifications.

To maximise the driver stage load whilst minimising driven stage grid circuit resistance the RATIO of driver stage load plate resistor to
driven stage grid circuit load resistor should be held within the range of about 1:1 to 1:2.
 

Most of us have a junk box stock of perfectly good tetrodes, pentodes or beam power tubes just waiting to be used - so why not
experiment and prove the benefits of OPTIMISED ELECTRON STREAM © TECHNOLOGY to yourself.
There is no restriction or cost imposition upon the home hobbyist constructor to using these concepts - the only restriction is on
commercial exploitation - so if you do not like it do not do it.

If you want your hi-fi to improve its performance at minimal cost to you then experiment. The concepts presented here do work and
cost very little to implement.

However to those who say that a product is only as good as what you pay for it, then these concepts are of no value to you because
they are free. You would be wiser to spend a hundred grand on a commercial system and feel better. While you are so doing, ask the
manufacturer to justify the circuit design and component choices to you - ie why the design is what it is and not some other alternative
approach. That is "why is it better?"

To those who consider relating these concepts to RF applications - this is an audio focused site - experiment at your own risk. It may
be safer to stay with the tried and true - waste a little power and live with it.

However RCA, in their Transmitting Tube Manual TT4, say at Page 44 -" The only restrictions on tube operating values are those
imposed by the published maximum ratings." The OPTIMISED ELECTRON STREAM © TECHNOLOGY design concepts presented
above should enable designers to withdraw operating conditions back into specification whilst improving performance.

Finally, an OES tube amplifier can only and will always sound like a tube amplifier. The OES concept is limited to and by the vacuum
tubes it relates to.

HAPPY CONSTRUCTING!!

MAY YOUR PROJECT BE A SUCCESS!!

 
 

REMEMBER:

- ALWAYS TAKE CARE WHEN WORKING WITH HIGH-VOLTAGE -

DEATH IS PERMANENT!!
 

© NOTICE:

INTELLECTUAL PROPERTY COPYRIGHT © D.R.GRIMWOOD 2002 - ALL RIGHTS RESERVED.

Intellectual property in the applied engineering concepts expressed in this paper remains exclusively with the author.
 
 

 
 
IMPORTANT NOTICE

THE AUTHOR MAKES NO CLAIM WHATSOEVER AS TO THE VALIDITY OR ACCURACY OF ANY

STATEMENT, INFORMATION OR OPINION CONTAINED IN THESE PAGES AND NO LIABILITY WILL BE
ACCEPTED FOR ANY ERROR OR OMISSION OF ANY KIND WHATSOEVER.

PLEASE NOTE NO WARRANTY IS EXPRESSED OR IMPLIED AS TO THE WORKABILITY OR

PERFORMANCE OF DESIGN INFORMATION DESCRIBED HEREIN.

For suggestions, critique or discussion re this page contact:

DENNIS GRIMWOOD
 

Email:   contact
 
 

This page is located at http://www.oestex.com/tubes/oes.htm

 
 

electron's Home Page

This page last amended 11 July 2012

 Telrodes with 5creen feedback
FURTHER LICHT ON THE SO·CALLED "ULTRA·LlNEAR" CIRCUIT

AITER a period do caution, amounting in some

quaners to undisguised scepticism, the .. ulua-
characteristics to the other. When the screen tap-
ping point is property adjusted the uansfer charac-
linear" output stage l . ' " , . is undoubtedly here to teristic il more nearly linear ~d distortion is less
stay. It was unfortunate, though, that Hatler and than that of either the teuode qr the triode connec-
Kerces in popularizing this circuit for audio ampli- tion. Obviously some factor is at work which is not
fiers should have chosen a term which, if it means present in either 01 the limiting conditions and
anything. suggests that the uansfer characteristic ce triode-teuode ~ is misleadingly simple. If it is
has been bent .. beyond the suaight" and is there- called the ee UL" circuit the special nature of its
fore still curved! performance is underlined, and we do not have (0
• Several alternative descriptions have been sug- grit our teeth over that .. beyond the linear.1t
gested, the most plausible being .. triode-tetrode .. The UL nom~nclature is, incidentally, adopted by
operation. This hardly does justice to the circuit. F. Langford-Smith and A. R. Chesterman who bavl'
since, although at the extreme limits of the screen recentlr carried out an exhaustive experimental
upping (Fig. 1) the valve is undoubtedly operating investigation of the push-pull circuit (Fig. 2). The
either as a triode or a tetrode, the intennediate tap- results of their measurements with KT66s are given
pinl points do not give a progressive transition. so in Fig. 3 and it will be DOted that they have taken
far as distonion is concerned, from one set of the ttOuble to adjust the load resistance and bias
for the best performance at each tapping point. This
--------~-------+Mr ensures that the effects of SCtccn feedback will DOt
TAPPING POINT be modified or obscured by unfavourable operating
(M_... rcd fl'OlD + bot.) condiuoos.
lmpedaac:e Tul'Bll The curve for manmWD power shows a clear
("'0) (%> minimum for a screeD tapping of about lS%, and
0 0 a similar though leas pronounced minimum occur!!
S
JO
22.4
31.6
at about 20% under minimwn distortion conditions.
is 38.' Both these minima are of lower value than the dia-
20 44.7 tortion present under optimum triode conditions.
25 50
30 50... Any reduction in iDhereal dlstoniou iD the output
50 70.S stage reduces the degree of overall feedback
100 100
requircci for a PVCJl amplifier performance and 10
I.,UT increases the stability margin, but the improvement
Left: Fif. I. So.le over triode performance by itself would seldom
connections of UL out· justify the expense of the extra primary taPpinlS.
put "t'olve. The real advantage of UL operaticm is that triode
~--------~~----H~ performance in the matter of low inherent diltomon
is achieved with a power
d6ciency pc:.rformmcc
approac:biq that of a
pentode. Por a given
audio-frequency power
output and distortion
level, smaller 0 u t put
valves and a leu expen-
sive power supply unit
can be used with the UL
circuit than would be
necessary with triodes in
the output stage.
For a given anode and
screen supply voltage the
available power output
from a pair of valves in
the UL circuit is always
less than that given by
the same valVCI operated
as pentodes (retrodes) (sec
Fig. 3), and the voltaiC
gain is also Jess. It is
sometimes argued that,
Fif. 2. Circuit used by loltf(ord-Smith ond 'ltestermon os 0 bosis for meOJurements provided the amplifier
of power ourpur and distorr,on liven in Fir. l. has a stability margin
capabJe of accepting the
higher overall feedback
necessary to reduce dis-
tonion, the same results
will be obtained by using
nonnal pentode opera-
lion. Langford - Smith
points OUt~ that tht volt-
.Ige gain characteristic of
.1 pcntode stage (Fig. 4 (a»
15 far from linear com-
pared with the UL cir-
cuit, and that with pen-
todes the feed back near
full power outpUt will be:
reduced-just where it is
most wanted. It is also
stated that since the
maximum-sisnaJ cathode
current is less with UL
than with pentode opera-
lion and the cathode
current efficiencies are
approximately the same, Fif. 3. Variation of total harmonic distDrt;on and ma.rimum fxrnr (fMlak InpuE_,r/~
il should be possible to biOI) with '''teII topp/n,. Load Gnd biG' odJusted (or optimum perfOlmonce at eodt
meosur~ polnE us;'" a pair o( 1<166 valves w/rh 300V atJOde and screen .upply_
increase the anode voltage
to bring the UL power 1 no .,
output up to the pcntode I ~F
level.
In the test circuit (Fig. ..-.
2) used by Langford-
Smith and Chesterman it i
2501-----1t----1IT-~!____fl..L----4 5 I
6
--
,.t
-I JOG

~
/
.
If
':,
,
,
--.
s I
will be seen that anti- 2
! I ~ I I
!
l'.lrasitic measures have
I 4 200
1
4
e
, I.. -
ZOO
h.-en liberally applied
...
0

/ I

,
1
...... 150
I
and the authors mention I
a tendency towards in- I
~
mbility which is attri-
buted to the multiplicity Z ...= ~ 100 I I
I
2
of tappings and their
associated switches. This
~
~ i
50
j
I TOTAL
1lANt000IC
DI5TORTlOil ,
•1 I
I
~ U
tendency to instability in
I~e UL circuit must not
h:: overlooked. It is S 10 IS 20
0
25
V-
0 , 10 I' 20
• 0
;. iosely related to the r.III.' INPUT VOlTS (£loCH $10[) r.lII.l. 111M YO\. T5 (EAcN 510£)
.k,ign of the output (4) ( b)
: : J nsf 0 r mer and is
Ji)cusscd in detail else- Fir. <t. Voltare transfer charocterl't/cs o( KT661. (a) a' tetrodes. (b) uttder 20% to~ (,II.
where in this issue. condirions. The vertical dotad line Indicates the le'lel at which peak input eqUQIs the
.. Mechanism" of the grid bias•
UL Circuit.-Although
the circuit behaves, so far as reduction of gain and An alternative and more plausible hypothesis
output impedance arc concerned', according to the recendy published·, takes into account the non-
known laws of feedback circuits and shows a smooth linearity resulting from multiplicative mixing when
:.ansilion from the pentode to the triode condition, feedback is applied to an electrode other than the
:he conventional feedback formula: fail to account for input grid. It is known that non-linearity can be
rhe dip in the distortioD curve at a critical screen introduced into an otherwise linear valve charac-
;.':'ping point (which varies from valve to valve). teristic by applying feedback to the suppressor grid.
I t has been suggested that non-linearity in the This fonn of distonion will be present also when
",reen/anode characteristic may offset curvature of the saeen characteristic is itself linear. It is shown
lhe cODtrol grid characteristic, but this cannot be mathematically that feedback can be critically
:asily checked as the screen characteristics of power adjusted to cancel a particular harmooic (in practice
:lutput valves are not usually included in the makera' the third) and that al all even harmoaics arc already
literature. But is this basically the right explana- amcel1cd by push-pull operation the residue mull
tion? If the screen curvature is sufficient to cancel comilt of higher-order odd harmoaics. The analyaia
the grid curvature at comparatively low levels of has not been extended to these higber harmonics,
feedback (5% in the case of the 6V6) why does it and althoush individually they are of amplitudes.
not predominate and cause more than the observed approaching the experimental threshold of measure-
distonion as the screen feedback approaches 100% ment, it is by no means c:enain that they may DOl
(triode)? ha\'e been increased by the same proceaa which
~'lRaEsS WORLD. JANUARY 1956 25
reduced the mucb atroDlCf third harmonic. In
prac:tic:e, ;udgiDg from the aubjec:tive quality frcm
UL amplifiers we have heard, this effect, if present,
is negligibly small; but it would repay invcsti,atioa
(assuming that clistortiOD measurements of sufficient
precision are fonbcoming) if only to throw more
light on the fundamental processes of UL operation.
Ack"O'Wl~dgmrnt. Figs. 2, 3 and 4 arc based on Figs.
6. 2 and 5 respectively of Radiorronics (Australia). Vol.
20, No.5, May. 1955.
References
I A. D. BlumJein. British patent No. 496,883 (1937).
I .. An Ultra.linear Amplifier," by D. Hatler and
H. I. Keroesj Audio Encinl!mng, November 1951.
• .. Amplifiers and Superlatives," by D. T. N.
Williamson aDd P. J. Walker, Wir~l~ss World, Septem-
ber 1952.
• Correspondence: Graham Woodville, Wireless
World. November 1954j P. J. Walker, N. F. Butler,
Wireless World, December 1954.
• .. Ultra linear Amplifiers," by F. Langford-Smith
and A. R. Chesrennan, Radiotronics, May, June, July,
1955.
• Editorial (W.T.C.>, Wirl!ll!SJ Engineer, August 1955.
♦ Verso Filler Page ♦
 Amplifiers and Superlatives
An Examination of American Claims for Improving Linearity and Efficiency

By D. T. N. WILLIAMSON* and P. J. WALKER§

Recent articles, particularly in the United States, have shown that some confusion of
thought exists about the "goodness " of an amplifier for reproducing sound. This has
been accompanied by a debasement of the terms used to describe the properties of an
amplifier; words that have a precise meaning when standing alone become meaningless
when qualified by superlatives. This article, which starts with a. discussion of what con­
stitutes a good amplifier, attempts to clear up some misconceptions and generally to clarify
the position.

HE most commonly specified parameter of an

T
ponents, and the reduction of the number of such
amplifier is its harmonic distortion or inter­ components to a minimum. Changes in effective gain
modulation content, and this has tended to be during " low-frequency" transients occur in amplifiers
regarded as the primary standard of "goodness." In with output stages of the self-biased Class AB type,
1944, one of the writers suggested I that a standard of causing serious distortion which is not revealed by
0. 1 per cent total harmonic distortion at maximum steady-state measurements. The transient causes the
output was a good level of performance to aim at. current in the output stage to rise, and this is followed,
Such a low distortion content is readily achievable at a rate determined by the time-constant of the
by modern design methods, and is, in fact, undetect­ biasing network, by a rise in bias voltage which alters
able by listening tests, however refined. This the effective gain of the amplifier.
arbitrary figure was chosen, not because it was the (5) Low output resistance. This requirement is
maximum permissible, but because, being rather concerned with the attainment of good frequency and
lower than was absolutely necessary, a good margin transient response from the loudspeaker system by
is left for any deterioration of the equipment when ensuring that it has adequate electrical damping. The
in service. cone movement of a moving-coil loudspeaker is re­
The requirements for a good amplifier were listed stricted by air loading, suspension stiffness and
in the article already mentioned, but are worth repeat­ resistance, and electromagnetic damping. The effi­
ing and extending. These are : - ciency of a baffle-loaded speaker is rarely higher than
( 1) Negligible non-linearity distortion u p t o the 5-10 per cent, and the air loading, which determines
maximum rated output. (The term " non-linearity the radiation, is not high. Electromagnetic damping
distortion " includes the production of undesired is therefore important in controlling the motion of
harmonic frequencies and the intermodulation of the cone. This effect is proportional to the current
component frequencies of the sound wave. ) This which can be generated in the coil circuit, and is
requires that the dynamic output/input characteristic therefore proportional to the total resistance of the
be linear within close limits up to the maximum ex­ circuit. The output resistance of the amplifier there­
cursion of any waveform, with frequency components fore should, in general, be much lower than the coil
inside and outside the working range, which is likely impedance.
to be fed to the amplifier. This should hold good (6) Adequate power reserve. The realistic repro­
under conditions of varying load impedance such as duction of orchestral music in an average room
are likely to be encountered in practice. requires peak power capabilities of the order of 15-20
(2) Linear frequency response within the audible watts when the electro-acoustic transducer is a baffle­
frequency spectrum of 10-20k c/s loaded moving-coil loudspeaker system of normal
(3) Negligible phase shift within the audible range. efficiency. The use of horn-loaded or other efficient
Although the phase relationship between the com­ loudspeakers may reduce the power requirement to
ponent frequencies of a complex steady-state sound the region of 10 watts. In an amplifier for the best
does not appear to affect the audible quality of the possible quality it is, of course, assumed that output
sound, the same is not true of sounds of a transient peaks required in practice will be comfortably below
nature. the maximum available. Even so, good design should
(4) Good transient response. In addition to low ensure that the overload characteristic will not suffer
phase and frequency distortion, other factors which temporary paralysis or " blocking " on momentary
are essential for the accurate reproduction of transient overload.
waveforms are the elimination of changes in effective (7) The level of hum and noise should be at least
gain due to current and voltage cut-off in any stages, 80dB below maximum output.
the utmost care in the design of iron-cored com- This is a formidable specification, and by no means
every amplifier styled as " high quality " will meet it.
* Ferranti Research Laboratories. However, provided that these requirements are met
§ The Acoustical Manufacturing Company. in every respect and that extraneous components
I M.O. Valve Co., Ltd. Internal Report No. Q253, later published as
"Design for a High-Quality Amplifier," by D. T. N. Williamson, occurring within the audio range (produced either
Wirefess World, April-May, 1947: by components inside or outside that range) do not
WIRELESS WORLD SEPTEMBER 1952 357
 e�ceed a small fraction of 1 per cent, any amplifier 0.75.
wIll so�nd as �ood as �my other amplifier and it be­ 0.5
�omes Impossible to "Improve " a power amplifier
III the sense of producing better sound. 0.25

Efficiency.- On the other hand it is of course Oscillogram showing

_.-::�""-- o
transitio n of Ia- Va
�bsurd to say that . amplifiers cannot b'e improved' I
a curves from tetrode to
III a�y sense. Effic.lency, compactness, and reliability triode form a s the
are l!11portant and It is with regard to these factors tapping point in Fig. 1
tha� Improvemef,1ts are possible and in many cases is moved from B to A.
deSirable. The size and purpose of an amplifier will
determine how much weight should be given to each
of these factors.
In small amplifiers with power outputs below
�wenty watts, power efficiency is not usually of prime
Importance, and other considerations such as ease
of c0l!str�ction af,1d certainty o� results may easily put, as in the vast majority of cases even the present
outweigh It. . With large amplifiers, efficiency be­ l�vel . of output cannot be fully utilized. The reduc­
comes a very Important factor. Sometimes it is not tIOn m power consumption and ht. voltage would not
pos�ible to produce a high output unless the effiency greatly affect the cost, although it would give a higher
IS high, because of valve dissipation limits. factor . of safety for the capacitors in the circuit.
There IS, however, no evidence that this is at present
Controlled and Uncontrolled Production.- The madequate.
desi�er ' s aim is (or should be) to produce the best
Circuits with Distributed L oads -Articles 2,3 have .
possible sound for those who will ultimately make use
o� his efforts. If he designs for construction not under recently �ee.n published in the United States claiming
hiS control, . h� must, as far as possible, ensure that the supenonty of a so-called " ultra-linear" output cir­
every a�pltfler made will meet his performance cuit in which the output valves are used as tetrodes
figures wI�ho.ut undue di�ficulty and with the employ­ with. negative feedback applied non-linearly by con�
ment of Itmlted measunng equipment. In such a nectmg the screens to a tap on the primary of the
case, the av?idanc� of cir�u!try which is not straight­ ?utpu� tran.sformer. It is stated that the performance
forward, or III which deViatIOns from specification in IS audibly Improved over that of triodes with similar
the values and construction of components is liable degrees of negative feedback.
to produce poor results, is paramount. The present writers do not believe this claim. The
The designer who has control over the production cir�uitry which forms the basis of these American
of the complete equipment has a different set of pro­ ?lalms for "ultra-linearity " and higher efficiency has
?l�ms: Re has much greater freedom of choice, and m fact, been familiar in this country for several years '
It IS hkely th�t. he wil1 arrive at different circuitry. and t�e technique �as been further developed and'
The amplifier which was described in Wireless us.ed m a comm�rclally produced high-quality am­
rr:or1d, April�May, 1947, is an example of the first phfier. 4,5 It consists of the distribution of the load
kllld ?f clrcUltry. The success it has achieved since imped�nce between the electrodes of each output
then, III the hands of people with widely varying de­ valve In order to obtain the optimum performance
grees .of skil!, is ind�cative that trouble-free perform­ from that valve.
ance IS re<l:dIly o�talllable with only simple adjust­ In its sim�les� form, and as used by Rafler and
ments. This ampltfier was based upon a triode out­ Keroes, the ClrcUlt arrangement is as shown in Fig 1
put stage with .a good �)Utput transformer, the per­ The circuit is normal, except that the screen of 'th�
formance of which was Improved by overall negative te!rode c�n .be tapped on the output transformer
pn�ary wIndmg, . thus coupling it to the anode in any
feedback. rattO, and rendenng a section of the load impedance
The only valid criticism which can be made about c0!11mon t<? .both electrodes. If the tap is affixed at
its performance is �n the score of efficiency. The out­ POlllt A, glVlllg. a co.upling factor of unity, the stage
put of about 15 W IS produced for an h.t. consumption behaves as a tnode, ItS performance being determined
of 56 W, an efficiency of only 27 per cent. By the
use of tetrodes as output valves this could be raised by the dynamic characteristic of the screen and con­
to the order of 35-40 per cent, which means that, for trol grid, and if the tap is shifted to point B the
c�)Upling factor is zero, and the stage behaves' as a
!he same power consumption, the output could be Simple tetrode. If now the screen is tapped at in­
Illcreased to 22 W, or, alternatively, that the power
consumption could be tervals b.etwc:en pO.int B and point A, there will be a
---+H . T .
reduced by about 20 W. progressive InclUSIOn of the load impedance in the
Whether this is worth screen circuit and a progressive change from tetrode
doing or not is problem­ characteristics at B, to triode characteristics at A
It is c�nvenient . to consider the stage as being a
atical. In the designer ' s tetr�de With negative feedback applied to the screen
opinion, the additional It differs from <l: tetrode . stage with feedback applied
risk of trouble in un­ to the control gnd, first In that the screen introduces
skilled hands outweighs a non-linear element in the feedback loop, and
the advantage of higher secondly, because as the degree of feedback is in-
maximum power out-
2 Hailer and Keroes: "An Ultra-linear Amplifier Audio Engineering
November, 1951, pp. 15- 17.
3 Hafler and Keroes " Ultra-linear Operation of tile Williamson Amplifier" .
Audio Engineering, June 1952, pp. 26-27.
4 First IlltroQuced III 1945; see advertisement of tile Acoustical Manufacturing
Fig. 1. Basic circuit used by Comp�ny in Wirless World , July 1945, P. 10.
Hafler and Keroes. 5 MOlf' , ReView of Bntlsh Amplifiers" , FM TV, October 1951 pp. 30-32.

358 WIRELESS WORLD SEPTEMBER 1952

creased, the permissible anode swing is decreased, due input resistance is increased, thus facilitating the de­
to anode current cut-off on the negative excursions of sign of the previous stage and reducing the value of
screen voltage, until finally the Ia-Va curves become the coupling capacitance necessary.
those of a triode. A more complete account of the The use of all the electrodes in this way gives addi­
behaviour of the circuit is given in the Appendix. tional flexibility in design, so that the parameters of
The curves of Fig. 2, which apply to a single valve, the output stage may be varied to suit the penultimate
show the effect of changing the tapping point. The stage -a technique giving greater scope in arriving
output at the point of overload, the relative gain and at a well-balanced ratio of stage distortions and hence
the output resistance decrease as the degree of feed­ an optimum final design. In .the outP.ut stage of the
back is increased, the output resistance decreasing Acoustical "Q.D.A.D." amplIfier, usmg two KT66
very rapidly because it is dependent, not on the gain valves, the values are so chosen that an output of 12
reduction, but on the amplification factor, which
undergoes a large change. SCREEN/ANODE LOAD IMPEDANCE RATIO

The distortion at constant output, which is pre­ 0'1 0 '2 5 I)-S 0·75
15,000 3 6
dominantly second harmonic, rises slowly at first until '\
the point of overload is reached. Since the maximum I--
power output is progressively reducing, the distortion r--
1"- r--..

�
curves for maximum output or a percentage of maxi­ "
mum output will show a progressive reduction from 10,000 2 4 1"-.. �
r-.,.-9c(
'/'
tetrode to triode operation. 'l � ,
In view of the increase in distortion, as shown in v \ DISTORTION
AT5W
f�� V
Fig.2, it would appear that there is little to be gained �� i:::.� 3
IDISTORTIOtt
rSI �1 -r-
by the use of this circuitry, since a simple tetrode, 5,000 2V u

with feedback to the control grid, would give a similar �l/rpl/) 2

z
o
:>:
reduction of output resistance, and this would be �fS/sr�lI�
-r;.:::,:f
'"
«
::<:
accompanied by a progressive reduction in distortion. DISTORTION AT I
0'25"11 I -'
«
This is indeed true in the case of a single valve. .­
o
o 0 0 0 >-
When push-pull connection is considered we obtain 0'25 0·5 0'75

a different state of affairs, because of the cancellation RATIO

COMMON WINDING <x)
TOTAL WINDI NG
of second harmonic components, and we arrive at the
curves shown in Fig. 3. The general shape of the
curves is maintained under varying loads. With up Fig. 2. Curves of output power, relative gain, output
to 30 per cent of the winding common to screen and resistonce and distortion for the circuit of Fig. I. The
anode the reduction in distortion is greater than the distortion combonents are predominontly second ond even
reduction in gain. This " something for nothing " is harmonics.
small, however, and can be lost or even reversed if
there are appreciable departures from precise balance
at any frequency. SCREEN/ANODE LOAD IMPEDANCE RAT 10
0'1 1)-5 0'75
It is, of course, not essential to have the screen r-..
>­
tapped directly on the primary winding, and an z
w

auxiliary winding tightly coupled to the primary u ,

'"
w I
(ideally in the form of a bifilar winding) may be used. 2
�
Z
This has the advantage that the screen supply voltage o
\ / I
>­
may differ from that of the anode to give optimum '" "- V /
o 12"11
operating conditions. >­
r-
Vl /
o

u ,� sW
,/
The" Acoustical" Circuit.-A further development
z
o --
f-r-
of this circuit (" super-ultra-Iinear," perhaps?) used :>: -- 4'W --
'"
in the Acoustical " Q.D.A.D." amplifier and its pre­ "'"
z 0"5"11. -
decessor, takes the common portion of the winding -'
«
>-
and inserts it in the cathode, giving the circuit of o
.... 0 025 0'5 0'75 1·0
Fig. 4. As far as the anode and screen circuits are w
o w

concerned, this arrangement is identical to that of o

'" RATIO
COMMON WIND�NG (x) o
o
>­ TOTAL WINDING a:
w t-
Fig. 1 (neglecting winding resistance and leakage reac­ >-

tance). The only difference lies in the grid circuit,

since that portion of the voltage appearing across the
Above: Fig. 3. Curves of
common winding is now applied in the grid circuit as �---�---+ H.T.
distortion for a push-pull
overall negative feedback. It is, of course, feedback arrangement (predomi­
of the most desirable kind, since it is the most prac­ nantly third and odd
tical method of applying voltage feedback over a single harmonics).
stage without either wastefully throwing away gain
or increasing the load on the previous valve.
Since the screen is now connected to a decoupled
point, it follows that the designer is free to use dif­
ferent voltages for screen and anode if required, with­
out the necessity of additional windings, with their
attendant disadvantages Further, the arrangement Right: Fig. 4. BaSic cir­
makes it possible to provide the optimum degree of cuit used in the Acoustical
smoohing for the anode and screen supplies. The Manufacturing Co m pany ' s
grid resistor may be taken to the cathode end of the "Q.U.A.D." amplifier and
common winding if desired, with the result that the its predecessor.

