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The fact that a work is unpublished shall not itself bar a finding of fair use if such finding is made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
 Web: http: // www. pearl - hifi . com
E-mail: custserv @ pearl - hifi . com Ph: + .1.403.244.4434 Fx: + .1.403.244.7134

Precision Electro-Acoustic Research Laboratory. Hand-Builders of Fine Music-Reproduction Equipment

A Little Input on Audio-Output Transformers

HE PRIMARY FUNCTION of most transformers The impedance reflected back to the primary side
T is to facilitate the transfer of power between cir-
cuits whose voltage-to-current ratios and/or DC levels
by an 8 load on the secondary is easily calculated: its
equal to the (turns-ratio)2 x (the load impedance).
are different. Transformers designed for use as imped- Thus, an 8 load reflects back to the primary as 2048;
ance converters do not function in purely DC circuits, a 16 load, 4096; and a 4 load, 1024. The point
although it is sometimes necessary to allow direct cur- here is that a transformer does not have a fixed input or
rent to flow through an audio transformer, such as a output impedance as such; it is a device used to match
push-pull or single-ended coupling or output device. circuits to each other in such a way that, for instance,
An audio-output transformer is usually required the maximum transfer of power can take place.
to couple a high source-impedance power-amplifying An irreducible limitation is that transformers are
stage capable of several hundred volts of swing at an band-pass filters, which, by definition, exhibit a finite
output of a few hundred milliamps to a low-imped- usable bandwidth. The geometric center frequency
ance, frequently ground-referenced load requiring a of the usable bandthe pass-bandis determined by
few tens of volts drive at currents of several amps. the design of a given transformer and is broadly
In Wattsrms, the power developed in a circuit is defined by the intended source and load impedances
the product of the current flow through the circuit, used as factors in the design. Other source and load
in Ampsrms, and the applied voltage in Voltsrms. This impedances can be used, provided that their ratio is
relationship holds for DC circuits and for AC circuits made equal to the source/load impedance ratio
where voltage and current reach their respective employed in the original design; then the geometric
peak and zero values at the same instant. center of the passband will simply move either up or
In the case of an ideal (i.e. 100% efficient, no dis- down in frequency. As this usually occurs without
tortion, no losses, etc.) single-ended, transformer- serious effects, the design center values should not
coupled, Class A amplifier capable of delivering 32W be taken as rigidly fixed absolutes.1
into 8, a 16.0V output forces 2.0A to flow through
the 8 load, yielding 16.0V x 2.0A = 32W of power. If A LAS R EALITY !
the turns ratio of the transformer is 16:1, the voltage Upon leaving the world of the perfect 32W, Class
swing across the primary is 16 x 16V and the current A amplifier, life becomes rather more complex, and
flow is 2/16A. The power developed is 162/8 = 32W. the difficult task of designing a high-power, wide-
If an attempt is made to force the tube output bandwidth, Class A(B) output transformer assumes
stage of this same amp to drive an 8 speaker its full proportions.
directly, a tremendous reduction in the power avail- A tube type power-output stage can be configured
able will be seen. The maximum current the output in many ways: triode, fixed screen-voltage pentode/
stage can deliver is limited to about 0.5A; as only 4.0V beam tetrode, swinging screen-voltage pentode/ beam
are required to force such a current through 8, the tetrodeultra-linearpartially or fully cathode cou-
volt/amp product yields a figure of only 2 watts! pled, Class A1 or A2, Class AB1 or AB2, etc.2, 3, 4 ,5, 6, 7
It can be readily seen that the ratio of primary Much study of the numerous options has shown
voltage to secondary voltage is 16 162= 16 while the that a properly executed, ultra-linear, partially cath-
current ratio is .125 / 2 = .0625. ode-coupled, Class AB1 output stage represents a
Either of these relationships can be used to well optimized, balanced solution to the various
determine the turns ratio of the transformer,
The geometric center frequency of any frequency band is equal to the
although the use of the voltage figures usually makes square root of the product of the lower and upper frequencies, i.e. the geomet-
measurements easier. In the present case, looking ric center of the 20Hz to 20kHz band equals the square root of 400,000, or
from the primary to the secondary, the ratio is 16:1, 632.45Hz.
If a piece of frequency-response graph paper with frequency scaled loga-
i.e. for every 16 turns on the primary, there is one rithmically is folded in half so that the 20Hz line is laid over the 20kHz line, the
turn on the secondary. crease will occur at 632Hz.