WIRELESS WORLD SEPTEMBER 1952 359

watts is obtained from a 320-V supply with a grid-to­ of triodes and distributed-load tetrodes (especially the
grid input of 72 volts peak, the input resistance being cathode-coupled variety), with the exception of effici­
1 mehogm. The total harmonic distortion of the out­ ency, in which respect the tetrode circuits are superior
put stage alone is not more than 0.7 per cent, and the to the triode.
output resistance is approximately half the load resist­ In order to avoid misapprehension, it should be
ance. stated that a similar order of performance is obtain­
By comparison with the same valves used as triodes able from a conventional tetrode circuit, by the appli­
to obtain-the same output, the following advantages cation of the appropriate degree of negative feedback,
are apparent: preferably in the form of multiple loops to ease the
( 1) Distortion is less than half that of triode connec­ stability problems. The advantage of the distributed­
tion, due to the cathode and screen feedback. load circuits is that, as a considerable amount of nega­
(2) Efficiency is increased from 27 per cent to 36 tive feedback is included in the output stage itself,
per cent, resulting in the h.t. voltage being lower by the design of the remainder of the amplifier is simpli­
about 100 V, thus simplifying reliability problems both fied and the problems of stability and restriction of
in the amplifier itself and throughout the range of scope in design usually associated with large amounts
pre-amplifiers, tuners, etc. , which may take their of overall negative feedback are avoided.
supply from it. Practical Difficulties.- So far we have only been
(3) Less smoothing is necessary for equivalent hum­ considering the ideal case. In a practical transformer,
levels. however, the windings are not perfectly coupled, but
Additional overall negative feedback can be applied are more loosely coupled by a complex network of
to the complete amplifier and with the ratios used leakage reactances formed by the distribution of leak­
in the above example, 8 db less feedback is required age inductance and self-capacitance throughout the
for a given level of distortion than with triodes. This windings. This departure from the ideal may mean
assists in maintaining a good margin of stability and that, at high frequencies, the circuit is not at all as it
reduces the effect of overloading due to the presence would appear on paper, and the effective sense of the
of frequencies in the input which are outside the coupling may even be reversed, producing oscillation.
effective feedback range. In a less severe case, peaks and troughs in the fre­
There remains the question of output resistance to quency response characteristic may occur, accom­
be considered, about which there appears to be some panied by " ringing " and instability when the
confusion. This is probably due to the general use, amplifier is excited by a transient.
as a measure of the efficacy of damping, of the load These defects are serious, and can only be avoided
resistance/output resistance ratio, sometimes called by designing the output transformer carefully and by
the damping factor, the scale of which becomes vir­ maintaining close control over its production, as even
tually meaningless at high values. It appears to be with transformers of the same nominal specification,
common practice to aim at as high a value as possible, wide variations in performance at high frequencies
with presumably infinity (zero output resistance) as may occur due to minor variations in the quality and
the ideal. An output resistance of zero is, of course, thickness of insulants, and in the positioning of wind­
only an arbitrary figure, which has ultimately to be mgs.
added to the speech coil resistance. The designer who has complete control over
Independently of the output-stage circuitry, zero­ production can arrange that the materials and con­
or any reasonable value positive or negative-can be struction of the transformer do not depart from
obtained by a mixture of negative voltage and positive specification, and even if they should do so he has
current feedback. It should be pointed out, however, facilities to detect and correct the deviation at an early
that the optimum value is dependent upon the loud­ stage. He is therefore able to make use of circuitry
speaker and particularly the intended performance of of the type discussed, without the dangers which
its enclosure, so that the doctrine " the more damping would attend its use in unskilled hands or where
the better," is not always sound. measurement facilities are absent. The advantages
Comparison of Output Circuits-To summarise, obtained have already been discussed.
Table I gives a comparison of the relative merits and The constructor with limited facilities cannot be
demerits of various output circuits. It will be seen too strongly advised to keep to proven circuits which

TABLE 1

Parameter Triode- Tetrodes Tetrodes with Tetrodes with

connected (ClassA ) load distributed load distributed
1
tetrodes between anode amoung anode,
(Class AI) and screen screen and cathode
(optimum value) (Q.U.A.D. arrangement)

Efficiency (per cent) 27 38 36 36

Relative power output 1 1.4 1.35 1.35

Relative distortion just below

the onset of grid current 1 2 1.5 0.5

Load resistance 2-4 0.05 -0.1 0.5 -1 2

Output resistance

360 WIRELESS WORLD SEPTEMBER 1952

keep to designs requIrIng the mInImum number of Output resistance.
coupled circuits in the output transformer, since the RL
Let RI. � 00, then 1
possibility of pitfalls is greatest in this component and R a + R I. �

Then gain (x 0) J.L

---
increases rapidly with the number of windings when = =

all these must be closely coupled. J.L

It will be appreciated from the foregoing that there Gain (x) =J.L
are a large number of solutions to the problem of 1 +x�
m
designing a first-class amplifier, and no one of these
solutions can be called the best solution. Each has Output resistance (x = 0) = £....
gm
its advantages and disadvantages, and the individual
designer must choose that which most nearly meets J.L
Output resistance = J.L' =
his needs. The "goodness" of an amplifier is not gm g m(l + X �
shown by its circuit diagram. Circuits have no in­ )
herent magic properties, but are merely the tools with
which the designer seeks to achieve a certain result,
and different designers-provided always that they
have the same high standards in view-may achieve "Q.U.A.D." case

the same result by different means.

APPENDIX Gain is given by
RL
Let
Ea E. A
J.L=p:=gmRa,m=E
g g

•
E" J.L
.. E. =m
X fraction of output fcd to
= Gain with feedback =

RL,
screen.
E = output voltage in load 1 + xA
produced by grid
voltage e.
Neglecting the effect of screen
current in the load, which is
normally less than 10 per cent,
Current in load due to e J.L Ra + RL
J.L
J.L 1 RL
=e
Ra + RL + x ID Ra -I- RL
Gain
Current in load due to xE = _ xE "
meR. + RL)

.
: Vo ltage in load = E ={e - J.L
Ra+ RL
xE
m
( J.L
R + RL)
a
}RL 1 + x

:. e J.L:La
= E ( 1 + X J.LRL)
mR a
Output resistance
Output resistance
without feedback
with feedback
1 + XJ.L
= _E_ = R a --'+_R=L'---_
__---=-
Gain
e 1 + .£ RL
X
mRa + RL

When X = 0, Output resistance =

J.L
E 1 +- x

e l+x£
m

WIRELESS WORLD SEPTEMBER 1952 36 1

♦ Verso Filler Page ♦
 Determining Screen Grid Dissipation
In "Ultra-Linear" Amplifiers
A difficult parameter to measure, screen dissipation yields to this method. Designers should
know what the dissipation is if amptiers are_ to operate satisfactorily over long periods.

LEONARD KAPLAN°

A N IMPORTA.NT OONSIDEBA.TION in the

design and adjustment of ampli-
fiers using beam power tubes in
eapacitor and high-impedance choke
was suggested by F. Langford-Smith in
Radiotro"icB for July, 1955. It is much
screen-grid (grid No.2) feedback cir- simpler, however, to use the arrange-
cuits ("Ultra-Linear" amplifiers) is ment shown in Fig. 3.
srreen grid 1iissipation under maximum- In this arrangement, the d.c. screen
signal conditions. In ampli1lers of this grid voltage is assumed to be equal to
type (see Fig. 1), grid No.2 of each Ebb, and measured with the d.c. volt-
tube receives a fraction of the a.e. plate meter M" The d.e. screen grid current I
voltage of the tube, as well as a d.e. is measured with the d.c. milliammeter
voltage, and contributes&. portion-of the M". and the rms- value of the combined
power output. As shown in Fig. 2 the a.c. and d.c. screen-grid currents t, is
screen grid dissipation varies with the measured with the thermocouple milli-
signal amplitude, and is always less than ammeter M,. The power-output meter
the product of the d.c. screen grid volt- is used to measure the maximum-signal
age and current. If this dissipation is power output Po delivered by the am-
Fig. 2. Typical screen grid voltage, cur-
less than the permissible screen grid in- rent and power waveforms in a screen- plifier into its rated load. resistance R L-
put for the tube type used, it may be grid feedback amplifier.
possible to obtain higher power output Procedure
from the amplifier by an increase in the proach is to apply a sine-wave signal
of maximum amplitude to the circuit, The circuit is adjusted to provide the
d.e. supply voltage Ebb' If this dissipa-
tion exceeds the permissible screen grid measure the a.c. and d.e. components desired operating conditions, and the
input, it will be necessary either to re- of the resulting screen grid voltage and maximum-signal values of POI Eo", and
current, and then determine the dissipa- It determined from the indications of
duce E"", or to increase the fraction of the corresponding meters. The rms a.c.
the a.c. plate voltage applied to grid tion from the relationship
screen-grid voltage 6 may then be detel'-
No.2 to assure that the permissible in- P=EI -ei (1)
put is not exceeded. mined from
where P is the grid No. 2 dissipation,
E and I are the d.c. sereen-grid volt-
Determination of Screen Grid Dissipation e=!..N, ~ (2)
age and current, and , and i are the rms 2Np'J - .
Although the screen grid dissipation values of the a.c. screen-grid voltage
in an "Ultra-Linear" ampli1ler circuit and the rIDS a.c. screen-grid current i
and current.
can be calculated by an involved integra- A method for separatmg the a.c. and from
(3)
tion procedure, a much simpler ap- d.e. components of the saeen grid volt- i=yI,' -I'
age and current by-the use of a blocking
• Bw"frOfl Tub6 Diviaioft, Badio Corpo-
raHOfl of .f1Mrica, HarrisOfl, New Jer'e1l.

OUTPUT
TlANSfORMEI
,.....--..,
BALANCEO
SINUSOIDAL
INPUT
POWl:R
-4-i1~~~-------+
OUTPUT
METEl

___ ....-+-+-1£.
N$ • GilD NO. 2 TUINS } ONE-HALf
. IlIMAIY
N, • TOTAL T\JIN$ WINDING

c-a- MI

Fig. 1. Simplified circuit of amplifier

uslna screen grid feedback. Fig. 3. Setup used to determine screen grid dissipation.
The screen grid dissipation P may then
be determined by substitution of the
values for E, I, 6, and. in Eq. (1), or
directly from the following equation:
N, - - ---
P=EI--yRLPo yl,'-I'
2N,
It is essential that M, and M, be cali-
brated accurately 8fainst. each other (in
a series circuit) because any di1ferences
in their indications will be magnified by
the squaring operation in Eq. (3).
The method deserlbed can be used to
determine the screen-grid dissipation of
any beam power tube or power pentode
in a screen-grid grid No. 2 feedback
circuit. JIB
 Ultra-Linear Operation of the
Williamson A m p lifie r
DAVID HAFLER':: and HERBERT I. KEROES:::

The Famous "Williamson" can be improved simply by replacing the output transformer and making a
few minor changes in other components. The results are well worth the effort and expense.

RO:\I Tllr. TIME an article I appeared

F descrihing the principle of Ultra­

Linear operation of power output
tuhes, considerable interest has been
evidenced in rhe application of this
new circuit improvement to the famous
WiIliamson amplifier. The William­
son circuit ha$ hcen publicized in
several arrangenlents including at least
one commercial one, and the configur­
atiou is undoubtedly the most popular The �uthors' Ultr�·
high-quality audio circuit ever devel­ Line" �mplifier com­
oped. For Illany people there is little bined with the power
necessity to attempt to improve this supply on a single
chassis.
basic amplifier circuit. Its listening
quality is excellent; it is easy to COil­
strtlct; and it provi(les top quality at
a cost comparahle wirh units which
cannot measure up to its capahili ties.
The one category in which the \Vil­
liamson amplifier is significantly de­
.ficient is with regard to e ffi ci ency and
power-output capabilities. Peak power
output is less than 15 watts, and it
takes a 450-vo I t supply at approxi­
mately 130 ma to achieve this power screen grids of tetrodes are energized TO-300, which was designed for use
output. It this limitation can he over­ from a tap on the primary of the out­ with tubes of the 6L6 type. Its 6600
come without deterioration of quality , put transiormer. This connection, on ohms primary impedance therefore, is
a change in the original design is justi­ which patents are pending, modifies the also correct for 51:\81 's and 807's in the
fied. If simultaneo n sly it is possible to operating characteristics of t.he tube. Ultra-Linear hook-up. In alhlition, KT-
improve the amplifier hoth in measure­ Proper location of the tap results ill 66's can be Ilsed without deterioration
able aspects and in listening quality, optimum input-output linearity simul­ of quality as the slight mismatch is in
then a change is not only justifiable, it taneously with ellicient operation, a favorahle direction with respect to
is mandatory. power capahilities approximatdy douhle distortion characteristics. Thereiore,
It is difficnlt to impr"ove on some­ those of a triode connection, alld low­ this transiormer can be used with the
thing which is really good. There .re impedance output such as is offered by tube types normally used in \\'illiam­
some audio enthusiasts who will scoff triodes. In short, it �>crmits better per­ son amplifiers w i thou t compromise of
at lhe illea that the \Villiamson circuit. fonuance than either triode or tetrode characteristics. It is of interest to note
can be improved. However, it has heen connection oi the tubes, and this is sub­ that the change in impedance to 6600
fi,'e years since l\[r. Williamson pub­ stantiated in comparative listening does not violate �[r. Williamson's de­
lished his circuit; and in the course of tests an(1 distortion measurements. sign considerations. The modi fied tube
five years, there is little which can The unique merits of the Ultra-Linear characteristics of the Ultra-Linear con­
maintain supremacy without change or stage are particularly applicahle to the nection require this impedance if we
renovation. \Vhen a basic circuit im­ Williamson circuit. The mating of the wish to preserve operating conditions
provement-the Ultra-Linear output two seems to have been ine,·itable. The similar to those of the original amplifier.
stage arrangell1ellt-<ame along, it was simple suhstitution of an 'output trans­ I n other words, the tuhes are still
natural to see how it could fit in with former with primary taps for Ultra­ matched for minimum distortion rather
the basic Williamsoll circuit. J .inear operation and a few minor than for maximum power output. The
The Ultra-Linear output stage is 110t changes in circuitry, which will be transiormer, therefore, can he placed in
a triode stage as is used with the 'ViI­ discussed he low, combined the basic the circuit directly and the screens of
liamson circuit-nor is it a tetrode or circuits into an amplifier which prac­ the output tubes connected to t he ap­
pentode srage. It combines the ad­ tically cveryhody agrees is an improved propriate taps as shown in Fig. 1. This
vantages of both triolle and tetrode hy version in all respects. Obviously, we eliminates the two lOO-ohm screen
using an arrangement in which the must gain improvement if we substitute stopper resistors of the original circuit.
a more linear outpnt tuhe and use a The plate and screen leads of the trans­
* Aero Products Co., 369 S/lIIrs LOIII!,
Philadelphia 28, PI!ll 110. transformer which exceeds the origin­ former are color coded to avoid phasing
1 Hailer and Keroes, "An ultra-linear ator's stipulations for performance. di fficttlt:ies.
amplifier," AUDIO ENGINEERING, November The original Ultra-Linear circuit Several additional circuit changes
1951. utilizes a transformer, the Acrosound have hecn found heneficial for optimum

26 AUDIO ENGINEERING • JUNE, 1952

 -'1---,-0 .

\10.'1---.0 •

ICI-----'O 4

c
",..+-1---:-91
=+---1-

acllO
To-�
'.TO�� __------�-------- t----- -I��
__TQG
'AlII

__ to-tO �
500 w"
11711

Fie. 1. Schematic of the Ultra-Lintar Williamson amplifier. The components in dotted circles life those which are changed from the origin,l circuit
i" making the conversion of an ellisti�1 amplifiir to Ultra-Linear oper.ltion.

performance. One of these is the all taps of a t apped - secondary winding. creasing one pair of coupling capacitors
change in value of the feedback resistor The amplifier, as converted, now from .05 Ilf to .25 Ilf gives a five-to-one
to 10,000 ohms in order to maintain 20 surpasses the original with respect to ratio of time constants for the two
db of feedback. In the Ultra-Linear response, distortion, and transient char­ pairs of networks and inc rea ses the
stage the gain of the stage is greater acteristics. In addition, it was con­ low-frequency stability margin at nom­
than for a tr i ode stage. In addition, the sidered desirable to make certain other inal increase in cost.
change in primary impedance changes slight changes which primarily increase The insertion of a lO,OOO-ohm para­
the proportion of voltage fed back. the stability under feedback conditions. sitic suppressor- in the input grid and
Thus the feedback is increased unless The low-frequency time constants of a 100-1111£ capacitor across the feedback
the feedback resistor is changed to the original circuit's interstage coupling resistor adds to the high-frequency
compensate. The readjustment of this networks were the same for both such stability margin and eliminates a slight­
resistor to the desired value then per­ networks. This is not particularly de­ ringing in the vicinity of 200 kc.
mits the added gain of the Ultra-Linear sirab le in a feedback amplifier since a One last optional difference from Mr.
output stage to increase the amplifier given frequency loss is accompanied by Williamson's or iginal circuit lies in the
sensitivity. It can now be driven with maximum phase shift. Separation of use of a bypass capacitor across the
a little over 1 volt as compared to almost the time constants permits less phase cathodes of the o u tput stage. This has
2 volts required for the original amp­ shift for the_ sam e frequency loss. In- been found beneficial in both the Ultra
lifier.
The feedback is taken from the 16-
oh m tap regardless of the speaker con­
nection. This tapped secondary arrange­ - -
ment is extremely convenient when Fig. 3. Squue-w�ve
shifting to speakers of different im­ performance �t 20
pedance as it does not require a change cps lIeft) and at 50
- -
in the value of the feedback resistor. It kc (right!.
is made possible -by special transformer
design (on which patents are pending)
which permits equivalent response on

I Linear conversion and in the triode

::f.'tt -: �'4I' I Williamson2 at high levels of operation
I as distortion at the overload point is
Fig. 2. Curn of in­ diminished.
te""odul.tion distor­ There are no changes required in
/ tion n. .Htctin the remaining stages nor in the power
/ sine-wave-power out- supply. Most of the publicized versions
put.
�
�
: Kerocs, H. I. "Building the William­
o :so sonamplifier.l" Radio and TelevisioK News,
December l �SO.

AUDIO ENGINEERING • JUNE, 1952 27

oi the circuit utilize power transformers on a 'scope. By observing the transfer
which iurnish �OO
volts at 200 ma. characteristic, it is possible to detect hy
Since the draio of the circuit does not eye harmonic distortion of less than 2
exce�d 130 to I�O ma, the voltage oh­ per cent. The power curve of the amp­
tained out of a capacitor input filter and lifier thus de"iates frol1l tlat by less than
5 V� recti fier is ahout �50 volts. This is 1 dh over the range 20 cps to 30 kc.
\s intimated above, the circuit has
the correct value for the circuit as cou­
excellent listening qualities. This is a
,·erted. Lower voltage will limit the
confirmation of the measurements. The
power output capahil it ies.
additional power available shows up in
cleaner and better articulated bass. The
Performance
overall effect is of greater smoothness,
Fig"re 2 shows iutennodulation dis­ more definition of detail in the soune!.
tortion versus power OIIlPIlt. I t can he and better transient response. Ultra�
seen that the power out)lut oi the cir­ Linear circuits seem to have a wider
cuit is effectivelv dOllhle,1 over that oi transient bandwidth-all audible benefit
the original circu' it for a gi\'en distor­ which is not readily susceptible to
tion. At low levels, around I watt, the lIIeasurement. The combined effect of
1:\1 hit� snch phenolllenal \"alues as the Williamson circuit configuration­
.06 per cent. It is only 0.3 per cent at a wide-band, low-distortion arrange­
13 watts. This curve is hased on eClui v­ ment-plus an output stage of decreased
alcnt sine wa ve power in order to make distort ion and higher power capability,
it comparahle with all the other pllh­ a stage which exceeds the original
lisht'd and advertised (lata on the Wil­ specification and operating parameters,
liamson circuit.J The \'allles graphed lIIust be heard to bc appreciated.
in Fig. 2 can he divided by 1.47 for
those wllo wish to have direct cOlllpara­
hility with the lIIeter readings ohtained
on the illlermodlllation test eqllipment.
Fig"re 3 shows oscillograms of
square-wave traces taken throllgh the
COml)lete amplifier with repetition ratcs
of 10 cps :11\(1 50 kc. Traces at mter­
mediate irt'qllencies appro.1ch theoreti·
cal periection. and t'\'en such a rigorolls
test as the 50-kc wave shows up ex­
trelllely well. The waveform has not
"sined off," ami the extent of ringing- is
It'ss than that exhihited hv the SO()()·
cps wave of many good < j nality alllp·
litiers. These square-wave tests were
made at a comparatively low level
which makes the fest even more rig­
orolls. At low excitation levels, the in­
ductance of an. Olltput transformer de­
creases, the phase shift increases and
the tOps of the square wave tilts. J\ high­
level square wave will appear hetter
than a low -level one at low i reqll('ncies.
Similarly, high powers at high fre­
quencies will clip any sllpersonic peaks
in the response and improve the ap­
pearance of the square wave. The use
of a high level of power can make a
relatively poor amplitier appear Ioetter
on square wave tests.
The frequency response of the con­
verte(l amplifier is flat ± 1 dh from less
than 5 CI)S to 200 kc. Its phase shiit
reaches 3 deg. at 20 cps and at 10 kc,
indicating symmetry of response with
resl)t'ct to the audio hand.
The aml,litier puts out 30 watts of
power o,'er a range greater than the
audio spectrum. IIo\\'ever, this type of
power curve, as measllred hy response
at high power levels, is not too meaning­
ful. The important consideration is the
amolint of Ill/distorted power available
at variolls freq\lencies. The Ultra Linear
Williamson arrangement Jluts out close
to 25 watts at 20 cps and at 30 kc with­
out clipping, attenuation, or other visual
distortion of the waveform as viewed

3 Sarser and Sprinkle, "Musician's

amplifier," AUDIO E:'lGINEERl:>G, No\"clllhcr
19 ..9.

AUDIO ENGINEERING • JUNE, 1952 43

♦ Verso Filler Page ♦
 High-fidelity unit featllres low-distortion
preamp and choice of speaker damping

Designing an Ultra-Linear Amplifier

Dy EDWARD S. MILLER·

OME recent trends in high-fidelity

S amplifier design have been in the

direction of smaller, more compact
assemblies that can be made into
attractive, self-cabineted units. This
type of design can be developed with­
out sacrificing performance in any
way; in fact-with new tube designs
and advanced circuitry-improved per­
formance has often resulted.
Our own Sherwood S-1000 (see
photo) is an excellent example. Its
features had to include -20-watt output
and single cabinet design. These were
just the two of the most important
factors governing its design. Others
were that it must have low heat dis­
sipation and be small. Within the
limits of these specifications it was
possible to include a choice of speaker
da:nping (including negative damp­
ing) ; low-distortion tone controls;
rumble and scratch filters; a low-noise,
low-distort'ion phono preamplifier and
a pushbutton-operated record equalizer Fi&'. I-Some members of the 6L6 family_
circuit.
The new 6 L6-GB tubes (Fig. 1) were where % is the percentage of turns In calculating a push-pull class-A
used, not only because they are smaller that the screen grid is tapped up from amplifier's maximum power output and
than their predecessors the 6L6-G and the transformer-primary center tap. optimum load line, it is necessary to
6L6-GA, but because of other construc­ Substituting % = 100 (triode opera­ know the exact zero g·iid characteristic.
tional advantages such as the more re­ tion), this formula becomes: Therefore, to eliminate confusion, only
liable button type base. To economize on E .. = Eh - E. + E, = E, the general direction of the other grid
B plus requirements, only tetrode and And if % = 0 (tetrode operation), the voltage curves are shown. Note' tha;('
tapped screen (Ultra-Linear) opera­ formula becomes: the curves for different degrees/"�f
tion of the output tubes was considered. E .. = E. screen tapping pass through common
Restricting the B plus- requirements to Values of E .. are shown below the points on the constant E. coordinate.
conform with the 125-ma maximum curves (Fig. 2) for each plate voltage Several interesting features of Ultra­
limitations of the 5Y3-GT not only . value. B supply voltage of 320 was Linear operation can be observed from
made it possible to use the smaller used in the final Sherwood design. the curves in Fig. 2. First, maximum
rectifier, but also saved the 5 watts of However, to simplify calculations, a push-pull class-A power output was
additional filament drain that would voltage of 300 was used in these tests. calculated using each of the zero grid
have been required for the larger 5U4-
G frequently used in such applications 2S
�

as this. °le\-£,£ (H.\AIo(!�1£l: (BB

ri3: I'IH
;
T��\UltEN' T(TlIOO(

Output tube operation was based on -+-- �LfSS n.t!Ctt�'u

an analysis of the typical plate char­ R
IOAXPWR
200I--s!Cu: ,M;''':
t"
() -
acteristics of the 6L6 tube family. Fig. , . �� /
2 shows these characteristics for triode
, --- L/ , ,
1/ Iloo- Fig. 2_
II/!III .t;oj.�7 -IOV f-
and tetrode operation. These are static I� Plate characteristics for triode,
1/ -
curves, taken with a laboratory power �
�-
�� i/
� III ..,
Ii- Ultra-linear and tetrode operation,
supply which varies the plate and I, -!lA

+
FJ l���
-f�� ',- - -7 '+V
- 7 Ir.. vLES� � -J
ISV f-

",.*r: x",""
screen voltages independently.
"

Also 101)
shown are four intermediate degrees
of tapped-screen (Ultra-Linear) opera­
f- f/ 7�JW � , J.. 7 - -lflY
'.L[!"" �L(SS -.I'..
�1--
-F- '-
---'-- J i.. -- �
� F= 2W 1-- -
_

tion.'- I, , � '--f-- ./ " .=:

,
In determining tapped-screen plate 1--- "-
7
XN --
/
-
.-
characteristics, the, screen voltage used -1�V f--
/
-

- ¥N -f-
in each case is a percentage of the
0
difference between the plate voltage 50 lOO 110 ZOO Z50 lOO 150

and the B supply or: lOO lOO lOO lOO lOO lOO 100 lOO lOO mAOO( SCA't'OlTAG(
2OX� '
•
250 260 ZIO Z90 100 110 120
E •• = E. - o/r (Eh - E,) 240
110 200 220
ZIO
Z40 ZIO 210 lOO 12O )40 -lOll'
120 ISO 180 210 Z40 Z10 lOO 1)0 160 eo.'
•""n"ral mana&er and chief en&lneer. Sher­ 60 100 140 110 ZZO Z60 lOO l40 110 eo..
wood Electronics Labor.torie�. o �o 100 110 ZOO 2S0 lOO ISO 400 TRIOOE

92 RADIO-EL
 X-ray view of the Z729_ Shield screen
around plate m inimi z ea hum and noise_

characteristics shown. This power is

approximately:
P. = � (Eo. - En... ) In, ..
where EM" and In, . . describe a point
either at the knee of a tetrode type
curve or at the intersection of a line
from E. which has the negative slope Left,-The Sherwood model
of a ,triode type curve. These points 8-1000 20 - w a tt amplifier.
have all been indicated in Fig. 2 and Below-Internal Tie w of
have been connected by a dotted line. the compact a mplifier.
Two similar lines have been drawn

• RIIS, 2O-A"S MPUT,1OI« COIITIS Ia. FLAT,

LOUOHfSS' UVEl CIlNTAOlS AT WAI

GNO

LPQ!!t!l2'!'�!l _______________________________ ,
(()ITAOL "'ASSIS PUSH BUTTON!
AfS
vzl2AX7 RIA. EuR

�.oo )9
M)() seRAT'" k-i Z72
..g_' __
.,

��'
L
: !--�
_
__ -..J

IOK 8+

rAl'{ �OJTM A 8 (TAl'{ M)()

Fig. 3-Schematic diagram of the power and control chassis of the 20-watt high-fidelity amplifier.

94
 T he "goodness" of an amplifier is not shown by its circuit dia-

gram. Circuits have no inherent magic properties, but are

merely the tools with which the designer seeks to achieve a cer­

tain result, and diffe,.ent designers-provided always that they

have the same high standards in view-may achieve the same re­

sults by different means.

-D. T. N. Williamson (1)

Hi g h Q u a lit y C ire u its

-

By JOHN K. FRIEBORN

INCE
S
Williamson published the
first description of his· "High­
Observations on Ultra-Linear, Quality Audio Amplifier"2, in
1947, other audio amplifier de­
plus circuit features of three signers have had two apparent choices,
"beating him or joining him." A popu­
lar compromise is to adopt his general
high-quality power amplifiers circuit arrangement, but to replace the
class-A triode output stage with another
type giving higher efficiency. The
Ultra-Linear version of the Williamson
amplifier is the solution which recently
has received the largest amount of
publicity in this country. Amplifiers
made by two American manufacturers
utilize a type of output stage which is
less well-known, but which Williamson
himself regards as a more substantial
improvement than the Ultra-Linear.
Still another American amplifier uses
'\ a type of output stage which apparent­
4711
IW ly can give results better than some
; of us expect of it.
..

Ultra-Linear
Where the original Williamson cir­
cuit uses triode-connected beam tetrodes
J
in this output stage, the Ultra-Linear
version has the screens connected to
taps on the output transformer pri­
mary. A typical example of an Ultra­
Linear Williamson is the Brociner
model UL-1. (Fig.!.) One explanation
Fig. I-Schematic of the Brociner model of the low distortion of this output
UL-I amplifier, an Ultra-Linear type. circuit is that, since connecting the
screens to the plate terminals of the
transformer primary (the triode con­
nection) produces a plate characteristic
which is nonlinear in one direction and
SEPTEMBER, 1953
 connecting the screens to the E plus distortion, is greater. Second, since izer and in the tone-control circuit (not
terminal of the transformer (the te­ the negative feedhack path is through shcwn in Fig. 4). The power-amplifier
trode connection) produces a character­ the capacitor, less feedback results at section has not only the output-stage
istic which is nonlinear in the other low frequencies and the gain is in­ cathode feedback previously mentioned
direction, screen connections to points creased to compensate for the low-fre­ and over-all negative feedback, but also
sf'mewhere between the center and the quency loss in the coupling capacitors. positive feedback between the cathodes
ends of the primary will produce ap­ The phase inverter in this amplifier of the last two stages.
proximate linearity'. (Sc� Fig. 2) is not direct-coupled, as it is in most This ampli!ier has the conventional
Williamson points out' that the variations of the Williamson amplifier. direct-coupled phase inverter (VI-b) .
Ultra-Linear circuit is in effect a No 1 % resistors are used in the push­ The plate voltage of VI-a and the grid
tetrode circuit with negative feedback pull stages, but other provisions are voltage of VI-b are adjusted with RI
from plates to screens and that the made to reduce the amount of unbal­ so that the distortion in the VI-b stage
same improvement in linearity can be ance. The push-pull voltage amplifier will be a minimum. Few critical com­
obtained with other circuits which give has an extrd retlistor, RI, common to ponents or adjustments are used. The
equivalent amounts of negative feed­ the cathode circuits of both tubes. The phase inverter load resistors, R2 and
back. In fact, he remarks that equally usual cathode potentiometer for bal­ R3, are matched by selection of pairs
good results can be obtained with an ancing the output stage tubes is used. from ordinary production tolerance
OI'dinary beam-tetrode output circuit by (The tube heater which is connected types. Other resistors are unselected.
using sufficient feedback around the in the cathode circuit of the output Tl.e common cathode resistor in the
entire amplifier. The disadvantage of stage belongs to the preamplifier. Each push-pull voltage amplifier stage im­
that arrangement is that sufficient over­ of the amplifiers described in this ar­ proves the balance in that stage. The
all feedback to enable a tetrode output ticle, except the Brociner, includes a use of multiple feedback loops in the
circuit to better the performance of the four-stage preamplifier, equalizer, and amplifier makes it possible to obtain the
original triode Williamson may be dif­ tone-eontrol circuit, using two twin
ficult to obtain without instability. It triodes. Only the main power amplifiers
can be done, as one of the amplifiers will be described in detail.)
to be discussed in this article demon­
strates. Inserting some feedback in the logen model DB20
output stage makes it possible to re­ The Bogen amplifier has other un­
duce the distortion any specified amount usual features in addition to the output
with less over-all feedback. stage design. It is particularly notable
Coupling from plates to screens is for its many uses of feedback. In ad­
only one way in which feedback can be dition to inverse current feedback in
produced in the output stage. Feedback six of its eight stages, due to unby­
voltage . can be inserted between the passed cathode resistors, interstage Fig. 2-Triode, tetrode and ultra-lin­
grid and cathode and it would be more negative feedback is used in the equal- ear circuit plate characteristic curves.
effective than feedback voltage inserted
between the screen and cathode. Such
an arrangement has been used in a
British amplifier, the Acoustical', and
in the two American amplifiers re­
ferred to above, the Bell model 2200
(Fig. 3) and the Bogen model DB20
(Fig. 4) . It can be seen from the dia­
:-rams that the voltage across the feed­
back winding of the transformer ac­
tually varies both the cathode-to-grid
voltage and the cathode-to-screen volt­
age (inverse feedback is applied to both IN

the grids and the screens) . FigurcJ

given by Williamson' indicate that this
arrangement gives less distortion in the
output stage alone than either the
triode or the Ultra-Linear tetrode cir­
cuit. Specifications published by the Fig. 3-The Bell model 2200 amplifier. Note that first stage is a.c. coupled.
manufacturers of both the Bell and the
Bogen amplifiers indicate a harmonic
distortion of 0.3% for a._ output of
20 watts. Vari-.tions of distortion in
the Bogen model DB20 with power and
with frequency are shown in the curves
of Figs. 5 and 6, which were furnished
by the manufacturer. AF IN

Bell model 2200

The Bell amplifier ha:.; a few notable
variations on the conventioaal arrange­
ment, aside from the output stage. The
arrangement of two resistors and one
capacitor in the cathode circuit of the
first stage serves two purposes. First,
the cathode resistor is partially by­
passed, so that there is less feedback
within this stage and greater over-all STAGES
gain without feedback. For the same
amount of over-all gain with feedback,
the amount of feedback which can be
used, . and the reduction in over-all ...·ig. 4-The Bogen DB20 a m p l ifie r uses both positive and negative feedback.

RADIO.ELECTRONICS
 srecified level of distortion without
i... critical b alancin g of the output stage .
2.5 &
" Stromberg-Carlson model AR-425
C
...
III

2.0 The power-amplifier section of Strom­

berg-Carlson model AR-425 custom am­
Fig. 5-A g r a p h plifier is shown in Fig. 7. A direct­
1.5 showing the varia­ coupled phase bverter circuit is used,
tion in harmonic but the push-pull voltag e-ampli fier stage
1.0 � distortion at vari­
ous power output
levels in the 80gen
does not have a common cathode re­
sistor and close-tolerance resistors are
not used for b alancin g . A simple beam
0.5
-�
) model D820.
tetrode power stage is used in this
amplifier, with no intrastage feedback,
yet the harmonic distortion at 20 watts
0.0 output is approximately the same as in
o 4 8 12 16 20 24 28 32 36 the other amplifiers described in this
Power Output - Watts @ 1000cps article, according to curves furnished
by the manufacturer (Fig. 8). Using
frequencies of 60 cycles and 7 kilo­
cycles in a " to 1 voltage ratio, this
amplifier develops only 0.7% inter­
modulation distortion at 15 watts out­
put. Using 40 cycles and 7 kc, the inter­
I .-
modulation distortion is 1.2% at 15
: WRttS. When making power runs to
determine the percentage of inter­
10
modulation distortion, the 15-watt com­
• plex waveform measured by the dis­
• tortion meter is equivalent to 23.5
Fig. 6-Harmonic watts of sine-wave signal having the
,
distortion i n t h e same peak value as the .resulting signal
• D820 a t the 10-, produced by intermodulation within the
15-. and 20-watt audio amplifier.
levels for frequen­
The low distortion is accounted for
cies up to 20 kilo­
cycles. partly by the large amount of over-all
S negative feedback used. To overcome
the resulting tendency to oscillation at
�I •• .-
I
,la.
lot ./ . •
....0 --
high frequencies, the gain of the first
two stages at high frequencies is re­
duced by two separate means.'
10 so 1)0 100 1)0 1000 1000 $000 ICPoo la,000
The output of the first stage is
shunted by a combination of two capaci­
tors and one resistor, Cl, C2, and RI, so
that the effective load impedance at
high frequencies is reduced. Also, nega­
tive feedback at high frequencies is
provided between the plate of the sec­
ond stage and the cathode of the first,
through C3 and R2.
All four of the amplifiers discussed
in this article have damping factors of
around 15, compared with 30 for clas­
sical triode versions of the Williamson.
That is, the various tetrode versions
have an output impedance of about one­
fifteenth the speaker voice-coil imped­
ance, against one-thirtieth for the tri­
ode circuit. Williamson points out'
that the effective damping resistance
is the sum of the amplifier output re­
-IOan,m TO't\J WETER. sistance and the speaker voice-coil re­
sistance. The tetrode circuits therefore
have total d amping resistances, not
twice as great as the triode circuit, but
only about 3% greater. END

1 References
)
Fig. 7 (abo\'e)-The Stromberg-Carlson
I"Amplifiers and superlatiyes-an examination of
,/ American claims for improving linearity and
efficiency." D. T. N. William.on. and P. J. Walker.
2
AR-425 amplifier in the schematic has
I� 8.V a sensitivity of 2 volts input for full Wlrele.. World, September, 1952, p. 357.
rated output. Fig. 8 (left)-A graph
'''D..ign for a high-quality amplifier , " D. T. N.
I
V l.,..o' >I--'" William.on. Wir.I ... World. April, IM7, p. Ill,
of the harmonic distortion variation in
�� � � 1"""6 the AR-425 amplifier at dill'erent fre­
and May. 1947, p. 161.
• "An ultralinear amplifier." D. Hafter and H. I.
K.roe., Audio fngln ••rlng. November, 1951, p. 15.
quencies as well as power output levels. "Ultra·linear operation of the William.oft ampli­
2 10 14 11 22 fier," D. Holler and H. I. Keroe •. Audio fngln.er­
.IUOIOOUT-.m Ing, June, 1952, p. 2'.
IlIN£ OJIMS IN CPS;- A·l) .12000; 8- 5000,10000; C- SO.IOO; • "Review of B"tish Amplifie"." J. Moir. FM-TV,
0'400,1000 October, 1951, p. 30.

SEPTEMBER, 1953
♦ Verso Filler Page ♦
Adapting the "Ultra-LJneaf....... .AA
...
...... ..
MII ...

to 6550 Operation
By HERBERT I. KEROES
Acro Products Company

feedback phase characteristics and in­

crease the stability margin of th e am­
plifier.
The output stage is coupled to the
driver through a resistance capacity
network which provides conventional
RO coupling at signal frequencies and
an attenuated direct coupling at sub­
sonic frequencies. This again intro­
duces an improved low-frequency phase
characteristic which adds to the sta­
bility margin of the amplifier. The use
Over-all view of the of thi s combined RO and direct cou­
"Ultra·Linear" amplifier that
pling is made possible by the choi�e of
has been adapted for Tunq.Sol 6550'8.
fixed bias operation of the output tubes
A new high-power output tube, the 6550, will find many whereby the required negative bias is
obtained from a separa:te bias supply.
The fixed bias supply consists of T., a
applications in converting present-day amplifiers and 6.3 volt, 1 amp. filament transformer;
a 50 ma, selenium rectifier SR.; resis­
in new equipment. Up to 100 watts push-pull can be had. tor Rc.. ; and electrolytic capacitors, O�

INCE its introduction many years

and 0.3•
In order to redu ce hum to a mini­

S
system and, in general, it has been ob­
ago, the Williamson amplifier has served that in amplifiers of analogous mum in preamplifiers that are to be
undergone a few design changes to design, the unit of greatest capacity powered from the main amplifier, a
further improve its performance. As will sound best. positive bias has been applied to the
originally described by Williamson, the The attainment of high power in heater line through resistors R", R..,
amplifier was a I5-watt unit designed audio amplifiers has become relatively and capacitor, OH. If a separately pow­
for low distortion, uniform output, and easy and inexpensive due to two fac­ ered preamplifier is to be used, this
small phase shift over the entire audio tors, the increased efficiency of the network can be eliminated, together
range.' Since the original conception "Ultra-Linear" output circuit and the with the hum balancing potentiometer
of t he Williamson amplifier, American introduction of new output tubes with R"", and the centertap of the 6.3 volt
manufacturers have jumped on the greater power handling capabilities. winding on the power transformer T.
bandwagon and today one will find One recently introduced tube, the can be grounded. "B plus" voltage for
many variations of the original circuit. Tung-Sol 6550, is particularly adapta­ operation of the preamplifier can be
Performancewise there is w ide varia­ ble to output stages of the "Ultra­ taken either from poi nt X or Y depend­
tion among the different units made in Linear" type and can be used to ad­ ing on the preamp to be used.
this country. One of the circuit im­ vantage in the "Ultra-Linear" William­
"Ultra-Linear" Output Stage
-provements made by American manu­ SOlt circuit to provide an amplifier of
facturers came with the application of 60-watt capacity having an intermodu­ The "Ultra-Linear" type of output
"Ultra-Linear" operation to the output lation content at maximum output of stage is characterized by output tubes
tubes, a mode of operation which dou­ 6/lOth of one per-cent. This amplifier of the tetrode type with the screens of
bled output power and further reduced differs only in a small degree in dimen­ the tubes connected to taps equally
distortion.· This amplifier has been sions and number of circuit elements positioned about the centertap of the
widely accepted by audiophiles with from its predecessors, and many WiZ­ output transformer. The operation of
the result that there are about twenty Ziamson-type amplifiers can easily be the stage can most readily be under­
commercial amplifiers on the market modified to take advantage of the im­ stood by the following considerations:
today which incorporate this design proved performance. first, if the screen of an output tube is
feature. connected to the plate, the tube func­
Amplifier Circuit tions as a triode, and the plate char­
The application of "Ultra-Linear"
operation to the WiZZiamson-type am­ An examination of the circuit dia­ acteristic curves are concave upward.
plifier increased the output power to gram reveals the basic Williamson cir­ Secondly, if the screen is connected to
30 watts using the same type of output cuitry of the first three stages. The "B plus," the tube operates as a tet­
tubes operating at the same voltages. first two, the input voltage amplifier rode, and the plate ch a r a c t eris t ic
When this circuit was first introduced and direct-coupled cathodyne phase curves are concave downward. If, how­
it was immediately noted that the new inverter, are familiar and unchanged ever, the screen is connected to a tap
combination provided better sound, even with regard to tube type, the on the primary of th e output trans­
even at the low volume levels which 6SN7. The driver stage also remains former, a t ype of operation is obtained
the ori gi nal amplifier could handle. a 6SN7, with but one change. Individ­ midway between triode and tetrode.
This phenomenon has resulted in a new ual cathode resistors have been added Depending upon th(l type of output
evaluation of the power requirements to provide a slight amount of local tube used, the tap can be chosen to
of an amplifier as a part of an audio feedback in order to improve the loop result in an almost linear set of plate
 consideration of voltage feed bac k only.'
Power Output SO watts @ 1000 Cps: within ±.5 db of 1 kc. level The output transformer is an Acro­
@ SO watts oYer range 20 cps to 30 kc. sound TO-330. This transformer is
Frequency Response ± 1 db @ 1 w att. 2 cps to 220 kc. ideally suited for the 6550 tubes, pro­
IntermodulaUon DlstorUon (SO and 10 watts-.07'}'o: 20 watts-.lO%: 30 watts­ viding the correct impedance match for
3000 cps mixed 4:1. equlv. sine· .IS,},o: 40 watts-.25%: SO watts-.40%: SO
maximum power and lowest distortion,
wave power) walts-.SO%
and primary taps located a t 40% of
Square.Wave Response Rise time on 20 kc.-2 microseconds: overshoot on
20 kc.-none observed: ripple on 20 kc.-approx. the total winding. The frequency re­
1%: droop on 20 cps-5% sponse of the TO-330 is fiat ± 1 db
Hum and Noise 80 db below rated output from 10 cps to over 100 kc., thereby
Nominal Feedback 20 db providing the necessary low phase shift
Feedback Stability Margin 10 db over the audio range for best feedback
Damping Factor 15 stability and faithful t r a n s i e n t re­
Sensitivity 1.3 volts r.m... for full output
sponse. The halves of the primary
winding are tightly coupled to make
Table 1. Perlormance characteristics 01 the converted "Ultra·Linear" ampllBer. available a full 60 watts of output over
the entire audio range 20 cps to 20 kc.
characteristic curves, and this mode of operation, it should be noted that the Although the nominal rating o f the
operation has been termed "Ultra­ feedback is of the power type rather transformer is 50 watts at 20 cps, no
Linear." It has been determined ex­ than the more usual voltage or current difficulty was experienced in obtaining
perimentally that the best operating feedback, and power is supplied to the full undistorted output at the low fre­
point for the 6550 is with the tap lo­ screen grids over the operating cycle. quency extreme. Too much emphasis
cated at 40 Cl" of the primary turns. It can be demonstrated mathematically cannot be placed on the fact that an
"Ultra-Linear" operation of an out­ that when power feedback is applied amplifier cannot be better than its out­
put stage has sometimes been described to the screen grid of a tube, the linear­ put transformer irrespective of the cir­
as the application of negative feedback ity of the plate characteristic curves cuit used. This component serves many
to the screen grids of the output tubes. can be improved over and above the functions in a feedback amplifier as
If this concept is used to explain the amount normally to be expected by a well as providing an impedance match
between the output stage and the
Complete achemaUc diagram of the SO·walt version of "Ultra·Llnear" WllUalDllon.
speaker. However, all of the necessary
conditions can be met by a propitious
choice of design and this unit can, in
fact, be improved in certain perform­
ance categories, for example, band­
width over :md above the circuit with
which it is associated.