AN 2.1 - Page 1
problems presented. Consequently Watts 300 turn produce a rough high-fre-
the PEARL SC280 transformer has quency response with ragged phase
been specifically designed to oper- 150 and group-delay figures that can
ate in this manner. This by no make the application of loop feed-
means precludes its use in other back a perilous undertaking.
output-stage configurations. The 60 If the amplifier is intended to
stage described presents serious operate pentode class AB , the
design challenges; as these have 30 difficulties are further compounded
been well met, the transformer will by the need for fantastically tight
work admirably in more straight- inductive coupling between the two
15
forward applications such as high halves of the primarywhich must
idle-current class A triode stages or be accomplished with an absolute
low idle-current, class AB1, fixed minimum of capacitive coupling.
screen voltage, pentode / beam 6 The need for low capacitive-
tetrode operation. coupling between the half pri-
No matter the output stage 3 maries must be met, as any half-
configuration, good low-frequency primary-to-half-primary capaci-
response requires a high value of pri- 1.5 tance actswith an increase in dri-
mary open circuit inductance, OCL. ving frequencyas a decreasing
A good rule is that for every value shunt load between the
1000 of reflected primary im- .6
opposing halves of the output
pedance, the transformer should stage. At high frequencies, a low-
develop 70H of OCL from very low ered impedance between these two
power upwards. This will yield a .3 points will excessively load the out-
small-signal low-frequency 1dB 1Hz 2 5 10 20 put tubes and adversely affect the
point in the 3Hz region. It should be high-frequency output at the sec-
Fig. 1. The decrease of low-frequency
noted that the power-handling output ability with decreasing frequency.
ondary. 9, 10, 11, 12, 13, 14, 15, 16, 17
capacity of any transformer reduces In a class A amplifier, current
to 14 of its previous value for every halving of frequency. always flows in both halves of the primary, whereas
Thus, a device capable of a 300W output at 20Hz can pentode class AB operation results in the cessation
meet only a 75W demand at 10Hz while at 1Hz its out- of current flow in one or the other half-primary as
put capability has dwindled to a few hundred milli- the output stage makes the transition from class A
watts. See Fig 1.8 into class B. Because the two half-primaries are
While good low-frequency extension is, relatively never perfectly flux-coupled to each other, a back
speaking, easily accomplished, smooth, extended EMF spike is generated in the half-primary-to-half-
high-frequency performance is always difficult to primary LL whenever current flow is abruptly termi-
achieve. The physical characteristics of the sorts of nated in either half primary. This voltage acts to
windings required for extended low- vs. extended stimulate a resonant circuit consisting of the half-pri-
high-frequency response are exactly opposite in mary to half-primary LL and some portion of the var-
nature and there exists an ultimate limit on the ious capacitances present.
usable bandwidth attainable for a given output This causes the appearance of a pernicious form
power and unit cost. If the very best of modern mate- of notch distortion, resulting in upper-frequency
rials are used and no effort is spared in the precise common-mode signals that can partially couple into
execution of the design, the bandwidth can be the load. As the feedback network samples the out-
extended beyond that previously considered to be put from the amplifier at the point of interface with
practically achievable. the load, this error voltage finds its way, via the feed-
Among the factors to be carefully minimized are back loop, throughout the entire amplifier.
the leakage inductance, LL, and the inter-layer and Unfortunately, feedback can do nothing to
inter-winding capacitances. Due consideration must reduce this particular distortion form. The circuit
be given to the inter-winding AC potential differ- exhibiting the distortion being downstream from the
ences, as these directly affect the net charge, hence output tube(s) biased into cutoff, feedback can only
the effective capacitance, and therefore the charac- serve to aggravate the problem.18, 19, 20
teristic impedance of a given section of the winding.
Triode and UL connections are not nearly as problematic in this respect
Mismatches in these values cause energy
as fixed-screen-voltage pentode/beam tetrode connections.
reflections within the windings and result in numer- Not to be confused with the crossover distortion exhibited by improperly