Feedback Stability
In a feedback amplifier it is always
desirable to m a i n t a i n a m a x i m u m
amount o f feedback stability in order
to assure complete stability under all
conditions of output power level and
output load. The degree of stability of
a feedback amplifier is generally rated
in terms of stability margin, meaning
the amount of additional feedback in
db that can be added before the ampli­
fier becomes unstable and oscillates.
y This design figure is usually taken
A26
under conditions of rated resistive out­
put load. However, loudspeakers are
not constant resistance devices, but
present to the amplifier an impedance
containing a large reactive component
over a good portion of their operating
range. Moreover, in the band outside
of the range of the speaker, the im­
pedance may be almost completely re­
active. It is desirable, therefore, to
have a stability margin of 6 db or more
R._18,OOO 0""', III W. '�I. to assure complete stability.
R" �IO,OOO 0""', Vz w. ,�... R._IOO 0"'" wi,�wo,,,,d pol
R�l '"�Io"m. y� w. ,«,.
The stability characteristics of a
C�7 pp/d. ",,,... ic e.p.ci,o,
R,-470 0""', IIz W. '�I. C•• C•• Cu-20/20/20 p/d., 4'0 ... �I�e. e.p.ei. feedback amplifier are associated with
R" R_22.000 0""', I w. ,,.. ( .... /c,,�tI ±I%) '0' the bandwidth and phase shift charac­
R�7,OOO 0""', I W. '�I. C" C•• C., C.-.2' p/d., 600 ... up.ci,o, teristics of the amplifier circuit and
R.-JJ.OOO 0""', I W. '�I. C...-:l p/d., 400 ... e.p.ci,or
R.-22,OOO 0",", I W. '�I. C" C,_IO p/d., 600 ... oil or �I�e. c.p.ci/o,
output transformer, and there are sev­
RII, R/�70.000 0""', IIz W. '�I. Cn. C,_20 p/d., 1'0 ... d�e. e.p.ci/or eral choices available to the designer
Rn-470 0""', IIz w. '�I. Tr-O..,p.. , ".",. (AaolOllnd TO·JJO, u. to increase the stability of a given
R", R,,-220 0""', Vz w. '�I. lul) amplifier. First, bandwidth may be
R ... R/�7,OOO 0""', 2 W. '�I. (",.'e"�tI T.-Pown ,,,," ,. 400·0·400 ... @ 200 ..... :
, ... @ 2 .mp•• ; 6.J ... . @ , ."'PI. traded for stability. In this procedure
±l%)
Rn. R,1-470.000 0""', III W. '�I. (",./c,,�tI T_Fil. ".", ., 6.J ... @ I .",p. loss networks are added to shape the
±l%) CH� ".,., 200 m .. fi"�' e"de amplifier response curve so that the
Rn. RI1-47,OOO 0""', III W. '�I. (....'e"�tI S.....-5.p ...,. Iwile" response of the amplifier falls off by
±l%) F r--' "mp. li,,� /"'�
RfI-IO,OOO 0""', Ii,,�., ,.p�, pol F., F_II••mp. e.'''od� /.. ,e the amount of feedback plus the sta­
R... R.., Rl.-IOO.OOO 0" "', III w. '�I. SR�.10 m •• 1�1�,,;um rrc,;fi«r bility margin before the phase of the
R... R..-IOOO 0""', % W. '�I. y" Y...-6SN7 I .. b� feedback voltage becomes regenerative.
R•.-2'00 0""', , W. '�I. Y" Y,-6,'O , .. b� (T.. ".·Sol)
Secondly, gain within the useful band
R._I',OOO 0""', % W. '�I. Y_'Y4GA ,.. be
may be traded for stability . In this
procedure local feedback may be added turned on. Correct phasing can be re­ and a rise time of 2 micl'oseconds. The
to stages within the amplifier in order stored by either reversing the trans­ intermodulation figures are particular­
to reduce their contribution to the former leads or by reversing the con­ ly good, being only 0.15% at 30 watts
phase shift. Thirdly, the bandwidth of nections of the coupling capacitors at and 0.6% at 60 watts, for a combina­
the stages may be extended by the use the phase inverter section of the first tion frequency of 60 and 3000 cps,
of certain design techniques, and the tube. mixed 4:1.
phase shift correspondingly reduced. The tota:l cathode current per output Much has been written about am­
tube will run about 75 ma. with a plate plifier testing and on the interpreta­
The first method is subject to the
supply voltage of 425 volts and a grid tion of test results. However, the
criticism that it restricts the amplifier
bias voltage of minus 48 volts. If the deeper one goes into the field of am­
band reducing the rise time with re­
cathode current differs considerably plifier design, the more apparent it be­
gard to square wave response and, in
from this figure, it may be advisable comes that the best test instrument is
this m a n n e r , affects the fidelity of
to change the value of R", until normal the human ear with music supplying
transient reproduction. The first and bias and plate current is obtained. the signal source� Unfortunately, the
third methods may be combined; the
ear cannot supply a numerical rating
bandwidth increased and then loss net­ Conversion of Existing Amplifiers
of merit, but only a comparison of
works added. It will occur to many that their WiZ­ "better" or "not as good," The ampli­
An appreciable increase in the band­ liamsons may be converted to take ad­ fier described in this article has been
width of the amplifier described has vantage of the increased power output subjected to comparative listening
been achieved by the use of the TO-330 offered by the 6550 tube by simply tests with both the older "Ultra-Linear"
transformer. The response of the am­ adding a few extra components and Williamson using KT-66's and with
plifier, with feedback, is flat to over changing the output transformer. One other good amplifiers. Listeners were
200 kc. An adequate stability margin precaution should be taken, however, generally agreed that this amplifier had
of 10 db has been maintained by the against overloading the power trans­ many points of superiority.
use of methods two and three. A small former. The plate cUl-rent drain of the The relative importance of the power
amount of degeneration has been added output stage has been increased from amplifier in a high-fidelity system has
to the driver stage by the inclusion of the 100 ma. drain of the usual KT-66 always been a controversial subject.
individual cathode resistors. The sub­ tubes to 150 ma., and a power trans­ There are those who maintain that a
sonic bandwidth has been extended and
former that is operating close to low power amplifier of 5 watts or so
shaped by the addition of the 470,000
maximum rating will not be able to is adequate for good reproduction and
ohm coupling resistors RI' and Rts• With
supply the additional drain. Many power qualify this by the indisputable state­
these added resistors the bias devel­
transformers will, however, be able to ment that the average sound power re­
oped on the grids of the output tubes
take it, and one should not rush to quired for good room volume is no
is partially dependent on the voltage
replace the transformer if it feels hot greater than this figure. Others state
developed at the plates of the driver
in service. A safe operating tempera­ that a moderately good power ampli­
tubes, and a plate current balancing
fier is a much more perfect device than
control has been added to the output ture for this component is 140 degrees
other elements of the reproducing sys­
stage. The procedure of balancing plate F. which is an uncomfortable tempera­
tem, in particular phono pickups and
current has been facilitated by indi­ ature to the hand. However, If the
speakers, Both of these schools of
vidually fusing the output tube cath­ power transformer ran at an uncom­
thought fail to recognize some basic
odes. To check plate current the fuse fortable temperature before conversion, facts relating to requirements imposed
is removed. The fuse clip serves as a it should be replaced. upon the power amplifier. In the first
convenient tie point for the connection
case, although average room volume
of a milliammeter.
Performance may require only a few watts, peak
Construction of the Amplifier The measured performance figures
powers may exceed the average by 10
times or more, and it is the fidelity
The amplifier can be constructed on of the amplifier are given in Table 1.
with which these peaks are reproduced
a chassis 8" x 12" x 3". A careful ar­ The frequency response at low outp�t
that contribute to the feeling of pres­
rangement of parts p e r m i t s d i r e c t levels is flat from 2 cps to 220 kc. MaXI­
mum power of 60 watts is delivered at ence. With regard to the second point,
point-to-point wiring of the stages and
all frequencies between 20 cps and 30 it is true that there is still room for
a short, direct feedback connection be­
kc. The square wave response at 20 improvement in pickups and speakers,
tween the output transformer and the
first stage. The axiom for wiring am­ cps shows 5% droop, and at 20 kc. the however, any additional contribution
plifier stages is to have leads as short square wave is clean with no overshoot to intermodulation distortion in the
and direct as possible. It is desirable
to twist filament leads; also leads to Underchassis "'lew of amplifier. Note the polnt.to·polnt wlrlnq of clrcult.
the power switch and preamp power
connector. A neater job will usually
result if filament, switch, and power
supply circuits are wired first, then
"B plus" circuits, then signal circuits
less the coupling capacitors. The cou­
pling capacitors are added last, and
since these are generally large, they
may be looped over the space from
stage-to-stage. The coupling capacitors
to the output stage can terminate on
pin No. 6 of the 6550 sockets, since this
pin is not a tube connection and can be
used as a tie point.
Care should be taken when wiring
the output transformer to see that the
proper color coding is observed for the
primary leads. Make certain that the
traccr leads are connected to the out­
put tube that is energized from the
cathode of the phase inverter. If these
leads are incorrectly connected the
amplificr will m o t o r b o a t when it is
power amplifier makes itself felt by the speaker and the power amplifier. A high
generation of new combination tones power amplifier of low and constant
which further change the identity of a internal impedance exerts better con­
musical instrument, and produce the trol over the speaker characteristics
effect of blurring the sound. This latter at high peak powers.
effect can be easily demonstrated in a
comparison test by playing a poor re­ REFERENCES
cording with l ots of surface noise. The 1. Keroes, Hervert I.: "Bulldino tIle 'Wil­
noise will be there in each instance, liam8on' Aml)/ijim'," RADIO & TELEVISION
NEWS, Decembcr 1950.
but will be much less objectionable l!. NaJ/e,', David .f [(eroes, Ne,'vert 1.:
with an amplifier of lower distortion. "Iml"'ovhtO the lV1.l iiamso't AmlJlijic,',"
RADIO & TELEVISION NEWS, Feb""ary 195.�.
Finally, one must recognize that a con­ 3. . . . . . . . . . . . . . . . . . . . . . . : Wi"clcss En-
dition of interaction exists between the gineer (Editoriall. A"!llIst. 195,;.

Reprinted from RADIO & TELEVISION NEWS November. 1955

--..... * The quality of Acrosound transformers

c:::::>
� is a cce p ted throughout the world
as being unequalled.
r--t-
:::r- * The finest amplifier quality come s
CC from the optimum integrat ion
of t h e best ou t put transformers
=t1 and the best circuits .
::::::s * Ac ro-developed Ultra-Linear circuits
CC
Co'-) are likewise accepted as the best,
r--t- and the combina tion of these f actors
-- make the finest sounding and finest
::::::s performing amplifiers obtainable.
* Excellence of Acrosound transformers
Co'-) and circuits is illustrated
c:::::>
t::: by specifications of the TO-330 model
::::::s designed for high-power Ultra-Linear
c:::l- amplifiers using the new 6550 tubes.
Frequency response± 1 DB, 10CPS
to 100 KC
Po wer 50 watts 20 CPS-30 KC
rating- 100 watts 30 CPS-20 KC
Impedance-3300 ohms plate to
plate to 4, 8, 16 ohms. Primary
taps located at 40% of best U.L.
p e rfor m ance .
Price: $39.75 at leading distributors
* Catalog available u pon request
* Distinctive features protected by
patent!

ULTRA-LINEAR : : :
OUTPUT TRANSFORMERS
ACRO PRODUCTS COMPANY
369 Shurs Lane. Philadelphia 28, Po.

PIUNTED IN U.S.A.
A 50-Watt Power Amplifier
By DAVID HAFLER
Dyna Company

� THE last decade t rcm('ndous

I sll'iell's havl' be('n m ade in Ih(' m·t of

amplifil'l' design. Distortion flgul"t:s
of .P;} at normal Iist('ning \('\"('Is and
\('ss tha n 1% at rated oulput aa'(' now
routine specifications in conh'ast to
I"OITl'sponding figUl"es which ran as
high as fi\':: times as much t('n V('III'S
ago. Even Ih(' I('ast eXJX'nsh'e am l)li­
:il'l's hav(' response nat OVl'r th(' 20
'ps to 20 kc. range and the cosI of
.mllliners with th('S(' exceU('nt spcci­
li("ations has declined steadily in I'e­
("pnl years.
Thel'e have be('n times wh('n P<'op\e
have felt that thl' audio amplifil'I' has
rI'ached thl' limit of d('sign capahilili('s
Top chassis ylew
··but every time this complac('ncy has
of the commercial
allllCared. new amplifi(,I' aITangl'm('nts yerslon of the 50·wall
eam(' along which xO/tlllled h('It(,I' and amplifier. Over·aIJ size is
-Im·tro a new cycle of d('sign \·III·iants. 9" x 9" x 63/'" Includinq Ihe
I-:\"('n though distortion has \)('en low­ perforated shield which has been
·.. ('d and frequency I'espons(' I'xll'nd('d. removed in photo 10 show layout.
.
! h(' . .(' has still been I'oom fo . impl"II\"e­
IIl1'nt in listening quality. The cus­
lI.mary specifications fOl' amr,lifi(' ..s do A high-power amplifier, providing high-fidelity performance,
nul cOlTelate closely wit.h audihh' JX'I'­
fOl"nlanc('. Low distortion and lIat r('­ that is well engineered and suitable for home construction.
"pons(' are ncces5al'Y fOl' good
1>(,1'­
Thl' I'l'spons(' oulsidl' 1I11' audio hand tl1I' san1l' as Ihc Jll'l'fo ..munce und(·t·
fo ..mance. but th('y do not insUl'(, high
must be smoolh IInd ('onll'OlIed so that slNtcly slall.· conditions.
'Iualily. Th(' factol's which indicate
Ih('I'(, is no ()\'('I'sholll lint! I'inging ('x­ This is onc ('xplanaIion of Ihe 1;l('k
t!ood Iistenability ha\"e not all hel'n
hihit('d on a squHl'('-\\'/I\'(' signal input. of cOI'l'('lation b('l\\'c('n list('ning qual­
id"nlifiNl. and man�' of Ih('m an' not
and pl1lls(' shifl in 1Ill' audio han" must ily and m('aslll'('(1 Iw .. fOl'munc(', '1'111'
, .. hkct to ('valuation hv com'l'nlional
'lI ..:tslll·ing tt'chniqul'5. j 10Wl'\"CI·. basic
be III a minimum. lahlll'allll'Y ml'aslII'l'llll'nls 1\1'1' mall('
TllI'sl' 1'I':luin'nll'nls
lIJ'(' all inh'I'­ wil h sin(' and sqUaJ'(' WII\,(' soun'C's
.l"il('rin fOI' good sound ar(' Iwing ('s­
relal('d 10 the slahililv dlaJ'a('I('l'is­ whil'h 110 nol duplkall' Ih(' \'m'ying
I a h Iished t1l1'ough ('xl l'nsh'(, conI 1'011 I'd
lics (If Ihc amplifil'I' u ; III('1' fc.·l'dhal·k (Iynalll ies IIf Il lus iea I SOU"l'('S, ] n IIddi­
lisl('ning ('xp('!'i('nCl' allll t hes(' criI ('!'ill
condilions and wilh IIl1' l'I'gulal io n I i"n. IIl1' lal lOl'aI Ol' Y Il'sl USI'S a fixl'.1
a ..('Imsic to the nl'W amJllifi('I' d('sign
chal'8cII'I'istics of Ih(' I II)\n'I' supply, ,\11 load iIllJlC'llanl'l' whil(' 1111' IIntJllilicl' in
I" h(' d('scl·jJlCd.
umplilil'I' which is on 1111' ,· .. ..gc of ill­ a PI'II1'1 i('al silualion wo ..ks inlo a IlI1ul
Thl'r(' ar(' 5e\,('I'al impol'lanl f(,HIIII'cS
slahilily uocl(' .. conditions of slwakl'l'
in addilion to disto ..tion lIIul fl'('quency of \"al'iahl.· imIK'c\ancl' like a loud­
which hll\'(' hc:'I'n id<'nt ifi('(1 liS iml lOl'­
loading Iwhich m'(' mOI"l' cl'ilical Ihan spC'ak(,I' Iwhil'h mny 1)(' flll'lh('I' ('om­
..('sisli\'I' loading I cannot l'xhihil gOIHI
lant in d<,ll' ..mining lIIulilol'Y (II ... lily, plicall'd hy 1111' associatc'd CI'OSSo\'('"
tl'ansiPnl pI'I·fo ..mam·,' sil1l'1' SiJ,:IHtls nl'l wOI'k I. 1'l'l'fOl'ma n('" wilh Iransit'l1l
Th.,s<, warranl som(' discllssion,
1. T,.,,,,x;,',.' 11('''/''''''''''11'/' is " ..lIh­ muy ('xcilc 10w-ft'I'(IIII'n('y SllI'g(·s. high­ wa\·l'fOI'lI1S. IIl1d .... \,al'ying load ('On,li­
f"('Ilu..ncy ..inging. mill IM,ssihl�' pal'­ lions. I'alllwl h"I'\,alllall'c\ wilh slc'lul\"
., hl�' till' amplifil't· dla t·at·1 l'I'islil' whidt
.·..nl ..ihuh·s mosl In Iisll'ning qllalily asilk oscillalions IIndl'l' high oUIPlI1 slall' l..sl nWlhncls "n a simplt� I'� ,­
'assllmin\.! low (Iislllrlion amI suil ahll' l'OI111i1 ions . ]n simila r fashion. if P"W- sisl i'·I· load, Thl' Sllllllion for IhI' Ih'­
1'1' sUIII'I�' \'olla\.!l's shifl IIncll'l' sigmtl slgnl'l' is III IISI' all a\'ailahlt' Ih·si\.!1I
I" 1'1'1 I 1I('nc)o' t'(·sl"lI1s. . l. 11 is Olll' of Ihus(!
ha . .eI-to-(lu:.nlif�' qllalilil's 1)I'('allsl' a ('()IlIlilions, tllI'l'I' is a dlHngl' in flpl· . . · I·llnsi ... ·,·al ions whil'h will inITI'asl' sla­
IIlin\.! cnndilions; alII I 1111' P"l'f"I'lI1anl'I' hility a l ld 11I'('SI'I'\'I' 011l'1'lIlional dlm'­
Il"IInsil'nt. hy (I('finilinn. is 11 non­
l'I'IIl'1ilivc wllv('fol'm whil'h I'all only wilh Ilymtmit= I"'ogl'am SOUI'("'S is nol al'l .. l·isl it's Imell'l' (I�'nllmic signlll inpllls
hi' 1tI'III'(lximat('ly "C'I, .. C'sl·nh·t! hy wilh 1'I';wl i\'1' loads. Tlwn hl' musl use
slIuaa'e wave's and similll" non-sinusoi­ Flq. I. IM distortion Yersus power output. his I·a .. s as 1111' linal Il'sl insl ..ument
.
dal signals, :!, "fh'l' Imnsi('nl 1,(' ..fOl·mllne('. tl1t'
GO()lt tl'ansil'nt \I(,I'flll'll1l1ne(' ('nhtils IlI'xl implIl'lanl faclOl' in Ihe sound
l'I'il iC1I1 damping of th(' lI'ml llilll'l' so qllalily of an umplifil'l' is ""f'tllltll,.
IhilI Jlulsl'-tYJI(' si�nllls do nol callSI' ,"'11"('" ",/mllillf/ (·"'H/{.';I.II,"· Although
oscilhtlOl'Y lIlII'g('S which 111'1'1'111' III IhI' 1111' 101111 III)WI'I' 1'('quiJ'('m('nl nec(,lIsar�'
for go()(1 1'('111'1111111'1 ion is 11 conll'O­
1111 IJIU I liS s P U t'iO lIl; signllls. 11 IIlslI 1'1'­
Iluil'(,s ('xlt'I'nll'ly wid(' hllndllllss so I Ij H'I'sial issu(·. 1 h(' l'I'Ilui!'l'rl I)()w('t· fOI'
---
Ihat sI ('('I' fl'ont('(1 signllls of the rt'U/;.�I it' l'l'III'oduction is in excl'SS of
0 40
sqUIII'('-WII\'(' lyP<' 81'(' not dislll .. I('11.

June. 1956 61
 I WATT --...._---
.. ANY L[VEL U� TO SO W"TTS ----....1.--1
. WATT
stage, without feedback, will put out
50 clean watts over the entire audio
band and has frequency response plus
or minus 2 db from S cps to over lOO
..
o kc. These characteristics arc further
impl'Oved by the proper use of nega·
th'e feedback.
The 6CA7/EL34 is a linear tube

IH1I11� g 11111 1 11] [II� Illlllllb£EIl1111

which is designed for operation as a
pentode. It has a suppl'essor gJ"id
which controls the space charge in the
'" tube �nd provides a high order of lin­

tllll lmll]l l! 111111 1 <'llfll�ftlllll l[

earity under reactive loading condi­
tions. This type of performance is su­
perior to that of bcam tetrodes which
t pnd to show increased distortion un·
del' speaker loading conditions. If
screen loading is used with this tull(',
. I() lO 100 IKC. lKC. IOICC. lOKC. IOOKC. it will aITect the space charge contl'Ol;
FREOUENCY RESPONSE-CPS but if the propOl·tion of screen loading
Flq. 2. (A) Frequency re.pon.e 01 Ihe amplifier at any le.el up 10 50 wall.. (D) is small, the beneficial aspects of the
The harmonic dl.lorllon alSO wall. n frequency. (C) Pha.e .hlll n frequency. pentode design can be maintained. In
this amplifiel', the screen load is about
30 watts unless speakers of very high in the discrepancy between lab tests lO'I'r, which preserves the inherent
efficiency arc used. This can be dem­ and listening tests. linearity of the tubes, The combina­
onstrated by recording a piano and 3. Even when excellent transient tion of plate and screen impedances
then playing back the tape at the performance and adequate power han­ was empirically determined as that
same level as the original piano. Scope dling are available, a circuit may still which gave optimum performance at
patterns show definite clipping with a not give the best sound because it is both high and low levels under a wide'
3D-watt amplifier and medium efficien­ critical as to parts and layout and range of load conditions.
cy speaker. therefore has insufficient reproducibil­ The output stage is operated in class
The reasons why more power is re­ itJ/. It is necessary to have tolerance AB.. Class AB. or class B operation
quired than most people realize in­ lutitude in the design so that any would have been more efficient, but
volve both psychological and electronic builder of the circuit will get the per­ both of these conditions of operation
factors. For example, a fixed level of formance specified. It is not suitable have relatively high distortion at low
acoustic power sounds louder when to have an amplifier in which the de­ levels. and they have such wide cur­
issuing from a stereophonic source sired performance can be obtained rent swings that operational condi­
than from a single-channel source. only by careful balancing and adjust­ tions shift with changing signal level.
Therefore, realism which parallels that ment. In some designs, the movement In order to maintain optimum tran­
of stereophonic reproduction requires of a lead can affect stability or hum. sie'nt performance, the essentially Iin·
at least twice as much power if the This is undesirable since the user has car operating conditions should be ad­
eITect is to be simulated from a single­ no warranty of continued high-quality hered to rigorously thl'Oughout the
channel system. Electronic factors performance. This problem received entire signal cycle at all power levels.
which bring about high power require­ careful attention in the design of the This entails good power supply regu­
ments are based on the wide increase circuit to be described since the am­ lation, and also involves the use o f
in bandpass and dynamic range of plifier is available commercially in kit fixed bias. The bias i s obtained from
present-day source material. To han­ form and many users do not have fa­ a supply of fairly low impedance so
dle wide dynamics over a wide fre­ cilities for adjustments and correc­ that it does not change at any signal
quency band takes power. If "ID watts tions in the effort to get peak per­ level, and operation remains linear
is enough" was correct 10 years ago, formance. under dynamic conditions. If cathode
then 40 or 50 watts is required now These considerations of tran,�ient bias were to be used, the operating
since the loudness range of records pel'!ormance, adequate power han­ point of the tubes would shift at high­
has been increased at least 4 or 5 dling, and repl'oducibility of charac­ er levels and give a form of transient
times. teristics were the guiding factors in distortion.
As the frequency band has been ex­ the' design of the amplifier to be de­
tended, the power requirements are scribed. The attainment of these de­
The Phase Inverter
more severe. At frequency extromes, siderata required an integration of The output stage is driven directly
the impedance characteristics of loud­ components and circuitry of unique from the phase inverter which is of
speaker systems change from their combination. the split load type. This drive, without
nominal values. This mismatches the an intervening stage, is accomplished
amplifier and causes a downrating of
The Output Stage readily with the tube used, the triode
power capabilities. On some speaker The output stage consists of SCA7/ section of a SANS. This arrangement
systems, a 50-watt amplifier may be EL34's in push-pull, matched at 4300 is an old tried and true one which has
limited to 5 or 10 watts at 30 cps even ohms plate-to-pla.te with the Dynaco now been revitalized because of the
though it measures full power on a A -430 output transformer. With the efficacy of the new tube type which
resistive load-this is another factor operating conditions selected, this permits adequate voltage ampliftca-

Fie,. 3. The .quare·wa.e performance 01 the 50·wall power ampllBer at .arlou. frequencle•• a. indicated.

RADIO & TELEVISION NEWS

tion (to be discussed later) and phase
inversion in the same envelope.
The sole disadvantage of the split
load. or cathodyne. type of inverter is
that its balance falls at high frequen­
cies. The high-frequency response
from its cathode section is better than
from its plate section since the cathode
is at lower impedance than the plate
and is less influenced by the following
grid input capacitance. Unbalance at
high frequencies results in some in­
CA) Amplifier response to 1.5 volt d.c. pulse from Lottery. Initial portion
cl'ease in distortion and also limits the
01 trace shows switching in 01 battery; second portion shows switch·
amount of permissible feedback. This.
ing off. Both impulses are damped on first hell cycle, even with this
in turn. lessens the margin of stability. high level signal which drives the a�plifier to full output. See text.
Fortunately. however. a simple expe­
dient can provide accurate correction CB) Amplifier performance under changing signal levels is demonstrated
of the unbalance so that the drives to here by switch from 40 to 6 watt levels. No bounce or surge is visible.
the output tubes can be balanced out
to ultrasonic frequencies. Flq. 4. The tremaleDt perform aDce 01 the 50·watt power amplifier deacrlbed ID .exL
The correction can be made by
adding a small feedback capacitor on could be induced. This is a substantial to move out, then back, and that is all.
the side of the circuit energized from margin of safety which makes a very There is no tendency for it to "rock"
the cathode of the phase inverter. In important contribution to transient back and forth-a form of hangover
the circuit used. a 390 ppfd. capacitor performance. The stability margin is which blurs the sound of percussive
is connected from the screen of the obtained both through circuit design signals. The corresponding high-fre­
output tube back to an earlier point. and through characteristics of the out­ quency effect is revealed with square­
This capacitor introduces more feed­ put transformer. wave testing. Many amplifiers exhibit
back at the higher frequencies s.o that The margin of stability which exi:;ts a spike (overshoot> on the leading
the circuit has less gain as frequency at both low and high frequencies can edge of the square wave followed by
increases. This correction is most ef­ be readily demonstrated. Touching the ripples (ringing) which are similar to
fective on the side of the circuit which input grid of the amplifier sends a low-fre:tuency hangover. These condi­
has the most high-frequency response. sharp pulse through the system. If the tions are aggravated by capacitive
and the net result is to. balance the speaker cone is watched, it can be seen (Continued on page 116)
signals f!"Om the two sides of the phase
inverter. Thus the basic deficiency of Complete achematlc of 50·wall amplifier. Part. are atandard except for traD.formen:
the split load inverter is rectified.

The Voltage Amplifier

"'0
The phase inverter is preceded by
the other half of the 6ANS tube which
is used as a high-gain pentode voltage
amplifier. This is direct coupled, (t 11'
Williamson, to the phase inn'l·ter. and
the parameters are chosen to give min­
imum intermodulation distortion over
the two stages. A gain of 200 can be
o
rC'adily obtained in the voltage ampli­
fiel' stage even with the unbypassed
('athode.
The use of the pentode tube intro­
�,
duces some subtle advantages. The
input capacitance is very low so that
there is little shunting capacitance to
attenuate the high frequencies of in­
put signals cven from high source im­
�,. RI7 'Ut
pcdanccs. Also, undel' feedback condi­
t ions , the low Miller elTect in the
pentode makes the circuit uncrit ical
47�V
as to inpu t source. In triode st.ag('s,
parasitic oscillations may be encoun­ Tbe 6CA7 and n34 are
I('rcd whcn the input source has high Impo�ted tllbe. and are
Identicat. The 6CA7 I. dl.·
capacitance such as that from a
trlbllted by Amperex Elec·
shielded input cable. tronlc Corp. and the n34
The high gain of the pentode tuhe (Milliard) by the Inter·
... UCH£D
eliminates the necessity for an addi­ national Electronic Corp.

t ional stage. This, of course, simplifies R.--470,OO(l olom, % w. re •• C,-J90 /Jp/d .• 1000 '. e_p.cilor
the amplifier and makes a substantial R.-680 olom, V. w. r,·,. CJ, C •. C., C,.-lO p/d.. '00 •• d«. up_eilo,
improvem('nt in stability characteris­ Rz-47 ohm, III w. n· .. . CJ-7'O pp/d. mic. CtlPtU;tt,,.
R.,-1.2 ," � .oh ". . Y.z w. rt'f. C., C,-.I' p/d., 600 •• up.eilo,
tics since it is easier to take feedback RJ-Z70,OOO oltm. �/l .' . n· ... ert-IOO ",Id •• '0 ... �/ec. c.".e;lor
,wer three stagC's th an over four. R,;, Rr--47,OOO ohm. I w. r�J. (nrt"C"lu,J) l,-S.p .•. t . ... itch
R1-47,OOO ohm, Ifl .', ",'S, £,-, .",p. /"Ie
The Feedback lotlp RIC-JOOO ohm. J w. r"s. SR,-20 m... ut""i",,. r�cI;��,
R,�680 ohm, 1 w. ,.. . . J ,- Phono j.elt
Twenty decihels of feedback are in­ R". RIt-IOO.OOO ohm. 1/.: .', , .... (,""uht.d) T,-O'''"ul I'.tI •• (Dyn.co II·4JO, J�� l('xI)
corpomted in an over- a ll feedhack loop Ru, Ru. R,t-IOOD oh"" Yl w. ,..••• T,-I'u .....r Ir.,,, . • ., 0 · ' 0· 0 ·440 ". @ 200 ",•. :
Ru-6ll00 o"m, I ... . rt' •• 6.J •. @ 4 _,"p •. ; , •. @ J _mp •. (07,,_eo
which goes f !"Om output to input of the
Rn-'OOO 010 ... p ol. P·78I)
amplifier. The amplifiel' is sufficicnt Iy 1(,,-10,000 olom, % ...... . V,-6ANS 'ub.
stable that it would take an ad ditional R,...-'O o"m, 10 ... . win.·ft'ou"J r� •. Vt, V.-6CA7/ELH lub.
20 db of fC'edhack before oscillation C t--. I "Id., 4DO ,.. Cdp.c;lur V.-,U4GB ,ub.

J ... e, 1956 63
loading such as is causecl hy speaker
The Power Supply
cables. Even as little as 500 ""fcl. of
cllpacitance aCI'oss the speaker leads The power supply is quite simple ,
may throw these amplifiers into oscil­ utili zing cllpacitor innut with resistive
l ation . filtering. Suhstantial decoupling is in­
In this design h igh - freq uency ringing cluded to preserve the low-frequency
is practi ca lly eliminated by proper stabil it y characteristics.
choice of capacitor across the feedback The general configuration and the
resistor. On a loudspeaker load, there time constants of the pow er supply
is a minimum I'ing or overshoot on have been intC'gl'ated to give ins ta nta­
sq uarc waves to 20 kc. Evcn thc new neous recovery on ovedoad signals.
clcctl"Ostatic tweetcrs with theil' high High level sign a Is at'e nl'a t Iy an d sym­
c a pacitive loadin g will not cletet'iomte metrically clipped when their peak
high- freq u ency pel"formance nor will power cXCl'eds 100 watts. IInd concu t·­
long leads to the loudspeaker ·i n tl"Oduce rent pl"Ogmm material is unaffected .
in sta bility. In the listening tests which accom­
At low f requ en cies the fact that panied the design work, it was ob­
there is on ly one stage with coupling served that some amplifiers collapse
capacitors leads to a wide margin of t e mpomrily on overload, and their
stability. In addition, the output tram;­ powet· capabilities are substanti al l y
former has about 200 henrys primat·y less on musical material than on
inductance so that its response is nat steady-state signals-another lack of
clown to 6 cps. This insures low phase correlation between dynamic a nd
shift at low frequen cies, which makes steady-statc test methods.
it practical to maintain low-ft'e­
quency response of the amplifier to
Amplifier Performance
below 10 cps and still h ave complete The amplifi er ' s performance in
low - frequen cy stability. Some designs term s of co n ventional steady-state
gain stability by re striction of hand­ measurements is il l us trated in Figs. 1
width. That has not been done with and 2. Th e intermodulation distortion
this circuit. (40 cps and 7 kc. mixed 4 to 1) shown
It is wOl'th mentioning that the de­ in Fig. 1 is extremely low, due to the
sign of an output transform e r to be linearity of the output stage and
used in a screen loaded circuit is quite transformer augmented by 20 db of
critical since capacitive transfer be­ feedback. The t'('sponse curve ( Fig .
tween windings can produce multi­ 2A) is of interest as it shows that
vibrator or oscillator action in the thcre is no peaking of response outs ide
coupling from plates to screens.' This the audio band and also th at response
is avoided in the present design hy us­ within the band is tlat at all levels up
ing four primary sections each of to full output. The curve of Fig. 2B
which has the same proporti on of shows harmonic distortion at 50 watts
screen load im pedance. A un ique para­ of output, over the entire audio spec­
coupled winding arrangement gives trum. Most harmonic distortion meas­
tight coupling between all sections urements arc shown for 1000 cps
without capacitive transfer between where d istortion is at a minimum.
w indings . As a result , low leakage re­
actances are ohtained w ith ou t hig h
intet 'winding capacitan ces , and hoth
bandpass a nd stability charactel'istics
are impt·()\"(·(1.

Und er chassis
view 01 the com·
merclal versio n
01 amplifier. The
prin ted circuit
ass embly. shown
at lo wer left.
can be omitted
in lavor 01 in­
dividual par ts
11 the home con­
struc tor dupli­
cates t h e cir­
cuit usinq his
own component••

RADIO & TELEVISION NEWS

while at frequency extremes distortion the extra cl a rity of the bass was ap­
is higher. In this amplifier, the in­ parent.
crease in harmonic distortion at 20 cps High frequencies were smoother and
and 20 kc. is very low, and harmonic had less colo ra l i on. Some of the test
distortion is kept below 1% over the group were not aware of this until
entire band even at the 50 watt level. alter they had lived with the new
This means that response at fuU power amplifier, grown accustomed to it, and
is essentially distortion free-a condi­ t ried to return to their pl'evious units.
tion rarely met in audio .amplifiers Then the ol d favorite sounded rough
even when they exhibit flat response and screechy by comparison. This is
at fuU power. an interest ing psychol o g ical phenom­
It is difficult to depict the perform­ enon which shows up in acoustical'
ance specifications which are related testing-change for the better is not
to good transient performance. As as a ppal'ent as change for the worse.
shown in Fig. 3, the square-wave per­ The irritation factor is more obvious
formance Crom 20 cps to 20 kc. exhibits in retr og ressive change while improve­
minimum ringing while preserving Cast m('nt does not seem t o make as great
rise time. This does not indicate, how­ a d iff er ence.
ever, how the amplifier responds to A no t h e r intC're sting phenomenon
non-recurrent s ignals. which t he listening comparisons re­
Fig. 4 shows two oscillograms which Top .,Iew 01 the printed circuit board vealed was tha t the high sta bil ity cir­
evaluate the performance of the circuit uHd I n the amplifier showinq the lo­ cuit was generally operated at higher
under dynamic transie.nt conditions. cation 01 ae.,eral Important components.
and more realistic volume levels than
Fig. 4A shows a d.c. pulse from a 1.5 comparison equipment. When inten­
volt battery-a signal which drives the ventional disto rtion and C requency sities w ere high. the comment "turn it
amplifier close to overload. The initial specifications. e x h ihits outstanding down" was n ot heard for the new
impulse is the switching in of the bat­ transient performan ce, adequ ate wide­ circuit though it was often applied for
tery, and the second trace represents ha rill Il Ow el' capability, and I'eproduci­ the other models. It has been no ted
r.emoval of the Voltage. Both impulses bility. Th ese arc the criteria sought before that both wider bandwidth and
are instantaneously damped on till' Cor listen ability. higher powers arc accepted only when
first half cycle without transient dis­ accompanied by lower distortion. non­
turbances. Listening Tests
peaked response. and genera ll y cJcaner
Fig. 4B is an even more rugged tcst The author has always insisted that sound. It seems evident that the sp e­
of transient performance. This shows the acid test oC a n amplifier is the cific performance qualities of t his
switching from 40 watts of lOOO-cy clc l ist('n ing test. Even the most supe l'­ amplifier produce less listener inita­
signal to 6 watts. This drastic chang(' lath'e spe c ifications do not guarantee lion which permits higher level use
in amplitude has absolutely no eff('ct that the amplifi C l' will soun d well. without corresponding listener fa tigue.
on performance, and there is no surg­ ThereCore, extensive listening tests All in all, some very reluctant test­
ing or bouncing which can be oh­ were ma de on this circuit through ers were converted to the new ar­
served. \'arious stages oC its de velopmen t and rangement even though they had ap­
These tests are flrst approximations after the design was CrOl�en. These proached the listening tests doubting
to what happens under musical listen­ wc� made on the hasis oC compal'isons the possibility that such a simple
ing conditions. They show that till' with other amplifiel's of high quality amplifier could he as good, much less
amplifier can handle suddenly chang­ using a vari('ty oC sp eak er systems better. than the amplifiers which they
ing high level material without violent with AB switchin g panels. In addition were using. These reactions prompt
voltage swings, grid blocking, or other the amplifier was put to the test of the author to repea t the same thought
actions which are involved in poor re­ liveability-u s e in normal home concli­ as he expressed in 1951 when d esc rih­
sponse to large transient signals. tions of various individuals for pe l'iods jng Ihe firs t "Ul tra - Line ar" amplifie r: '
The power output of the amplifier is r a nging up to six months. !teactio ns "For sheer listening pleasure this
adequate for practicaUy any home in­ of listeners confirmcd the design amplifier represents the hest tha t can
staUation, even with low efficiency pl'emisl's of the ampli li (, I' as most t e st­ be achieved at the present state of t he
speakers. Since power is un�jj storted ers a gr(, l'd tha t t he new a mp lifier art. Others who have had the o ppo r­
at frequency extremes and momentary 1I01l1uied brttcr. These lis tener reac­
tunity to hear and try the circuit
overloads have minor effect, the useful tions agrl'ed in sev('ral rlircctions: agree with this; and these beliefs will
loudness levels can be pushed up to "DiI·ty" progl'am matl'l'ial sccmed to
not be shaken until somethin g come s
higher intensity than can be obtained b e c1caned up. The passagcs on rec­
al ong which .'Iomld.'I bette r. or at le as l
from some amplifiers of nominally ords which had appeared to he s li ght ly
sounds as good and can be built fOI'
higher power rating based on mld­ overcut no\\' scemcd smoothcr am! le ss
lower cos t. "
band sine-wave testing. I'aspy. This was intc l'llI'(' I ('d as mcan­ As p oin t ed out. this p o wer amplifil' l '
It is as difficult to quantify the re­ i ng I hal I husl' hig h Ic\'('1 passagl's I'C­
c an be e as il y duplicated by the home
producibility characteristics of a cir­ qu i l' cc! till' r\lhlecl pm\'!'r ancl Iransient
builder. Any of the special compo­
cuit as it is to specify its transient per­ r('sllOns(' of I h(' 111'\\' a mpl ilil'l·.
nents arc available direct from DJ/Ill!
formance. This amplifier, as a com­ TIl<' mid-hand "g;\I'hll'd" ('rf('cl which
Company, 5142 M ast er St.. Philadel­
mercial kit, Includes a printed circuit is fl'('quenlly associa tI'll wilh spcakcl'
ph i a 31. Pa. For those who would
panel with all the wiring up to the syslcms which h",'c C l'o ss m '(, I' net­
r athe r buy the unit in kit fo rm (the
grids of the output stage. This elimi­ work s was diminislll'd wil h I hc new
"Dynakit Mark n"), it is avail a b le
nates a major source of variability and ch·cuit. This appal'('ntly sl('ms fmm
from local part s jobbers or fr om the
insures reproducibility. However, the thc facl Ihal tilt' ml'thoc! of opcl'alion
company direct COl' $69.75.
stability margin of the circuit is such of (lCA 7/F.L.1'1's Iloe s not clclcl'iOl'ale
that almost any din'ct layout can be p<'r fol'manc(' on a compll'x l'I'active REFERENCES
used without d an ger of motorboating load to th(' samc ('xll'nl as occu rs 1. ",,/fa. Tlo"ill: ":1 60 11""" UIIl'o-l,ill­
cor it UI/;/i/icr," H,\()10 & T.:l�.:\"1810� .:":I-;w ... ,
or high frequency oscillation. Within wilh conv('ntional Cil·cuill·Y. Fel,,",,,,,',, NI.;:•.
the normal 10 r,? tolerance range of ll<�a \'y low-fl'cqu('ncy passag('s ha d �. Ilrr."cl'. 1>0"i(/: ".l{,,,lcl'lIi:c }'<>I'" 11'il­
components, the IM distortion wi\1 not liams(J)I.. .411tIJlijict·." �t udiuc"('lt, JClJllUU'Y
he 1\ ('I' dc·finition. This can hp a Ill'Ih­ 19;;6 .
exceed 1 % at 50 watts. With parts ut<,d 10 s('vI'l'al fal'lors: gl'l'all'r power .'. l,c"I"CII. D. 111. "1ll1 r.iI�o>,. R. 8.: "1'[,
trimmed for minimum distortion, the 0"'1111' 'rl'(ill�for",cr." lI·i'·clc .•.• lVorltl, J<II,­
cfillahility, low dislortiOIl at 1111' low­ "n,'1I 19;;6
IM can be made as low as ,25% at 50 fr('clul'ncy (.'!lcl of Ill«' hand. and mm­ ·1: IT"IIcr. D'"'I,I ,,',,/ lie,.",.." IT. I.: "A"
watts. "Il'l (' low-fJ'C'qu('ney s I ahi Ii I y. )o:\'l'n on llll,·" [,illcar Aml,li/icr," AuC/io, NovclIIl,cr
1951.
Thus the circuit, in addition to con- speakel's oC nominal 10 watt rating,
RADIO & TELEVISION NEWS 119
♦ Verso Filler Page ♦
 120 WATTS ••
'-@-.

OF
.. .. . ,

, . .- ... .�
... . - "

. .'

HI-FI POWER I.

Top·chaaaia and undet-chaaala viewa of Ihe h l'lh.power ampll.

fier. A aeparale power aupply. nol ahown here. la required.

By DAVID HAFLER Construction of an amplifier with IM as low as .15% at

Oynaco Inc.
full power, but be careful of your speakers with this one.

HE necessity for high pO\�er con­ "oice coil in the magnetic field creates

T
theil' preconceptions and founrl that
tinues to be onc of the major contl'o­ music sounded bettel' when higher am­ a "oltage across the amplifier output.
vel'si{'s of the high-fidelity field. plifier power was used. E"en the very If the amplifier has a low output im­
Much has been written and discussed conscrvative British. who fo r m e r l y pedance (as is common in high-quality
on this subject without final conclusion, rarely considered the need for mOI'e amplifiers today I, this voltage is short
but with the over-all effect of shi ft ­ than 10 watts, have now begun to use circlliterl, acti!"� liS a brake to the COl1e
ing power l'equirements upwards. A amplifiers which arc capable of 30 to movement. Some loudspeakers with
few years ago many people felt that 50 watts of middle-frequency power­ large magnet structures and compliant
10 watts was adequate while others ar­ and they justify this new viewpoint on cones arc shipped with a shunt across
gued for 25, and a few held out for 50 the basis of better sound. the voice coil to prevent movement of
watts anrl even more. Now, however, Many of the reasons why it is de­ the cone while in transit (as is also
the number of those who accept 10 sirable to have higher power than rlone with meters l. The ability of the
'
watts as sufficient is very small; and some people have believed necessary ampliCit'r to hold the cone free from
the minimum acceptable power for have been rliscussed previously' and ul1rlcsirerl excur�ions makes a rlil'ect
high-quality reproduction is generally it is not the purpose of this article to contribution to the clarity of low-Cl'e­
considered to be 20 to 30 watts; while go into these reasons again. Howl',·er. qut'ncy passages.
many people have come to feel that 50 onc of the reasons why higher powt'r Thel'e is nothing new thus hll' in this
or more watts arc required in most often makes better sound has not been discllssion until wc consider what h<lp­
home installations where maximum widely recognized; and discussion of pens if there is a heavy speaker cone
fidelity is desired. this will readily illustrate the SUbtle and a puny amplifier. In this extreme
There are two major reasons why differences which can be attributed to case, the amplifier cannot hold the
there has been this gradual acceptance higher power. cone in place. The undesired voltage
of higher power. First. speakers of Onc o f the functions of the amplifier generated by the overshoot of the cone
comparath'ely low etliciency have be­ is to provirle t'lectrical damping of the is more than the amplifier can control.
come popular and these require more loudspeaker cone.' For example, if the It would take a higher powered ampli­
power for the same acoustic output. cone is pushed in and then released. it fier to get the desired damping of the
Second, many )"X'opl{' ha,'e disregarderl snaps forwarrl; and the motion of the loudspeaker. Even if the cone is not

118 RADIO & TV NEWS

 h .".�. and the amplifi('r is not limited ing high po wer , tlH're is a limit<'C1 IJy the sIngle ell\'l'IU)Il' t),\,\'b <UI·
I" ; , f('w watts, the damping of the choice of circuitry. Onc(' it is d es ired 6ANSA I tuhe which functiolls as a
sl,,'akel' will be decreased as the am­ to go above 60 or 70 watts, there arc high-gain voltage amplifier and split­
pIiJi"I' gets near its full capacity-and serious problems of high-voltage··­ load phase inverter. The rectifier is a
lit Ih(' low frequencies where damp:ng power-supply re g u I a t i o n, expensive single GZ34 tube. Fixed hias is sup­
is impol'tant. the speaker impedance tube types, and choice of Olltput trans­ plied throu gh a sdenium rectifi('r and
rises above its nominal \'alue and the formers, These problems are mini­ is adjust ed to obtain 70 ma. of current
;wlplifi(,I' can no longer deliver its mized by the us(' of paralleled output through each output tube. Transfor­
rall'l! pow('r, tuhes so that cOI1\'('ntional circuitl'Y mation of t h i s circuit to 120 watts re­
:'\alurally, any consideration of how and volta�es can give hi�hel' powers.' quires the use of four output tu\)('s,
Ill1' COIII' motion is hdo in check 1IIso Th(' use of paralleled output t u\J<'s is two r('ctifier tub('s, and output and
;11,!>!iI'� 10 the problems of startin� the n ot thl' most el('gant solution to thC' pow('r t ransfOl'mC'I'S w hi ch can handle
' ''11'' in motion. Transient response, problem of high power, but it giv('s a th(' adrlitionot I·('quirements. In addi­
"hility of the system to get in mo- simple and comparatively i n(' xpe nsiv e tion. allY change ill transformers gen­
11"11 "I' to stop mO\'in�, is adv('rs('ly way to do the job without g et t i ng in­ erally necessitates some minor circuit
aIT.,ctl'o hy insufficient powcr. A firm, \'oh'ed with pot('ntially lethal \'oltages alt('rations to obtain proper compen­
ulll1luocliro bass" r cqu i res mort· powel' or monstrous pow('r supplies. sation of the f('('dhack loop.
than is indicated by m(,Hsurements of Th(' decision to use push-pull parallel Fig. 1 is the schematic diagram oC
a\·cragc or peak powers available in out put tub('s leo to experim('ntation t h(' 120-wa 11 amplifier using !his push­
music. Th e high-powered ampli fie r , w ith the DYIlIlh'il "Mark ,Ill" 1IS t he pull parallel output arrangem('nt with
thcrC'fol'c, has inhcrent quality which hasis for the new high-powerl'd oC'­ KTHX's. (The new EL-131 can also be
cannot \)(' o bt a ined in a low-powered si�n. The "Mark Ill" provioes up to us('d as the output tube .• The four out­
uni'. This doc� not answ('r the qu('s- 60 watts at low distOl'tion ann with put tuh('s an' matched with 2200-ohm
1 inn of how much pow('r is r('quired, wid(' fl'C'qupney response,' Whl'n this plate-to-plate impedance and the D:tJ­
j. rI".,s inoicatt' that the prohl('m is is us('d with push-pull para111'1 output IlItrl) A-450 output trans f ormer c:ln
', '. , . ill w hich the suhjective Iist('ning tuh('s, it is ca pa b le of pro vi d in g a full d('\i\'('r thc' full 120 watts from 20 to
1,'''1 <:an prove more than the convcn­ ] 20 \\'a Us, 20.00Q cps \\'ithout attenuation 01' de­
tiunal nwasurement. ThC' hasic 60-wall cOllfigurat ion USt·s formation of t h(' \\,;)\·('fo r m. Th('
FOI' tho se who ar(' int('I'ested in 115- KTR.'l tulX':; in tll(' output st ag l' dl'iw·1I 5.6-ohm l'C'sistor, R" in th(' common

+ C2

Cl

COOII

I
I
I
47lC
•
I
J
I
L __ ...J

{:;o-��A�I7N- ----------�----��RWI8�----�-------------------------------------------------t
AI6

��:r:
�AKE-OFF o-------���--�

""�u.:J "

Ft
·

GZ34
I. V6
Fiq. 1. Complele achemallc diaqram
aDd parla UatiDq lor Ihe 120·waller.

R,-IO.(}(}O ull ... , III ... ,� •. R•.--ICJ.U(J() "hm "'.r. ,•.•. 7,-0 .. , ,. .. , ',.,u. ( D"n_ ,n A·4'0 J
It . H4, 11.:. R .. , #(,.,.-1.'00 01.,.., 1ft ... r�,. 1(.,.-,".0000""'. Ih .. . r.· •. 7,-."n .... ' ,,_" •. 410·" ·0·4'U •. fj) 400 "._.:
6.J .. . @" 11 _ ... p.: , •• @. 4 _ ... p •• (D)"".Cf'
k ·�·'.6 ulrm, 1 .. . 't" . ("0,, ,�.,) C" er-lO pIJ •• -lOO ... . .., .. ". c""",;,or
/'A,1I/jl
A' . H�. 1(" RIf. 11.", RIl. Ru. R,;-4i ohm, 1 w. C,-..OZ IIIJ .• ZU" .", ('II,,"C'"''
CII,-4"0 ", •.• '(IV •• " " ,", '''t",,:
c •. C.-'O lA/tI., " 0 t', .. l.,c. ('"",uilur SIl.,-24J In_. .dt·"i.,,.. ,r,';",",
#(,.-6 MO "It".. 2 ... . n' •• C.-16 ,,/J .• 60fJ .. . .. I.. c. e",." ..;,,,, S,-S." .•.•. ••"il,h
Itlo._/(, t"J,m. I,.:z ... "("I,
C?-'J() ,,/J. . '2' .", ,,·1.,,', c""ft.";,,,, I"r-' ."'1' /"""
1t,:-Jl.(}()O ult".. 1 ... '�J. PC r--P,;nh·1I ci,," .. i, b".,1I (It'r J;_�,_", lor V,. V" V,. V.-KT'8 ,.. ,,�
11,-6800 oil .... 1 ... ,� •. ,. " . ... /wr.) V" V.--CZH , .. be

'February, 1959 119

 output tubes. If the KT88's arc op.

I.D J f1q. 2A. Th. Inter·

crated at 90 ma. each, they will bc
n m odu l a l i o n dl.tor. slightly over maximum rated dissipa.
&
rI tlon 01 the 120·watt tion, but the distortion will be rec\llc('d
NORMAL ampllll.r I. .hown to truly negligible proportions. For
I,..-- TII_D IS([ TUT!
here. The circuit may \ ' ery critical usage, lik" cutting mastel·
be car.lully trimmed
10 '20
discs, suc h an expcdicnt is justified
lor lower dl.tortlon.
POWER OUTPUT-WATTS Even with norm",,1 bias the distortion
(AI is only .25',� at 20 cps at 100 watts,
This quality of performancc at this
power ll.... cl is unique.
Flq. 2B. Pow.r r.· Frequency response of the amplifier
I I I I
�W(II OUT�T '011 2'" TOTAL MAIlIOOMC DtSTOIITlON .pon•• 01 the hl9h. is :=.5 db from 5 to OH'r 50,000 cps with
, pow.r amplifier I. smooth roll·off at cach cnd. Tht'

110
, .hown h.r. at a
level that /u.t pro·
smooth roll·off results in a tlat-tol'
square wave of good waveform. Th,·
due •• total h armonic amplifier has hecn checked on clectro.
d l a t orl lon 0 1 2%. static loudspeakers with no problems
10 100 200 .. ZK
FREOUENCY-CP$ of stability and with definite ad·
(81 vantages in hi�h·frcquency clarity on'l'
designs in which power is limited I"-
ca t hodes oC the out IJU t tu hes Itwo 11.2· reduced to .5',; hy the adllition of 20 the capacitive nature of the lourl.
ohm resistors are paralleled in the pho­ db oC negative feed hack. The circuit The listening quality of the unit COil·
tograph) has the dual Cunction of shown has several components which firms the theory that most of the hpnl"
degeneration oC out-of-balance signal provide Cor stable operation under fits of high power on conventional dy·
components which lowers the distor­ feedback conditions. These compo­ namic speakers occur ut low f,·equl'n.
tion' and also provides a convenient nents are similar to those used in the cies. Organ tones and other low fn"
check point Cor bias setting. If the "Mal·k Ill" with the exception of the quencies of prolonged duration exhihit
voltage across this resistor is 1.56 volts. 600 I'l'fd., volt mica capacitor
1000 more fundamental anrl less harmonic
the total output stage current drain is which connects from the screen of onc sounding as if the,·c is more de('p 10\':'
280 ma., which corr-esponds to 70 ma. output pair to the cathode resistor of frequency response. Low·f req u e ncy
per output tube. It is desirable, al­ the fi,·st stage. The similarity of the sounds of short duration, like d,·ulll
though not essential, that the four two amplifier circuits makes it prac­ beats. sound sharper. cleare,', nncl mol'"
output tubes be matched so that tht'y tical to use the printed circuit board widely separated from simultaneousl�
draw equal currents. However, differ­ of the "Mark Ill." which is also avail­ playing instruments. These differencl'"
ences among tubes up to about 10'/� ahle separately, in the 120-watt ampli­ appear on low·efficiency spl'akel·s ; t ,
will not advt'rsely affect perCormunce. fier with the single change of this one fairly high (not higher than original,
The d.c. bias on the KT88's is about mica capacitor. Naturally, the ampli· !e\·els. However, many listene,·s agree
50 volts. The r.m.s. drive to the output fier can also he built with point·to­ that a 120·watt amplifier sounds bet 1<'1'
stage must be about 35 volts (which point or conventional wiring without than those of less power on any spca k·
�ives a peak signal oC 50 volts) for difTiculty as long�as normal good wi,·­ er and at any level. Some of these rlif·
('ach grid. This is easily handled by ing practice is followed. The circuit is' fercnces arc subtle ones and for muny
the 6ANS driver, which can put out up not critical in any rpspect. The photo­ people they do not jOstify thc exp('ns,'
to 60 volts r.m.s. beCore distortion be­ graph shows the amplifier, without of a higher powered amplifier-- bllt
comes significant. It may seem un­ power supply which goes on a sepal·ate they do exist for those who do nol
usual not to have an additional driver chassis. on a standard 10" x 14" x 3" want to limit the quality of their high­
tube. but there is sufficient sensitivity chassis. (When lIsing a separate power fidelity systems.
and voltage swing to do the job with­ supply, the hl'ater wiring should be at One word of caution for those wl:o
out it and the omission oC an unneces­ least # 14 gauge to avoid drop in tli/e experiment with power of this magni·
sary stage eliminates problems which heater voltage.) tude--it is possible to blow out a
are introduced by the phase shiCts The power supply used in the ampli­ speaker by putting a full 120 watts of
which each stage contributes. fier is conventional capacitor input hum or other signal through the sys·
The split·load phase inverter con­ with a filtcr choke. The pair of GZ34 tem. Under normal musical listening.
tributes practically no distortion oC its rectifiers is rated for a full 500 ma. and even a 25-watt speaker can be uSl'd
own and does an excellent job oC phase the amplifier uses only about 300 ma. with the amplifier w i t h o u t WOITY,
inversion. Although more complex in­ quiescent and sli�htly over 400 ma. at However, if an input plug is ·pulled. (11'
verters have heen popularized in recent full rated power. Thus the supply op­ if some other tremendous surge is pili
years, the split-load type does every­ erates with comfortable margin. At through the system. any speaker mighl
thing requh'ed oC a phase inverter and full output thC're is a drop of ahout be damaged. This can be prevented hy
has one important advantage over all 107- in the "B+" (from 500 volts to fusing the speaker. A 1 amp. fust'
other types which has not been men­ 460 volts). If the reg:.:lation were such with a 16-ohm speaker will permit only
tioned by tnose who promote the n('w­ that there is no drop. the amplifier 16 watts of continuous power. How·
t'r arrangements. Its characteristics would handle a Cul\ 150 watts. How­ ever, short peaks of higher power will
are dependent only on the accuracy of ever, a supply with almost perfect not cause any problem. Since then'
the plate and cathode resistors and are regulation would be quite expensive is little likelihood oC exceeding 16
independent oC the age or condition of and cumbersome. \.... atts continuous (if you did the peaks
the tube used. All other types oC in­ Fig. 2 shows the distortion char­ would be well o\·el" 100 watts), this
verters rely on the gain characteristics acteristics of the amplifier. It operates shoulrl p,·o\'ide acll'quate loudspeakt'l­
oC one or more tubes; and unless these in class AB, without the high bias as­ protection when this high power ampli·
have means of balancing and rebalanc­ sociated with class B operation. There­ fier is usec\.
ing the a.c. signal output as the tubes fore, the distortion at low levels falls in
age, the performance will deteriorate the residual range of the instruments. REFERENCES
as time goes on. This disadvantage Although it is not ,·ecommended for 1. ITaliC". 1:"'I"i,·"","1I1.<t fu,'
nll";,l: "/'"".,."
lIi."·;:' 1." Al)IO .. 7',,· .\* Jo: "'.�. J."",,,,',"
does not apply to the split-load in­ normal use since the output tube life is 1!'';;
verter. shortened, it is practical to make the 2. '",,1,",,1 uJ
·" .lmlll;/; ... ,·
])nUlI,iug Factor:' ;I,;d • ."".11 '�':;!i
Without feedback and with ordinary distortion as low as .15% at 120 watts. .f. loA ,jIJ.l\·,,1t ·1;J"·fI·/�;t'-
c�hlll;tijtcr," iI,f,I, "'cl,rlff""!' nl.;:;
components and construction, the am­
plifier will deliver 120 watts at an IM
This can be done by trimming the re­
sistors for minimum distortion and in­
,
$. /lUll."" tl'J"iltrH: " ',." . ·.Ufl"�· Ill: ;I,id.
-'Iare". l!';S
distortion level of about 5'fr. This is creasing the cathode current of the

120 RADIO & TV NEWS

♦ Verso Filler Page ♦
 "ULTRA-LINEAR" OPERATION OF 6V6 TUBES

One 01 the best designs in

recent years covering an audio

amplifiN using 6V6 tubes.

The author, in this case, has

converted a G rommes unit to

Flq. I. The Gromme. JOOBA power amplifier after con·
... erslon to "Ultra·Llnear" operation. See diaqram of Fiq. 4. illustrate his design idea.
VER since the introduction of "Ul­ loudspeaker damping, Lastly. and Linear" aJTangement with 6V6· tubes

E tra-Linear" circuitry,'
I)('cn a steadily growing interest in
there has most.
lil/!'I/(
important of all. it has Cl
input-output relat ionship
more
at
in
heen
medium-powered
im'estigat.ed
amplifiel's
cal'efully,
hecn found that. the tube is well suited
It
has
has
amplifiel's utilizing this type of out­ most pOWCl' le\'els than ei ther I ..iodes
put stage coupling, The basic arrange­ or tett'oQes which n1l'ans that ils dis­ for this mode of opel'ation since its
ment has become popular in ardent tOl'tion is lowel' than othel' met hods of dynamic input-output char'acteristic
auoiophile circles and has also found op<'I'ation, This alone justi tics the use can hc lilleadzed hy pl'opel' selection
commercial and industrial applications of the circuit in those cases whe ..e lo\\' of a tapping point fol' screen connec­
where I.'xtremely low distortion is rc­ disto ..tion is the guiding c ..itel'ion, tion,
quirl.'d, The "Ultra-Lineal'" circuit has The charactel'istics of the 6V6 are
Essentially, the "Ultra-Linear" cir­ achie\'ed populal'ity in deluxe ampli­ not. at all similal' to the 6L6 family.
cuit is iIIustJ'ated in Fig. 2, The screens fiel' alTClngl.'menls s u c h as cOI1\'el'sion and the connection alTangement which
of beam power output tubes are con­ of I he Williamson cil·cui!." It has heen is oPI imum fOl' 6V6's is quite diffel'­
nected to taps on the pl'imary of the widely used with tuhes of the KTI,6, ent f/'Om that which can be used with
output tl'ansformer; or if it is desil'cd to H07, and 5R81 type for cil'cuits in the the lal'ge tube typcs, As a tetrode. the
opel'ate the sCl'eens at a diffel'ent a,c, 20 to 30 watt power hmcket· fOl' cil'­ 6V6 pennits 10 to 15 waUs of output
potential than the plates, to a tel'lial'Y cuits of truly outstanding cha ..aell'l'is­ d<'pending on plate supply \'oltage and
winding on the output transforml.'l', tics suitable fOl' the most critical us­ bias, These I'atings arc ba sed on the
Eithel' alTangement I'equil'cs a trans­ agc, Naturally. 200/' 30watts is a lot. point \\'hcl'(, clipping of a sine wa\'e
former with the COtTect ratio of sCl'cen of powel' for li\'ing room use- -just as hecomcs \'isible-·-which happens when
load to plate load if oplimum results 200 hOl'sepower is a lot of powel' for the grids start to go posili\'e. and the
al'e to he obtained. and a mismatch a deluxe automobile, Howe\'er, there d dving soul'ce cannot fUl'nish POWCI'
will lead to inefficiency and/ol' in­ are dcfinite advantages to high pow­ to the tuhes,
cl'eased distorlion, (,I'cd amplitiC'rs which arc opC'I'atl'd at a If the same tuhes arc tdode con­
The "Ultl'a-Linear" atTangement f .. action of tht'ir potential output juSI nect.ed 1 hy st I'upping I he scrcl.'n t.o the
has been mistakenlv referred to as a as tl1(' ..e an' ddinitt:' ad\'antagl's to plate), [>o\\'er output. using the same
feedback cil'CUit, This is not COITect high [>O\\'('I'pd ea . .s which an' 1'1111 at a crit('I'ia. is I'eouced 1021" to 3�!! watts,
since negath'e feedback would pmduce fl'action of tlwil' capabilitil's, Wlwn thC' "Ultl'a-Lincal'" connection is
a n'ducl ion in gain which does not Ne\,C' ..thel(·ss, not all of us want, 0 .. used, t hl' IlO\\'f''' ou t put r1l'lx'nds on
occur with the "Ultra-Linl.'ar" circuit, can a ffol'd, 200 hOl'sC'I)(lWI.''' ca .. s: a ne\ 111<' position of the s(,I'C('n taps, If a
It would be just as incolTect to I'cfer not all of us rppl thl' Iwt:'d fo .. , 0 . . wish SO'.; tap is used, P<)W(,I' is I'educed
to a triode as a tetmdl' with feedhack to spend thl' money fOl', itlllplifil'I'S of 10 a hout ollP-half of thc tet ..odC' capa­
as it is to analyze the "Ultm-Lineal'" :,10 01' l1lon� watts pO\\'('" . . itt ing. :,\lany bility, If it gn'alt'r I ha n SW'; tap is
dl'cuit as a feed hack circuit Instead audiophilt's and music Im'el's an' \'l'I'Y uSf'd, pow c .. is 1't'c1ucC'd towa I'd the
.
it must he considel'ed as a new and happy with ampli f iel's in th(' 10 10 15 tl'iode limitations. Ai Cl lappin g l)(lint
dilTl' .. ent tYPl' of tube st ructu ..e which wat t power h . . ackt'I, Tlw popula I'i t Y of of about 24'/(. powel' output is within
is neithe .. tl'iodl' nOl' tetl'Od(', this I'HngC' is demol1stl'a1<.'d hy thl' sales !)()(/� of till' tetI'(xlp condition, and dis­
The circuit (ll'ovidl's some of the ad­ success of Ihousands of \Villiamson­ tO/'tion at all le\'els up to maximum
vant ag('s of hoth tI'iodl's a III I tl't mdps. type amplifiel's as well as tens of t hOll­ is minimized, This point. thcl'cfore, has
and it O\'I'I'comes some of t he disadvan­ sands of lowel' cosl amplilit'I's using hpcn sclectecl as the optimum opc ..at­
tages of each of these types, J<'or l'xam­ 6V6 tubes p/'OviclinJ,( 10 to 15 watts of ing point fOl' "Ultra-Lincal''' uSP,
ph', it is more efficient. a III I pl'Ovid('s powel' output. Undouht<,r1ly, 1111' g ..eat­ It would be possiblc to take an I.'vl.'n
0101'1' IlOW('I' output. than t ..iodt's, Its est number of amplili('I's in honll' ust' lowl'r tapping point. and obtain slight­
capahilitit·s in this J't·sP(·ct. parall d the utilize the type (;Vl; t IIhl' in ont' of ly mOI'(, pO\\'CI' outpuI tha n t hc tetrode
capahiliti<,s of tet /'(Jcles, 1I0wl'\'('r. it s('\'('1'1\1 popular cil'cuil a .. . . ang(·nlt'nls. cOI1lH'ction. hut thl.' distortion at low
has low internal illlp(·t\ance. almost as all of which ha\'(' ('SSl'nti ally silllilal' le\'pls and the intel'l1al impl'dance hoth
low as triodes and a hout one-t l'n t h p" .. rormanc(' chal'act pristics, ht:'gin to increase as thc tap is hl'ought
that of tl'll'Otles; this IlI'ovides good Thc possihilities of usinJ,( the "UIII'a- clos,,1' to the Z(,I'O [X'r-cent point which

Ju" •• "54 43
 mum power output, distort:on at \'al'i­ Circuit Considerations
ous levels, and internal impedance There are many 6V6 circuits which
have all been weighed against each have become popular. but by far the

Eo
other. It must also be mentioned that most commonly used is that in which
listening tests at various tapping a twin triode phase inverter is used
MER points with no feedback around the to drive a pair of 6V6's; and feedback
amplifier validate this selection of the is carried from the output winding of
tapping point. This was done without the output ti'ansformer to the cathode
feedback on the assumption that the of one of the triode sections. This
best amplifier without feedback would basic configuration is simple. practical.
also be the best after the application economical. and adequate. The a.c.
of feedback. In these listening tests, gl·id·to-grid voltage requirements of
FI.,. 2. Ba.le "Ultra· Unear" arran.,emenL the triodes fell behind because they the 6V6 output stage are not stringent.
could not handle the power (after all. and the phase inverter supplies ample
3 watts is insufficient for musical (h-ive without the need for an inter­
peaks). the tetrodes were somewhat mediate push-pull stage such as is used
screechy and boomy (too much in· in the WiIliamson-type circuit. Since
ternal impedance for satisCactory thel'e are only two stages. the prob­
speaker damping). and the 24% point lems of utilizing feedback are simpli­
sounded natuml and smooth even fied (as there is less phase shift in
without connection of the amplifier the circuit). and the designer can use
feedback loop. less elaborate circuitry and compo­
Use of the "Ultra-Linear" circuit in­ nents while preserving a satisfactory
volves utilization of an output trans­ mat'gin oC stability.
former with the correctly placed taps. Generally the phase inverter tube is
A special transformer. the Acroso1tnd a high mu triode such as the 6SL7 or
TO-310. has been designed specifically 12AX7 in order to obtain as much gain
for this application; and its parameters as possible within the two stages. Ac­
were selected so that it would not limit tually. except for gain considerations.
the ultimate capabilities inherent in the specific type of inverter is of com­
the circuit. For example. its band­ pal'8tively little consequence-circuit
Ft.,. 3. Varlou. leedback arran.,ementa width has been set at ± 1 db from 10 performance is determined almost
from yolee coli 01 the output to the cathode cps to 100 kc. so as to provide the low completely by the mode of operation
01 an early .ta.,e. See text lor dlacuulon. phase shift and good transient per­ of the output tubes with respect to
formance desired in the most critical bias. supply voltage. and impedance
is coincident with conventional tetrode applications. Similarly. its distortion match; the quality of the output trans­
connection. characteristics complement those of former; and the proportion of feed­
Thus the "Ultra-Linear" operating the "Ultra-Linear" circuit and permit back. The voltage amplifier stage con­
point has been set at a compromise low distortion at both high and low tributes relatively little. as compared
level in which the factors of maxi· levels from 20 cps to over 20 kc. to the contribution of the output stage.
to the over-aH quality of the amplifier.
Ft.,. 4. The Gromme. lOOBA power amplifier conyerted to "Ultra·Unear" operation. Conversion of these circuits to "Ul­
tra-Lineal'" operation can be done by
substituting an output transformer
which has properly placed taps for
connection to the 6V6 screens. Gen­
('rally. this substitution will make an
immediate decrease in distortion.
••
If the original amplifier used a
screen dropping resistance. this is re­
RI
moved for "Ultra-Linear" operation;
11 C=-.!!:..._ and the screens are connected to the
tapping points on the primary of the
G output transformer. It· is important to
observe polarity and to connect the
screen to the I/(f·me primary side of
the transformer as that from which
the plate is energized. Otherwise an
oscillitory condition will be provoked.
fORMEIILY 8+ Similarly. polarity must be observed
fOA OUTPUT STAGE between upper and lower output tubes,
fORMEAl.Y
or the feedback from the secondary
All
100
elr
SCAEEN side of the transformer may be in the
1 IOW CONNECTION
incorrect phase and cause regenera-·
� <
I fORMERLY 54ICIO
tion.
ZW fOR SCREENS When the screen resistor of the
"=" ...
. original circuit has been removed. the
.
screen bypass condenser must also be
·
disconnected. This can be readily put
to good use by paralleling it across
one of the filter condensers of the
R,�70.000 0"'". 'IJ ... .�•• C.-.OOI ,./d. c�'.mj< co .. d.
R...-200 0",", '11 .. . .� •• C,-40 ,.Id.• '0 �. d�<. <o .. d. power supply for extra filtering and
R., a .. Ra. RI' R..-220.000 0",", '11 ....... CT·C.-40!JO I'ld.• 4'0 �. dc<. <0 .. 4. lowered power supply impedance.
a.-l000 0",", '11 ... ,� •• T,-O .. ,p .. , " ..... (A<'o.o .. .. d TO·JlO) The only other changes which need
R.-JOO 0",", 10 w • .,;r••o,,,.,1 ,�,. T...-Po ... , " ..... J40·0·J40 �. @ 120 m•• ; ,
�. 0 :1 .mp •. ; 6.1 �. c.,. • 2 ."'p"
be made involve the feedback resistor
R • .-'600 0" .... 1 ... " ••
au-lOO 0""', 10 ... ..;.� .. 0... 4 ,,,. ].-1 .. , .. , i.clt and feedback compensating co ndenser
C.-lOO ,./4.• U � .•IH. to.. 4. V,-ISL7 , .. b� which shunts this resistor (or in some
C.. C,.-.l ,./4. . 400 �. co.. 4. V•• Y,,--6Y6 , .. b� circuits bypasses it to ground). The
C.-IO ,./4 .• 400 �. �/.c. co.. 4. Y.-, YJ , .. b.
ratio of series resistor to shunt re-

44 RADIO & TELEVISION NEWS

sistor in the feedback path determines
both the lotal gain in the circuit and
the proportion of feedback. For exam­
ple, with'a 6SL7 phase inverter and
feedback ,rom the 16-ohm tap of the
Acrosoun4 TO-310 transformer, the
power amplifier will have 17 db of
feedback and require a maximum in·
put signal o f 3 volts to drive it to full
output when the ratio of feedback to
cathode resistance is 5 to 1. If the
ratio is changed to 7.5 to I, the am·
plifier will be driven with a 2 volt in·
put, but the feedback is cut down by 3
to 4 decibels. Similarly, a 12AX7 has Fiq. 5. Bottom view of converted Grommes amplifier showlnq new output tranaformer.
about 50% more·gain than a 6SL7. If
this tube is used with a 7.5 to 1 ratio is used, there is some possibility that version since the "Ultra-Linear" cir­
of resistance, the amplifier can be the amplifier response will peak in the cuit is less critical as to supply regula­
driven to full output with 2 ,'olts of ultrasonic region even though the re­ tion than the t('trode circuit, In the
signal while still maintaining 17 db sponse without feedback is fiat over a Grommes l00BA no filter choke is
of feedback. In the original construc­ very wide range. This peaking can be used, and the converted circuit works
tion, it is recommended that the eliminated with a consequent improve­ just as we\l without one although a
12AX7 be used so as to obtain this in­ ment in transient response, by adding single 100 ohm resistor was added in
creased sensitivity. However, in con: a network to change the phase of the converting in order to reduce the hum
verting an existing amplifier, the con· feedback voltage in the peaking voltage.
structor can leave the 6SL7 tube in the region. One simple arrangement is to The converted amplifier has extraor­
circuit and can adjust for the required add a sma\l condenser across the feed­ dinary specifications for its size and
sensitivity by varying the feedback re­ back resistor. A suitable condenser price, In fact its specs read amazing­
sistor. If necessary, he can sacrifice value in the type of circuit under dis­ ly like those of a com'entional triode
a portion of the feedback in order cussion is one which makes the prod­ WilIiamson amplifier. Frequency re­
to maintain sufficient gain for the pre­ uct of the feedback resistor in ohms sponse is fiat ± .5 db from 20 cps to
amplifier slages which are being used. and the condenser in microfarads over lOO kc. at a 1 watt level. (By
In many commercial amplifiers, the equal to unity. Several typical circuits increasing the size of the cathode con­
power amplifier section must be suffi­ using a 5 to 1 resistor proportion are denser of the 6SL7 the low end re­
ciently sensitive to be driven by 1 volt illustrated in Fig. 3. In these arrange­ sponse can be made fiat to below 5
of input because of the relatively low ments, the feedback connection is cps.) At 10 watts, response is fiat ± 1
gain of the earlier stages. If this is the brought to the cathode or pair of cath­ db from 20 cps to over 60 kc., and
case, it is necessary to diminish the odes of the phase inverter stage. All clean waveform is preserved from 20
feedback (by increasing the feedback of the arrangements have the same cps to 30 kc. even at this high a level.
resistor). However, the most modern pl"Oportion of feedback and the iden­ The transient response as evaluated
preamp designs are intended to supply tical phase correction. by square waves is shown in Fig. 6.
about a two volt input such as is found Thel'e is a minimum of transient dis­
on Williamson-type amplifiers. Any of
Circuit Conversion
tortion and phase shift at these two
these preamps will handle the converted These conversion considerations are extremes of the audio band.
6V6 amplifier and still permit 14 or exemplified in the conversion of the Intermodulation distortion is ex­
more db of feedback. This is sufficient G"ommes 100BA amplifier, Fig. 1, the tremely low. It runs about .1 % at 1
feedback to reduce distortion, hum, circuit of which is shown converted in watt, rises to .4% at 8 watts, and to
noise, and internal impedance to low Fig. 4. This amplifier is typical of . 5 ', k at 10 watts. It is still below 1 '70
values suitable for top quality applica­ many which come both with and with­ at 11 watts. These tests were made
tions. Thus the more common front­ out preamps in the $40 to $60 pl'ice with 40 and 7000 cps mixed 4 to 1 and
end arrangements will serve with the bracket. Both former values and con­ are based on equivalent sine-wave out­
"Ultra-Linear" 6V6 amplifier while verted ones are indicated on the put. This is the conventional method
preserving an adequate proportion of schematic. There are only three elec­ of rating which is used for practically
feedback. When the 12AX7 is used, the tronic parts changes in addition to the al\ commercial amplifier equipment.
designer has an additional 3 or 4 db of new output transformer. The quality of a low cost 6V6 am­
latitude in his choice of gain t;er�tts Physica\ly, it takes only two addi­ plifier is normally not up to the top
proportion of feedback. . tional holes for mounting the output high-fidelity standards which have
In some amplifiers which are of the transformer-the remaining holes line been set by the WilIiamson-type am­
public-address type rather than the up without alteration. The trans­ plifiers produced in recent years. How­
high-fidelity type, inadequate feedback former fits rather snugly but inas­ ever, it is now possible. by using the
is used which is limited to 6 db or less. much as it contributes no heating, its "Ultra-Linear" circuit arrangement
Conversion of these amplifiers with proximity to other parts causes no and a top quality output tJ'ansformer,
the increased feedback which results difficulty. to convert these run-of-the-mill ampli­
from a 5 to 1 resistor propol'tion will Any power supply which is satisfac­ fiers into ones whose quality is com­
pl'oduce insufficient gain. In those tory for the original cil'cuit is also pat'able with the best obtainable in the
cas('s, there must be either a sacrifice suitable for the "Ultra-Linear" con·
of feedback or the addition of more
gain in the early stages. However, in Fig. 6. Square·wave performance (A) at 20 cpa and (B) at 20 kc. See text for detalla.

these amplifiers the original quality is

genera\ly so poor that the substitutioll
of the "Ultra-Linear" output arrange·
ment will make a decided improve,
ment in performance even if bnly 6 db
of feedback is used. The relative im­
provement in a Jow grade amplifier is
even greater than is achieved by con­
verting a fairly g ood amplifier which
has a high proportion of feedback.
When feedback in excess of 12 db

J •••• 1954 45
10 to 15 watt power range. For many
people this power range is ample for
all home requirements.
Cal'eful listening tests have borne
out the justification for the "Ultra­
Linear" conversion. Partirularly in the
low frequency range there is substan­
tial improvement. The solidness and
clarity of the heavy bass passages
is a revelation when one contrasts old
and new amplifiers. The silkiness and
smoothness of the treble range also
stand out in a side-by-side comparison.
In short. the improvement in meas­
ured characteristics is confirmed and
substantiated by a corresponding im­
provement in listenability.

REFERENCES
1. lTa.fler, Davitl cf Ke,·oc8. ITerbcrt T.:
uAn ''''lfrt&-/�itl'·(Jr A1I",Ii/icr", Audio j')'UfJin­
cc,·jJlg. Novcnaber J !lS1
1. : hlml""fl1�hlg tile ll'ill;(lIl1!'UJJ&
A JI'}llijic,''', HADIO " T£L£nslo:-; :>;EWS, Fcll­
rlwrJ/ 1!JiiJ
 O'nr-all view of the "Xn!ght" 2S-wa\l
linear ampllfter_ It la o ff ered ID kit form
lor a••embly by home audio builder.

.
�\

� . ::::��;�;':'�'�':'.:.-;'.�
The underlying design concept was
to keep the distortion in the amplifier
as low as possible before the addition
of the negative feedback, the thought
behind this being that upon the addi­
tion of the single loop of negative
By feedback the inherent distortion would
virtually vanish. The input circuitry
NORMAN KRAMER was made as simple as possible with
Prolect Engr .. AlUed Radio CorporaUon the signal level kept at a minimum to
keep the distortion low. Two 12AU7
dual triodes are used as input, phase­
inverter, and driver stages and all of
Circuit details on a new li near amplifier kit which uses this circuitry is combined on an etched
circuit board. This insures that each
two 5881 beam power tubes and a special output transformer. unit is a photographic reproduction of
the engineering prototype and that
there is no possibility of additional
UDIO amplifiers in recent years' other words, variable damping which feedback loops which could degrade
have reached a practical limit of enables the listener to optimize the the amplifier through undesirable in­
A perfection. The present trend in source for any loudspeaker; (5) Linear creases or decreases of signal levels in
amplifier design is to reduce the cost frequency response throughout the en­ these stages. Furthermore, the printed
of such units while retaining the high tire range from 10 to 20,000 cps; and circuit board makes for ease of wiring
standards of frequency response and (6) Complete freedom from hum and and a layout of components critical to
undistorted output power. microphonics_ hum minimization which cannot readi­
The new "Knight" linear-deluxe There are numerous factors and de­ ly be achieved by conventional wiring
basic amplifier is one such unit since sign features which have to be consid­ methods. Circuit parameters and op­
the cost has been reduced by offering ered in meeting these requirements_ erating voltages are carefully adjusted
the unit in kit form. The design of Choice of circuitry influences distor­ to minimize IM and harmonic distor­
this amplifier is centered around the tion and output power as well as sta­ tion.
following requirements: (1) Negligible bility at both the high and low ends A great deal of care went into the
harmonic and frequency distortion as of the amplifier bandpass. The output design and selection of the output
well as negligible IM and phase dis­ transformer, which is actually the transformer used in this amplifier.
tortion; (2) Maximum output power heart of the amplifier, governs the IM Low distortion and wide bandpass in
rating far in excess of that required and phase distortion as well as fre. an output transformer are not enough
to reproduce the transient peaks found quency response and the power han­ to insure optimum performance. This
in music; (3) No tendencies toward dling capacity. Freedom from hum is bandpass must be free from any res­
low-frequency instability or ultrasonic most dependent on circuitry and com­ onances, must be smooth, and must
oscillations:(4) Negligible incremen­ ponent layout. Actually, all of these have a minimum phase shift over a
tal output impedance and a method for factors must be considered together very wide band. Therefore, such trans­
making this a variable in both the since many of them are interde­ former parameters as shunt induct­
positive and negative directions. In pendent. ance, leakag� inductance, and capac-

Fig. 1. Method of getting poalUve (A) and negative (B) current feedback. (C) Simplified diagram of the feedback circuit.

RA

NEGATIVE VOLTAGE FEEDBACK

R.

RC

IPOS.CURRENT
HED8ACKI
IAI
ICI

90 RADIO & TELEVISION NEWS

itance, and even tube types to be used
I'fIIlNT[O C'''CUtT BO&JtO
become factors to be considered in the
r------ --- - ------�;.:..-- - - --- ---- - 1
choice of an output transformer for
a given amplifier circuit. !
I )'�
t------il---:�-;,====��
The power output stage consists of I
I
two 5881 beam tetrodes and the out­ I

put transformer just described. The "",""TJ.O'�.

5881 tubes were selected because of
their special grid construction and their
excellent characteristics of higher per­
missible plate dissipation and rugged­
ized construction. These factors make
for very stable operation. The output
transformer has taps provided on the
plate windings to which the screens
·
of the 5881 tubes are con n ected. Thus,
the screens as well as the plates are
.471\. .!fA
energized with both a.c. and d.c. and
their currents are combined in such a
way as to "linearize" the dynamic
plate characteristics of the tube. In
this way the efficiency and power
sensitivity of the tetrode are pre­
served and the low internal imped­
ance of the triode is maintained.
The secondary of the output trans­
former not only supplies the single Si'lnai input ahou!d b. 1.1 yoll. \0
ob:ain the max:muDl power output_
loop of 20 db negative voltage feed­
back to the input cathode, but is also
O,,'pu, ",," •.-7000 ohm. 10 4, 8, 16 01,,",. For '881 OP�rt";O". pri. I.p. for .cr�"'''' m •••
brought to ground through a resistive .."J;J'o,'�J po., .. , OUlp'''' 30 .."lIs, Ir�q ...."cy re'potJu ±Vl db, 10·120,000 cp ••
network which includes the variable I'owrr ".",.-'00·0·'00 �. @ 160 m •. ; , �. @ 3 .mp •. ; 6.3 �. @ ., .mp •.

damping control. Amplifier damping Fillrr clooltr-6 Ioy. @ 160 m•• , d.e. ,r.i.'.nu 160-170 010,",

factor is defined as the ratio of nomi­

nal load impedance to actual output Flq. 2. Schematic of "Xnlqht" 2S-wal\ amplifier which I. a... allable lA ltIt form. Porta ... alu ••
impedance. It is apparent that ampli­ ar. .hown on dlaqram, .xcept for .p.clal part. who.. .p.clfication. ar. vl....n abe.....

fier damping factor can best be con­

t.rolled by changing the actual output The damping factor is continuously and jack arrangement is provided for
impedance since the load impedance is variable from +35 to -4 with a nor­ the metering of cathode currents and
fixed by the speaker system used. mal damping factor (network shorted a bias adjustment can be made for
The most effective method of chang­ out of the circuit) of +16. balance.
ing the output impedance is by a com­ Fig. 2 shows the complete schematic Frequency response is ± % db from
bination of voltage and current feed­ diagram of the amplifier. A 5V4G is 10 to 120,000 cps. Harmonic distortion
back. Fig. lA shows a simplified used as a full-wave rectifier with a is .15% at 25 watts output. IM distor­
diagram of one method of achieving pie-type LC filter to supply the output tion (60 and 7000 cps, mixed 1:4)
positive current feedback. RA is the stages. Additional RC filtering and de­ measures 1 rl<. at 20 watts and is less
voltage feedback resistor and Rr. is the coupling are used in each of the pre­ than .25% below the 18 watt output
cathode resistance of the stage receiv­ vious input stages. All the power level. Output taps are provided at 4.
ing the feedback. R� is the current supply ratings are sufficiently high so 8, and 16 ohms.
feedback resistor and is in series with that power may be taken from the This amplifier is being marketed in
the load. Current through R. produces amplifier to supply a remote preamp kit form by Allied Ra.dio Corpol'Utioll.
a feedback, proportional to a positive if necessary. 100 N. Western Ave., Chicago SO, Ill.,
current, which decreases amplifier out­ Optimum performance requires that (or $41.50. It comes complete with
put impedance. Fig. 1 B shows a some­ the output tubes operate in a bal­ full instructions for assembling th"
what different arrangement whereby anced condition. Therefore, a switch unit.
the phase of the feedback through R,·,
the current feedback resistor, is op­ Under chass!. ylew of a •• embled ampl:fie:. Not. u •• of prlnt.d circuit board (top).
posite and is now proportional to a
negative current. In this way, ampli­
fier output impedance is increased.
Fig. 1C is a simplified schematic of
the feedback circuit actually used in
this amplifier. The two CUlTent feed­
back resistors, RN and Rc, are shunted
by RIO, the damping factor control po­
tentiometer, which permits a continu­
ous control over both negative and
positive feedback. Furthermore, there
is a point where amplifier output im­
pedance approaches zero and th:­
damping factor approaches infinity.
This control is capable of adjustinJ:: the
damping factor between fAirly wide
limits at the lower frequencies to as­
sure optimum damping of virtually any
speaker and enclosure resonanees. In
this way, the listener may adjust the
amplifier damping factor to achieve
optimum damping for his own system.

April. 1956 91
 ANY high fidelity amplifier circuits

M have been made available in the

low and medium power brackets.
O.,er·all .,lew of Ibe 60·watt "Ultra·LlDear"
amplifier. AD ACro.GUDd T0-330 !rODS­
However, there has been very little former la u led 01 Ibe output IlGDdormer unll.
activity in development and public dis­
semination of high powered amplifier
circuits in the power range over 50
watts.
There are many applications in
which high powered, high fidelity am­ Construction detaiLs on a commerciaLLy-designed, high-power
plifier capabilities are desirable or
even mandatory. In addition to non­ ampLifier for the serious audiophiLe. It uses four KT66's.
sound reproducing functions in labora­
tory and industrial applications, high
powered audio units find application it operates at only a fraction of its -approximately 10 audible steps of dif­
in recording work, in auditorium and potentialities and
' is rarely pushed to ference. The extremes are not as far
amphitheater work, and ,in systems the limit. apart as it appears on the surface, and
where power is divided over many dif­ On the other hand, many experi­ it is probable that an intermediate
ferent speakers in different rooms or ments have been made to demonstrate rating will serve the needs of the most
areas. There is also a school of that only a few watts are needed in critical home installations. However,
thought which believes that high home installations. This is generally it is difficult to achieve intermediate
power is required in home high fidelity done by measuring the average power power levels of 50 watts or more using
installations for the ultimate in realis­ used in a specific installation and al­ conventional parts and conventional
tic reproduction. lowing for an estimated ratio of peak circuitry.
The necessity for high power in to average power. This type of test
indicates the customary listening level
Approaches to High Power
home use has been doubted by some
,
authorities, but its proponents have at a specific audio installation. It does There are many designs available for
several valid arguments which must not actually indicate how much power powers up to about 25 watts. When
be given some weight. They claim would be required to simulate the this point is passed, ol'dinary operat­
that high power is required for proper l'ealistic auditory loudness of live ing conditions and circuits are not di­
reproduction of musical transients music. Few people are interested in rectly applicable. When it is desired
since transient waveforms require reproducing the full acoustic power of to exceed 50 watts of clean power and
more power than the sine waves which an orchestra in their living rooms. still preserve high quality, the prob­
are used as the basis for conventional However, many want to have the same lems of design are rather acute. There
amplifier ratings. For example, repro­ sound pressure at their ears that they are several possible approaches which
duction of a square wave of given am­ would get in the concert hall. Any­ warrant consideration:
plitude requires twice the power of a body who has listened to a full scale 1. Cl<j,,�s B (overbiased operation)
sine wave of the same amplitude. symphonic orchestra, or even a large provides high efficiency and delivers
Another cogent argument for high dance band, is immediately aware that high power at low operating cost. How­
power is that amplifiers have to work it takes tremendous power to dupli­ ever, overbiased operation causes ex­
into loudspeaker loads of widely vary­ cate the same sound intensity even in cessive distortion at lower levels even
ing impedance. At the bass resonant a smaller room where power require· though high level distortion can be
frequency a loudspeaker might show an ments 'a re far below those of the con, made relatively low. A Class B ampli­
impedance of many times the nominal cert hall. fier with 1 per-cent distortion at 50
value. No amplifier can deliver its If the goal of high fidelity is "con­ watts may have 1h per-cent at 1 watt­
rated undistorted power into a load cert hall realism in the home" as many an excessive amount for high fidelity
which represents such a severe mis­ people agree, then a lot of power is re­ standards. Class B amplifiers also
match. A wide reserve of power ca­ quired. Estimates of requirements run have inherently poor regulation, and
pability is thus required to maintain extremely high. One published figure 1 careful attent ion must be paid to the
high fidelity operation under normal calls for 100 watts of clean power. In design of power and bias voltage sup­
listening conditions. The argument, contrast, estimates, based on average plies or peaks of operation will shift
which always is used as a "clincher" usage generally run about 10 watts. voltages and operating conditions caus­
for high power, claims that like a high Superficially, this divergence of opinion' ing transient distortion.
powered car, a high powered ampli­ looks extreme. However, the range of 2. Class AB. op'eration provides high
fier is smoother at all levels because 10 to 100 watts is only 10 decibels wide power by driving the output stage into

Febr.ary. 1955 CS
 EffiCiency is of minor importance un­
less portable operation Is required
where power Is costly.
The choice of type of output stage
to parallel is restricted. There are no
triode stages with conventional receiv­
ing tubes which provide sufficient out­
put. Tetrode stages have excessive low
level distortion, are very sensitive to
impedance mismatch and, therefore,
are subject to nonlinearity using
speaker loads since these have a re­
active component and an impedance
which varies with frequency. The
choices left are the Ultra-Linear stage
or the tetrode with local feedback such
as is achieved with a cathode winding
on the output transformer. This latter
circuit has had a recent revival of pop­
ularity after favorable comment by
Williamson.
Obviously, the use of local feedback
will imprpve the characteristics of the
output stage and seems to be an at­
tractive arrangement. However, there
is a penalty attached to the use of
feedback. Sensitivity is reduced in
proportion to the amount of feed­
back. This puts a severe limitation on
the use of feedback over an output
lIr-l0,OOO oft"" Yl ... .. .. C" C_20 "Id., 4'0 ... elec. c.,..<i/or. stage alone. In the circuitry under con­
R..--l ,"�&Ol".. , '11 w. r�J. C_JO "Id., 4'0 .. . elu. c.,..ci/or. sideration, the driver stage must sup·
1I�7,OOO oft"" 1 ... ,. . . C.-82 ""Id. mic. C.,.'l<i/or.
ply about 30 volts r.m.s. to each output
II_JJ,OOO oft"" 1 .. . ,�•• . C" C" C" C..-.2' I/old., 400 ... c.,..ci,o,.
. grid. If 6 db of local feedback is used
11., 11--'70 oft"" % ... ,.•. C_2'0 "Id., '0 ... elec. c.,.«i,or.
R., R�22 ,OOO olem, 1 w. re'. (mlftched "tli,) CII, C,r-16 "Id., 600 " . elu. c.,..ci,or. I
on the output stage, the driving re­
11.-12,000 oft... , Yl ... 'e'. Tr-Po .. e, 'r." •., 400·0·400 .. . @ JOO "' .. . quirement is 60 volts r.m.s., or about
11_22,000 oft... , 1 ... ,e •. , ... @ f .m,. •. , 6.J ... @ 7 .", ,. •. (Tftord.,·
88 volts peak, for each grid. This
11", 1I,�70,OOO oft"" % ... re •. .0" No. 2211J' U" b� ,,,�4).
11", 11,_220 oft"" Yl ... re •• T_Ow'"w, I,"" •., (Acro SON"d TO.JJO). brings the driver stage to the point
·
RI" Rl....7.-4 ,OOO ohm, 2 Wo re,. (mfllched ".i,j CHr-Filler cftolu, f ft.,., 2'0 m•. , 60 oftm •. where its distortion becomes signifi­
1111, 11,_100,000 ohm, Yl ... 'e'. (5,""co, No. Clf12 U" H .. ,ed). cant.
R", RUt R ... R,.-1000 ohm, Vz w. ,r,. 5...-5.,. .•. ,. /o"le ...i,cft.
Good design calls for integrated per·
Ru--Z'OO o"m. , Wo .,..;rewound '�J. F�L;"e /r.ue. , tlmp,.
Ru---l'O olrm. 20 w . • i,eJt'o"nJ r(! •. V" V.-65N7GT , .. bc. formance of an amplifier, and the driv­
Ru, R... Ru, R.r-27 ohm, Yl w. re,. V" V .. V" V.-KT66 , .. b�. er and output stage must be con­
Rn-lOO ohm .,ire.,ounJ pot. V" V..-'V4G , .. be.
sidered as a unit. On this basis, a
conventional type of RC coupled driv­
Complete schematic diaqram of the Acro·deaiqned 60·wall "Ultra·Llnear" ampliBer. er combined with an Ultra-Linear out­
put stage will give less distortion than
the grid current region. The driving eventually _ Present efforts have not the same driver and a tetrode stage
stage must be able to deliver power at surmounted all the problems of using with cathode feedback. Therefore.
low distortion into the output grids. these tubes in reasonably-priced, effi­ Ultra-Linear operation has been select­
These grids present a low impedance cient circuitry. ed for the. paralleled output stage.
when they are driven positive, and dis­ 5. Parallel operation of tubes is rela­ The justification of this choice is
tortion is created in the driving stage tively inefficient and requires careful shown by the extremely low distortion
unless transformer coupling or other design to avoid parasitic oscillations. characteristics of the push-pull paral­
involved and expensive circuitry is uti­ However, there can hardly be a better lel amplifier.
lized. This mode of operation also has way of getting high power than by Experience with paralleled output
poor inherent regulation as the Class taking a stage of desirable character­ tubes has indicated that some parasitic
B stage with consequent problems of istics and paralleling it with another oscillation is probable unless "stop­
power supply design. Lastly, it utilizes stage. Williamson mentions using suf­ pers" are used for partial isolation of
a lowered plate to plate impedance to ficient pairs of output tubes in his paralleled elements. The present de­
get higher output, and this creates famous circuit to ach ieve 70 watts of sign, therefore, uses grid stoppers of
high distortion at low levels similar output. The inefficiency of this ar­ 1000 ohms each and plate stoppers of
to what happens with a Class B stage. rangement renders it rather impracti­ 27 ohms each. These eliminate the
3. High voltage transmitting tubes cal; but if the basic stage is efficient usual· difficulties encountered when
can be used for high power output. and can do better than 25 watts, par­ tubes are paralleled.
However, the associated voltages are allel operation with two pairs of tubes The four tube output stage requires
in the lethal range of 1000 volts or will produce the desired output power the same signal drive at the grids as is
over, which is definitely undesirable. ahd will still be practical. needed in a two tube circuit. There­
In addition, the tubes and necessary fore, a push-pull driver stage similar
power supply make the amplifier quite
Circuit Considerations to that of the WilIiamson circuit will
costly. Consideration of the various alterna­ handle the drive requirements ade­
4. There ar.e "hot" tube.� on the tives for getting high power and high quately and with low distortion. This,
market such as the RCA 6146 (and fidelity simultaneously points to the in turn, can be preceded by any high
the Tung-Sol (>55V which will be avail· paralleled stage as being the simplest quality phase inverter. Again the WiI­
able shortly) which can provide high and most foolproof and bug free. It is liamson arrangement of voltage ampli­
power at reasonable supply voltages. not efficient, but this means merely fier directly coupled to the ca thodyne
Experiments with these tubes for high that a slightly higher drain is put on inverter is satisfactory.
fidelity circuits are being carried out. the electric meter than would exist The basic tube alignment described
and rcsults w ill be made available with Class B or Class AB, operation. must be powered by a fairly heavy
power supply. The use of four output
tubes doubles the normal current con­ AMPLIFIEB SPECIF1CAnOIfS
sumption of the Ultra-Linear output POWER RAtING . . . • • • • • • • • • • • . . . 80 watta at Ilot oftr 1 per<ellt IM dlatortloll
stage, and the B+ requirement Is SENSITIVITY . . . . . . . . . . . • • • • •• • • . • 1.3 .01.. r.m.a. for 10 watt8 output

FREQUENCY RESPONSE • • • • • • • • • • @ 1 watt piu. or mlIlua 1 db: 2 cpe 10 200 kc.

about 250 ma. To exceed 50 watts of
clean DutPUt, the B+ vDltage should @ 20 watla plua or mlIlua 1 db: 10 cpe to 100 kc:.
be in excess Df 425 vDlts. A 300 ma.
@ 80 walla plua or mlIlu 1 db: 20 eps to 20 kc.
DAMPING FACTOR . . . . . . ..... . _ . 18
transfDrmer Df 400 VDlt rating, alDng
with two 5V4G's paralleled, and ca­
pacitDr input can supply the amplifier
2 ...
\\·ith the desired requirements. There sign at IDW frequencies. If there are
two. RC stages and the output trans­
!
is no. need fDr chDke input Dr Dther
means Df imprDving power supply reg­
ulatiDn because the amplifier has good
fDrmer included in the feedback loop
(as } s true in the WilIiamson type Df
circuit), there are three SDurces Df IDW
I
.1
...
/
J
inherent regulatiDn with practically no.
change in current drain as the DUtPUt frequency phase shift_ If each Df these
level is varied. This good regulation is has the same time cDnstant, the 180 " .5'"
a very desirable characteristic Df the degree point is reached when each Df --�
.%5'"
amplifier as it Dbviates the possibility these three SDurces cDntributes 60 de­
Df transient distDrtiDn caused by mD­ grees of phase shift. In this hypotheti­ o 10 20 so 40 50 10 70
WATTS OUTPUT (EOUIVALENT SINE WAVE POWElU
mentary changes in tube Dperating cal case, the gain has decreased 18 db
InlermodulaUon dl.lortlon •• power oul­
conditiDns. It makes the use Df cathDde at the critical phase shift point, since
pul (40 and 60 cp. mixed at ratio4:1)
hias practical, just as with Class A tri­ 60 degrees of phase shift correspDnds
ode stages, since the cDnstancy Df to. 6 db Df gain reductiDn in a single tinuously or during some intervals of
cc:.thDde current in the DUtPUt tubes stage, and there are three such stages use. This is certainly an undesirable
maintains a cDrresponding cDnstancy invDlved. Thus 18 db is the limit to. the condition which caused many WilIiam­
of bias, with a fixed drop acrDSS the permissible feedback if instability Is to. son amplifiers to. have lDW frequency
.
cathode resistDr. be aVDided. instability.
The final element in the design Df HDW can 20 db of feedback be run if When the Ultra-Linear version Df
the amplifier, and a very important the amplifier gDes unstable with 18 the WilliamsDn type circuit was de­
element, is the feedback circuit. In a db? The answer is that the phase shift signed, this problem Df low frequency
drcuit of the quality sDught, it is de­ must be reduced_ This can be aCCDm­ stability was investigated, and steps
sirable to. utilize about 20 db Df nega­ plished by introducing phase shifting were taken to extend the margin Df
tive feedback in Drder to. prDvide the networks Dr by staggering the time stability. This was done by 8taggering
\,·ell knDwn advantages Df IDwered dis­ constants Df the variDus stages so that the interstage RC networks by length­
tortiDn,reduced hum and nDise, and a given gain reductiDn is accompanied ening Dne Df them. The same expe­
imprDved damping Df the loudspeaker. by less phase shift. went has prDven beneficial in the
However, feedback must be integrated The IDW frequency stability Df the push-pull parallel version Df this cir­
into the design and cannot be added original Williamson circuit was de­ cuit and has, therefore, been included
haphazardly Dr indiscriminately with pendent Dn two. RC netwDrks with in the design.
the expectatiDn of experiencing no dif­ equal time CDnstants and the Dutput The high frequency stability prDb­
ficulty. It is probable that the lInin­ transfDrmer time CDnstant. When the iem is not as simple to analyze be­
formed use of feedback will result in inductance Df the transfDrmer made its cause of the fact that the performance
the degraded performance Df most time CDnstant equal to. that Df the of the output transformer at high fre­
circuits since the feedback may nDt be interstage netwDrks, 20 db of feedback quencies is more complex than that of
stable, and regeneration rather than cDuld nDt be accomplished with the de­ a simple RC network. As a general
degeneration will be experienced_ This sired degree of stability. Since the rule, the amount of feedback obtain­
can cause increased distortion, hang­ transformer's time constant is variable able at high frequencies for a given
over, boom, and screechy effects depending on the d.c. balance of the circuit is directly dependent on the
caused by high frequency ringing. output tubes, a.c. excitation level of quality Df the output transformer
Even if the amplifier is stable under the Dutput stage, variability in core which must have both low and smooth
quiescent cDnditiDns, it is possible fDr material, production quality control, phase characteristics over a very wide
instability to occur momentarily under etc., there was always the possibility bandpass. Many criticisms leveled at
SDme operating conditions. This tran­ of running into instability either con- feedback circuits made condemnations
sient instability is one Df the factDrs
which makes for discrepancy between Undercha••I. ylew 01 the 60-wall ampUner. Part. layout ellmlnale. hum problem ..
the measured performance of an am­
plifier and its listening characteristics.
They can measure well and not sound
well.

Considerations of Feedback Stability

In any feedback amplifier. regenera­
tion will Dccur if the phase shift
reaches 180 degrees before the ampli­
fier gain has dropped by an amount
equal to. the amount of feedback. For
example, if it is desired to use 20 db of
feedback, the 180 degree phase shift
points at high and low frequencies
should not be reached until the gain
of the amplifier without feedback has
decreased by at least 20 db. In other
words, phase shift must be kept down
until the normal roll-off of the stages
has taken effect.
A quick appreciatiDn Df the design
prDblems invDlved can be had by Db­
serving the prDblems Df feedback de-

hbnciry. 1951 47
 which are not the inherent faults of these amplifiers at 24 watt s is .14 per­
the circuits, but are due to the fact cent (based on a 40 cps and 7 kc. sig­
that the optimum transformer was not nal mixed 4:1). With the mismatched
used. . condition, the total power output· is
In this push-pull parallel design, cir­ reduced to less than 40 watts, but the
cuit considerations and transformer IM at 24 watts is .07 per-cent-a re­
design have been integrated to provide duction to one-half. Such low values
a circuit ot excellent stability-a cir­ of distort i on exceed the capabilities ot
cuit in which the feedback could be most measuring equipment. .
increased by 6 db without encounter­ The amplifier passes more than 50
ing instability either on resistive or watts of power without visible distor­
speaker load. tion on the scope at any frequency
, However, some combinations of load­ from 20 cps to 30 kc. Frequency re­
I ing can lead to ringing or other tran ­ sponse, of course, is far greater than
, sient disturbances under conditions of this power curve indicates; and be­
high level operation. This can be pre ­ cause normal frequency response is
vented completely by the addition of a plus or minus 1 db from 2 cps to 200
250 I'l'fd. capacitor across one-half of kc., the square wave transmission at
the ou tput transformer primary (from any frequency from 20 cps to 20 kc. is
blue-white lead to red lead). The need excellent.
for this expedient is the exception The amplifier requires only 1.3 volts
rather than the rule. to drive to full output. This makes it
Special circuit considerations such practical for use with any preamplifier,
as staggering at the low end and the as any commercial unit can supply the
introduction of local current feedback necessary Signal voltage.
by splitting the driver stage cathodes The combination of attributes avail­
in order to increase stability at high able in this push-pull parallel design
frequencies have been incorporated in is unique. It offers more power, at low­
this design. In addition, the output er distortion, over a wider bandpass
transformer was designed to have a than can be obtained by any conven­
bandpass l1at plus or minus 1 db from tional circuits.
10 cps to 100 kc. with a smooth phase It is reali zed that measured per­
characteristic tailored to the phase formance does not make an amplifier
characteristic of the amplil1er. The sound good. Of course, poor measure­
transformer, the Acrosound T O -330, ments generally indicate poor listening
also was designed to provide high pow­ quality; but the converse is not true.
er at low distortion over a very wide Therefore, since measurements corre­
band. The transformer combinations of late imperfectly with listening quali­
bandwidth, preservation of operating ties, it is necessary to make the listen­
characteristics at all power levels, and ing test the final and most important
maintenance of push-pull balance over test of merit. The perfonnance of the
a very wide band all combine with the push-pull parallel circuit remains su­
basic circuit configuration to give good pe ri or on listening also.
stability and unusually outstanding Heavy bass passages have better
perfonnance characteristics. definition and sound better damped on
The push-pull parallel circuit pro­ the push-pull p arall el amplifier than
vides better results than would be an­ on others. This is due apparently to
ticipated on the basis of doubling the more than the additional power capa­
output stage capabilities. Since the bility alone. It seems to be also due to
TO-330 transfonner is more efficient the high damping factor of 16, the fact
than the lower powered unit designed that varying speaker impedances do
for a single pair of output tubes, more not pull the power capability down to
than twice as much power is available below a satisfactory level, and to the
from the push-pull parallel circuit fact that distortion is at as low values
than from the regular push-pull ampli­ as have ever been achieved outside the
fier using the same design criteria. laboratory.
Thus, it is practical to reach levels in In the middle and upper frequency
excess of 60 watts before the inter ­ regions, the absence of intermodulation
modulation distortion reaches 1 per­ effects and the undistorted reproduc­
cent-if lour norm a l quality KT66 tion of transient signals make for a
tubes are used. With selected tubes, sm oothness and clarity which is appar­
the IM can be kept below . 5 per-cent ent in less listener fati g u e." After con ­
at 60 watts. Conventional commercial tinued exposure to the quality of the
advertising of the power capability oC push-pull p ara llel circuit, it is difficult
the amplifier would classify it as a 65 to listen to other circuits without be­
or 70 watt amplifier with a specifica­ coming aware of their previously un ­
tion of "peak power in excess of 100 noticed shortcomings.
watts."
As with all amplifiers, a mismatch BIBLIOGRAPHY
of output impedance such as putting a 1. Sarser and Sprinkle, "T"e Maestro­
A l'ower Amplifier," Audio EllgineerlllO,
64 ohm load on the 16 ohm tap to sim­ November 1951.
ulate speaker behavior at the bass res­ ll. WilIiamson, D. T. N., ",1 High Qualitll
A mplijier Desion," W i r e les" World, S ept e m ­
onant frequency results in a decrease ber 1'17.
of power capability. However, this also J. lVilliam.!on, D. T. N., "Amplifierll (lnd
Supel"l(ltives," IVlreles" World, September
results in a decrease in distortion. For 1952.
example, under matched impedance 4. Ha/fer, David, "Improving the WiI l iam­
conditions, the IM distortion of one of son Amplifier," HADIO & TELEVISIOS' N.,;ws,
f'cun",,.y J95J.
 .....::::::::::::::::::::::::::.:.:.

the ultr�,f'n'ea�r

All of the workinq

controla. Jacks. and pluqs
are mounted on the front panel

A new 60-watt power amplifier kit that employs an

unusual feedback circuit with exceptional stability.

E HAVE TESTED quite a large

Wnumber of high-power amplifiers
ly high capacities, usually due to long
lines or extreme variations of voice­
and presented our reports in past is­ coil impedances with frequency, result
sues of RADIO & TV NEWS. As a gen­ in phase shif ts that may cause oscilla­
eral rule. we have found most tion. Although this is a major problem
amplifiers in this classification quite in all feedback amplifier designs, it is
similar as far as performance is con­ not a serious one. Most ampl ifie rs on
cerned. specifically regarding harmonic the market today have a sufficiently
and intermodulation distortion. Vari­ large stability factor to a point that,
ations involved mostly appearance, if they are used under reasonable con·
special features, and, of course, cost.. ditions, they will present no problems.
The results obtained from the "Ul­ On the other hand, should anyone dc·
tra-Linear H," whieh is being mar­ sire to go to the extreme and, (or
keted in kit form by Acro Products example, use an electrostatic speaker
Company of Philadelphia, are similar which is basically capacitive in load,
in many respects to the other ampli­ some power amplifiers would encounter
fiers tested, but the design incorpo­ difficulties.
rates a novel method of obtaining The new Acro method, which has
feedback. It is a new method and its been called hy br id feedback. does
" "

greatest advantage is that it provides not employ feedback directly from the
an especially good stability factor. The secondary winding of the output trans·
manufacturer believes that the ad­ former. It does incorporate a sl'parate
vantages are so great that many fu­ wi ndin g which is tightly coupled to
ture designs from other companies both primary and secondary windings
may incorporate this technique. and is connected to the secondary in
Almost all recent amplifiers use a series a iding. It is this winding t h at
-

voltage feedback loop from the sec­ supplies all the necessary voltage and
ondary of the output transformer back current feedback. The diagram clear­
to the first stage of the power ampli­ ly shows how this particular type of
fier. In these designs the degree o f feedback is used across all three
stability i s dependent upon the type of st ages. It is of considerable interest
load connected to the output. Ex t r eme- to note that th is new design, since it

The total time required to

build the kit I. approxi.
mately 2 '12 h ours. The
conatrucUon I. e x tre m .ly
, slmpl. sine. a printed
, board of basically Ihe en·
tlr. circuit I. fumlahed.
All that la r.ally left 10
the constructor la Ih. wlr·
Inq of both the Ir cmaform ·

.r. and the controla mount·

ed on the front panel.

RADIO & TV NEWS

 leee'/lI •• ' ICCUII''''''

•
ao

)0 10

Note: • f)e�iJ(nah'K Numbered

f:yel�tM un I)rinted cir­
cuit hoard.

Diaqram of the 60·wall pow.r ampllfi.r. Not. particularly the feedback loop.

does not tie in directly with the sec­ change and a dv ise s us that al l future 2',;, h armon ic distortion is the maxi­
ondary winding of the transformer, kits will incorporate this modification. mum tha t should be permitted for
is n ot readily affected by variations In addition, a us eful suggestion was ideal performance. With this assump­
in l oa ds. It also adapts itself very passed on to us. To el imina te any pos­ Iion , the a m pl i fi er output is 63 watts
nicely to a design incorporating a sible squeal in the speaker which ma y at 20 cps and 5 1 watts at 20,000 cps.
variable damping control. We have occur w ith input plug disconn ecte d and This l atter figure is somewhat lowe !"
checked the per form ance of this unit amp l ifier turned on or off, solder a Ihan ra te d output. However, we have
an d can certainly say tha t it is e x­ I-megohm resistor from input jack con­ Ilot as yet found any amplifier which
tremely stable. Unfortunately, we have tact to its own ground termi nal . Ill('els published specifications for this
not tested other power amplifiers to With these changes wc found that the part icu lar condition. Onc poin t worthy
the degree where we could give com­ hum a nd noise were now down to 71 db of comment .·C'ganlillg the tests is that
para t iv e figures. The manufacturer from 2 w a tts output with op en-inp ut Ihey were taken with an �-ohm outp u t
claims tha t an electrostatic speakcr circui t. load. Were our tests made with 16
was trie d with this power amplifier Frequellcy Rc,�polI.�e: Tak en at 2 ohms instC'ad. all performance char­
and performance was excellen t. watts level is ::!:: 1 db f rom 7 to 52,000 acteristics listed he.·C'in would, w i thout
cps. At maximum 60-watt output, the a douht, show considerable im prov e­

Lab Report frequency response is flat w ith i n ::!:: 1 nH'nt. This is not unusual and it applies
db from 9 t o 26,000 cps. to a l l power amplifiers with th e e x­
Sensitivity: 1.3 volts input required I M DiMortion: (60 and 6000 cps, 4 ce pt i on of tho s e designs that have sep­
for maximum 60-walt output. to 1 ratio). For an equivalent sine­ :,.·ate fpedback loops for each of the
Hum and No i .�e : Down 56 db from 2 wave output of 2 watts, thC' IM distor­ output taps.
watts output. open-circuit input. For tion is 12 ',:,. At 60 w atts the distor­
. . Damping Factor: The damping fac­
sho r te d inp ut , the hum and noise is tion is 7 1 'fr. T his is wel1 within our
. tor v ar ie s from .35 to 7.9 with the
down 70 db from 2 watts. requirement of It;,. permissible dis­ damping co ntrol variC'd from min im um
The -70 d b hum and noise figure is to.·tion. In fact, the l�; fi gure appea.·s 10 maximum. Th e maximum figure, of

as good as they come. While the open­ at an equivalent sine-wave output of cou.·se, is obtained with the da m ping

circuit value of -56 db is not of th e 64.68 watts. control in "01T" position. 'fh e manu­
bes t, it is not customary for the us er 1I(o'7Iwnic Distortion: At 2 \valts lact III"e1"S fi � u l"es on the damping fac­
to operate the pow er amplifier under output, .09�;' at 20 cps; .08% at 1000 tor, as published and as print ed di­

open -clrc uit conditions. Improvement cps; an d 1 'f a t 20 kc. We believe that
.
r rectly on the chassis, show a possib le
can be made and for those who already \ariation from .5 to 15: Th ese are
have a unit of this type it is not a diffi­ ha sed on the assumption that a 16-
Printed circuit board
cult task to make the c hanges. ohll1 speaker load is used. We w ould
furnished with the kit
Remove the red-white and green­ :.ctually have lik e d to have seen a
Is aupplied with all
black wires from the ground lug to compoDe n t a mounted
maximum d.L of 15 with an 8-ohm
which they are connected. Run them and wired In the circuit. load as well. However, th e dilTerence
through the power transformer cable 11"0111 wh at was o btain ed is not !"eaHy
clamp and attach to another ground "f too great s i gni fi ca nc e in actual
lug at the n eares t printed circuit board pract ice.
mounting screw. A new soldering lug
should be used a t this point and it
should be soldered to the ground bus
o f the printed circuit board This ground
.

bus is a broad band of metallic mate­

rial on the bord er of the printed cir­
cui t board nearest the ground lug. The
manufacturer has recommended this

RADIO & TV NEWS

 V!l

Ov.,·all vi.w of the 4O.watt

of hi·fi amplifi.r built by a
I�T
, -' :-:. ..
!.�.!'·..vZ
��" �. ", / �;� �It_� :
r ..�1.:·f' I��
SI R20 ...
:'7:':-

40- .O;'·W�tt'··'H ,i :Fi':; m'plifier

�

or ":..-,(_ ...

with'l'ertia· r)' . Feedback

,.. ... ·r. _

By THOMAS F. BURROUGHS
/ ;�·:t�· '. . . .. ,' -. j ,
. . �., . ' _

Complete 'co�nstruction details on a stable. hiCJh-power

audio amplifier w ith output-staCJe cathode feedback.

OR years the most popular power nected to the screens of the output, tive in reducing distortion and helps to
amplifier circuit design was based on thus.providing the inherent advantages lower the internal impedance of the
the famous "Williamson." Some of which have been adequately covered in amplifier.
I he factors leading to its popularity are the literature. While the distortion of present-day
its low distortion, simplici ty . straight­ Some interest has been shown in amplifiers is very low when measuring
forward design, and the ri s'i ng interest tertiary feedback. which has been a sine wave, it is only of late that at­
in high-fidelity reproduction. Even to­ called "super Ultra-Linear." Up to this tention has centered on the stability
day a "Williamson" amplifier that is time, there have been very few circuits of the amplifier. In a feedback am­
carefully constructed and tested with published using this type of feedback, plifier instability will occur when the
the speaker with which it is to be used but it is believed that its popularity feedback beeomes regenerative. This
will give top quality results. will grow as more transformers be­ regeneration takes place when the
Of late the trend in amplifier design come available. Some of the ad­ phase shift reaches 180 degrees at some
has been toward higher power and, as vantages of tp.rtiary feedback are the point where the gain of the amplifier
the trend has developed, use of the lack of prob!ems involving instability­ has not dropped by an amount equal
"Ultra-Linear" output stage has grown which may occur when feedback is to the feedback. Sometimes instability
with it. In this circuit the primary taken around three or four stages. In will appear when a transient signal is
winding of the output transformer is fact. an amplifier with tertiary feed­ fed to the amplifier. This transient in­
lapped at approximately 40 per·cent of back seems to be more tolerant of an stability is one of the reasons why an
the primary turns. These taps are con- over-all feedback loop. It is very e/Tee- amplifier often shows up on te�t in-

20 CPS. I KC. 15 KC. I KC. 15 KC.

60 WATTS 60 WATTS 60 WATTS I WATT , WATT

o
. �
.J

o
z

<r
�

'\ A��. "

Cl)
-0
�c(
erO
...J

V :V
Fig. I. Sin. and squa,. wav.fo,ms und., va,iou. condition •.
Not. that the .in. wav •• w•• ,. all obtained at the full powe,
output of the amplifi.r. whil) the 'qua,e wav •• we'e at I watt.
struments a lot better than It actually The first problem which arose was from the plates are almost perfectly
sounds. the high signal needed to drive the balanced and the impedance looking
6550s when. using tertiary feedback. back into the tube is the same. This
Circuit Used
The required signal is 135 volts r.m.s. stage is quite degenerative and very
The circuit of Fig. 2 Is a very stable grid-to-grid. The need for this high little distortion is introduced by it.
40-watt amplifier (or a 6O-watt ampli­ signal arises because the voltage must The EF86 is a low-noise pentode used
fier with the higher voltage power sup­ be increased by the feedback factor in as a voltage amplifier. This is direct­
ply in Fig. 4), using a new transformer the output stage. This ruled out the coupled to the phase-splitter/driver,
made by Chicago Standard Transform­ use of a 6SN7 as a driver. The tube thus eliminating some phase shift that
er Corp.-the BO-15. This transformer found best suited for this purpose was would be introduced by a coupling
was designed for 655Os, EL34s, or the 5687, a dual-triode used extensively capacitor.
KT88's using screen taps plus 11 d b of in electronic computers. It is capable Stability of the amplifier is very good
tertiary feedback in the output stage. of exceptional output into the low im­ under practically any type of load. as
There is also 12 db of feedback taken pedance necessary when using fixed ca n be seen by the oscillograms of Fig.
around the whole amplifier. Many con­ bias. When used in the Mullard circuit 1. Over 20 db more of over-all feed­
figurations were tried and discarded for a!' a phase-splitter/driver, as it is here, back could be added before instability
one reason or another before the pres­ it will deliver 180 volts r.m.s. grid-to­ in the form of ringing or motorboating
ent design was arrived at. grid into a 50.DOO-ohm load. The outputs sets in. If the amplifier is to be used

V3
6550

RI4 5

RI6

RI'

RI9
Cl

R2
RI7 -:
JI
RI5
RIO

C3
R7

TI V2 V3 V4

W
��.,t.3".(ji)
C
B A
..,,,.(ii)UA.
SRI
+
lA. .-i -��WW"'-oNI/IJw"....,
......
R27 R28
VI
CII
FI + R29
V5 CIO
GZ34 R26
CH2
"Y

R25 Fig. 2. S chematic diagram

of original40.watl amplifier.
+ +
To convert to a 60-watl
+
CI3 CI4 CI5 unit a diff.rent power sup·
ply ( Fig. 4) must b. con·
nected to points A, B, C.

T..--O",,,,,, I,d'lf. 4JOO olt"", c.I., .. i/h ,O'�e"

RI. R..-l m�. olt m, '11 Wo �.J. R,.-1',OOO ohm, 10 w• • ",,.,0.,"4,,,. lap' a"d elllh04, 1"db.cI, .. ;"4i",; S. 16
R.-JjOO ohm, J Wo " ,. RI1",-j6D nh,". 1 ,.,. " ,. ohm. " c.; 6' .. .11, (CIt;CII,O S, ."4 .r4 80·
R..-JJO,OOO olom, 1 ... " •. R,r--27,OOO ohm, I ... " •. I' or eq u i .,. . )
R., R" R .. , R,,.-I7,OOO olom, 1 ....... er-.D' JlI., 400 P. cflpfle;Io, CH�8 107" 200 m•• lilt., cloo •• (Cloic.,o
R.-IIIOO o"m. 1 w. ,••• C..--'O JlI., l' ... . ,I,e. etlpfle;Ior 5,."d.,d R C8 1 00 0' .q .. i�.)
Rr-'IO ohm, I .... r••• C.--.2' JJI •• 400 ..... edpfle;Io, CHt-2 Ity., 200 ". •. , 60 oltm . 4.e. or ",.
R�4f700 01" , 1 .. Wo r••• C., C,-.' JlI .• 600 ..... eflpfl e;Ior �It'r cltol" (S'."cor C- 2 J2 ' 0, �qu;.".)
R ",...--11000 ohm, .. Wo c.rboN r ••• (J •• I.",) e" e 1-'0 JlJlI. e,trflm;, cflpflcilor SRt-20 md., 1 JO "' . "I,ni .. m recri�'r
Ru--l' ,000 oh". • .. Wo c.,.bo" r... (J•• I•.,) ell. C..-.40 Jl/.. J'O ... ,/�C. C_P_C;Io, F 1-2 dmp. tu"
R ,.-JO,OOO olr". • .. W. c.,.boll ,.,. (, •• I,.,) C,�.04 JlI .• 1000 ..... e_,_cilo, F..-22' m •• I .. u
Ru-180,OOO ohm, I Wo ,,,. CH. C,....'
.- O JlI.. 1'0 ... �/�c. c_P_CiIO, 5" St---S.p .•. r. " ... 1.'''''''0'' , .. il,,., rlorm,""
Ru. RII, Ru. R..--IOOO oh".. 'Il ar. r••• C 11, CH. C u-40 JJI., '00 ... ,I,e. c_,_,;Io, op,n
R", R,..-JOO.OOO oh,", 1 Wo r••• C ,.-2' �JlI. ce,."mic '_'flC;IO' M-11ft •. d.e. ""t".
R... R.....-I0,OOO ohm pot Tt-Pow" Irll'''. '20·0·'20 ..... @ 200 md.; / t-PI.ono i.c"
R.,--6 80 0""'_ '11 Wo ,.,. ( ". ,•., ) 6.1 �. @ 4.' .m/H.; 6.1 �. @ 1 _p.; , �. II�EF86 ... b.
RH-lZDO ohm. 1 .. . ,�,. @ 1 .mp •. ( C lo iu, o 5,."d.,,1 PCR-200 0' 11..-'687 , .. b.
R,;-6tJO ohm. 2 w. r�'. �q,,;"'.) 11., 11.-6"0 , .. b.

44 ELECTRONICS WORLD
 ·
,

·
_
_ I I
'10
·
/
·

..
... ·
... 50
�
...
� 40 ·

I­
...
1'", ·
'"
... ·
"'>0
l ·
v
10

· ,
v
o

10 100 IKC. lOot. IOOt<c.

""[0.- C"I

Fi9. 3. Pha •• charact.ri.tic. of the hi 9 h . fi d.li ty amplifi.r as

m e a.ur .d b.tw •• n fr.qu.nci •• of 20 cp •• nd 200 kc. by author.

Und.r·ch...i. vi.w of th e hom.·built uni t .howin9 the u •• of •

ground bu., lac.d wirin9, .. nd th e t.rmin.. 1 b o ar d con.truction. �

with a highly capacitive load, such as

an electrostatic speaker and a woofer
calling for a low damping factor, it
C 11 .11 might be ad vantageous to omit R .. and
C.. This leaves out the over-all feed­
back loop, but there will still be 11 db
SItS of feedback in the output stage' where
R35
it is most needed. This will give a
117V..... C. lower damping factor and an amplifier
that is very sensitive.

Construction
Construction of the amplifier is fair­
ly standard, being built on a single
Sv,," x 15" x 3" chassis. Most of the parts
arc mounted on a terminal strip to fa­
cilitate testing and ea sy access. All
signal-carrying leads, including the
feedback loop, were kept as short as
R"" Rn. R.u. 1 w. r" . po ssibl e . A ground _bus was used and
H.;-21.000 ohm, 10 ... . .. ;r� .. o""d rl:l. connected to the chassis at one po int .
N. .-160 ohm. 1 w. r('" CH�2 It.., •• 100 md. fo/h'r cholte' (S'anco.,.
In the top chassis view of the ampli­
11 -10.000 ohm po'. C-l]l' or ,qui.".)
H. '-27,000 ohm, I w. rl:'. SR" SR.t"'"--l O ,"4., 130 " . u-/.'n;um r,'ctifu:r
fier you will notice a meter mounted
11 '. R". RI" RIr--<S8,OOO ohm, 2 f�lllmp.luu on the side panel. with push-buttons by
H .:-600 ohm. 2 w. ,r,. f�-12" ma. I..,u- each output tube for b alanc in g the cur­
C,�-.07" 1-'/" 600 ... ctlpflc;/or T.-Po.,,,r /rall" "'0·0.450 .... @ 2 00 m•• ;
C 1�-40 pi., 2'0 " . '/('Co Cdpdc;/or
rent in the output stage. The meter is
6,J .. . @ 4 dmp,.; 6.J ". @ .6 11,,"1'.; ,. ".
C fr-50 JA./" 1 '0 p • .-/('c. cap.c;/or @ 2 amp" (Chicllgo SIoIIJtlrd PSC lV' or connected so as to read the voltage
CN--10 #/ .• 600 ,.. . "'('c. clJplJc;/or 'Cl",;,,·) drop across the tertiary winding. A 1-
en. ell, Cu. e'I-80 J.I./ •• J50 ... .. I,c. Ctlp. V.-GZJ4 'wb,
ma. meter is used and R�, se lecte d so
that with 100 ma. flowing through the
Fi9. 4. C i r cui t of t he c .. pacitor.input pow.r .upply u•• d to o btain the hi9h.r volt. tertiary winding t he metcr will read
"ge n..c . .... ry for the bO·watt v...ion of the amplifi.r. Filter c .. pacito ... ar ... er i••. full-scale. In t his way the tertiary
conn.ct .. d .. nd bleed.r re.i.ton are u.ed to eq ua li •• voltage drop•. Bec.. use of this wimling a cts as the meter shunt. The
.. rr.. n9.ment, tha ne9 ativ. t.rmin.. l. I usually the c .. n.) of C", and C", mu .t be
"value of R�, will va ry , depending on the
in.ulated from the cha ..i. and th.s. c .. p .. cito .. must be .ep .. rate unit.. Also for
d.c. ,'esistance of the tertiary winding
.afety t h e y should be mounted b.ne.. th th e chassi.. T he 47·ohm resi.tots across
the b.3·volt he .. ter windin9' .. r. r eq ui red .inc. the windin9' are not cenn,r.tapped. and the meter used. This was an extra
"frill" on the o ri ginal amplifier but can
Fi9. 5. l A) Tot.. 1 h ar monic di.tor/ion of the bO·w.. tt amplifi er at .. n ou t p ut wi thin be \'ery handy if you have an old meter
I db of full output. I B) Pow.r.r ••pon•• c urv. of the bO, watt er . Full·powe, re·
:!:: in the junk box that can be utilized.
'pon •• of the 4O-w .. tt v. .. ion i. just a li ttl. flatter .. t the fr.quency ",tremes. The parts list calls for some 4-watt
carbon resistors. The ROOO-ohm unit

�B Illllll 111111111 lllfHtfJ

was made up of two 1 6 ,OOO -o hm, 2-watt
resistors in parallel. The 25.000-ohm
unit is 12,000- and 13,ooo-ohm, 2-watt
resi stors in series while the 30.000 ohms
is made up of two 15,ooo-ohm, 2-watt

-81111111 11 nnlll 1II]JIg

resistors, also in se,·ies.
All of the pal'ts arc conservatively
rate d and should give long, twuble-free
pe,·for-mance. The "B+" supply bus
was fused to prev!'nt damage to the
\f"REOUENCY-CPS
-
,e.

4S
6550s if anything goes wrong with the
bias supply. An RC voltage divider was
used to obtain the bias voltage. It Is a
simple, yet effective, method of obtain­
ing the desired voltage and one that
dissipates very little heat. A choke­
input power supply was chosen for the
4O-watt version because of its excellent
voltage regulation as compared to a
capacitor input. With a swing in cur­
rent from 165 to 220 ma., the d.c. volt­
age will dr op less than 20 volts. In the
case of the GO-watt version, capacitor
input was used to get the higher volt­
age needed for the extra 2 db of power
output.
The bias adjustment (R .. ) should be
such as to produce a cathode current
for each output tube of about 75 ma.
for the 40-watt amplifier and about 55
ma. for the GO-watt ve rsi on . The bal­
ancing pot (R",) should be adjusted for
equal currents in both output tu bes .
When connecting the output trans­
former, careful attention should be paid
to the color coding of the primary leads.
When the winding is connected proper­
ly, it will take approximately 70 volts
r.m.s. on the grids of the 655O's for GO
watts of output. When the over-all
feedback loop is connected, the output
should drop by some 12 db. If motor­
boating or oscillation occur, you can
correct the phasing by reversing the
transformer leads or trans pos ing the
connections of the coupling capacitors
at the driver.
A word might be said here about how
the amplifier sounds, since the final
word on any amplifier is whether it
pleases the ear and sounds good after
prolonged l iste ning sessions. The highs
sound clean and the instruments well
delineated. The bass is solid and we ll
damped. A large measure of the Quali­
ty obtained with this circuit should b e
attributed to the outstanding charac­
teristics of the output transformer and
its ability to reproduce full output over
such a wide f requ en c y range.

D. C. YOLTAGES
PIN 2 1 4 5 6 7 • ,
V, 110 2.5 12 2.5
Va 240 " '2 92 12 250
v,. V. 410 410 -48 .5 -

Table I. Voltages m • .sured by author in

"O-watt unit. Voltages in 6Q-watt version
ar. fairly similar elcapt for somawhat
hi9her ".-plus" and bin on tubes V.. V •.

Ta.ble 2. Important .pecs of both versions.

Output 40 wolh 60 w.II.

Son.itivlty .7 volt fo, full I volt fo, full
output oulput
Ito.pon.o Down .1 db .t 00wn.7 db .t 20
20 cps; down .5 cp.; down I .)
db .t 20,000 cps db .1 20.000 cp.
,.t 40 w.lls' ,.t 60 w.II.,
Fe.db.c. 12 db ov.,·.1I 12 db ov.,-.II
11 db t.,ti..y 11 db t.rtl."
F.odb.c.
St.billty M.'9111 20 db 20 db
O.mpin.
hctor 10 10