ous high-frequency, self-resonant modes. These in biased solid-state amplifiers

AN 2.1 - Page 2
 Two of the cures for this situation are: lab is one in which all of the factors above and numer-
ous others have been very carefully considered.
thevery difficultreduction of LL to vanish- No matter how admirable the standard set by the
ingly low levels circuitry preceding the output transformer, much per-
the use of the McIntosh unity-coupled circuit formance can be lost if the transformer has not been
designed for excellent high-frequency common-mode
In their early papers McIntosh suggested that a performance. Such a specification demands extraordi-
ratio of OCL to half-primary-to-half-primary LL of nary AC and DC balance within the windings, and
80,000:1 would be required to effectively eliminate these have been achieved in the SC280 transformer.
the Class B notch-distortion problem. In the PEARL See the Specifications section for more details.
SC280 transformer this ratio is 40,000:1. In summary, the present transformer is the fruit
Any push-pull amplifier must show extreme pre- of a thorough-going and often arduous research and
cision with regard to the balance of its push and pull development effort spanning a four year period.
halves.2, 21, 22, 23 Without such balance, common- Much study of the prior art revealed many fasci-
mode error voltages are generated that cannot be nating aspects of the work done during the heyday-
extracted by any known corrective measure.24 era of the vacuum tube. We have been able to look
Additionally, such voltages can be generated over the shoulder of some of the most talented
through any of the following actions: designers of the day.
In this work we have used what we consider to
capacitive coupling of the AC line to the core of be the very best ideas expounded by numerous
the power transformer, to the chassis, and workers and writers in the field. To this admixture
hence to signal ground of knowledge and experience we have added our
poor noise/ripple rejection or high internal own insights. Our finished product incorporates a
impedance in the various power supplies host of design and construction features never
improper phase-inverter stage AC balance 25 before simultaneously embodied in any output
unbalanced gains in the push and pull halves transformer of which we have knowledge.
of the driver and output stages
tube-aging and thermal-drift effects 26 N O T E S O N T H E S P E C I F I C AT I O N S
AC and DC winding imbalances and, if the out- The transformer is specd overleaf in just one of
put transformer is intended for Class B opera- many possible configurations. The device can be
tion, high half-primary-to-half-primary leak- wound to meet almost any requirement, based on
age inductance LL. the clear understanding that all the elecrical para-
meters co-mingle: one cannot be changed without
The amplifier design presently in progress in our affecting most of the others.

REFERENCES
1 Using Audio Transformers, Herbert Ravenswood, Radio-Electron- 14 Output Transformer Design, N.H. Crowhurst, Audio, September
ics, April 1960 1956
2 A Multi-Loop Self-Balancing Amplifier, J.R. Macdonald, I.R.E. 15 Audio Transformers CAN Be Good, N.H. Crowhurst, Audio, May
Transactions on Audio, July-August 1955 1956
3 Class-A Push-Pull Amplifier Theory, H.L. Krauss, Proc. I.R.E., 16 Leakage Inductance, N.H. Crowhurst, Electronic Engineering,
January 1948 April 1949
4 New 50-Watt Amplifier Circuit, F.H. McIntosh & G.G. Gow, Audio 17 Winding Capacitance, N.H. Crowhurst, Electronic Engineering,
Engineering, December 1949 November 1949
5 Realistic Audio Engineering Philosophy, N.H. Crowhurst, Audio 18 Quasi-Transients in Class-B Audio-Frequency Push-Pull
Engineering, October 1950 Amplifiers, A. Pen-Tung Sah, Proc. I.R.E., November 1936
6 Ultra-Linear Amplifiers, F. Langford-Smith & A.R. Chesterman, 19 The Distortion Resulting from the Use of Center-Tapped Trans-
Radiotronics, May, June, July 1955 formers in a Class-B Power Amplifier, R.G. De Buda, I.R.E. Trans-
7 Amplifiers And Superlatives, D.T.N. Williamson & P.J. Walker, actions on Audio, July-August 1960
Wireless World, September 1952 20 Extended Class-A Audio, H.T. Sterling, Electronics, May 1951
8 Puzzled About Transformers, N.H. Crowhurst, Radio-Electronics, 21 Push-Pull Balance, W.T. Cocking, Wireless World, November 1947
December 1960 22 Balanced Amplifiers, F.F. Offner, Proc. I.R.E., Waves and Elec-
9 Output Transformers, James Moir, Audio, February & March, trons Section, March 1947
1960 23 Push-Pull R/C-Coupled Amplifiers, F.F. Offner, Review of Scientific
10 Design of the Output Transformer, R.C. Hitchcock & W.O. Instruments, January 1937
Osborn, Electronics, November 1930 24 Understanding Common-Mode Signals, R.N. Marsh, Audio, Feb-
11 How Good is an Audio Transformer? N.H. Crowhurst, Audio ruary 1988
Engineering, March 1952 25 Distortion & Phase-Splitter Unbalance in Push-Pull Amplifiers, N.
12 Measuring Up an Audio Transformer, N.H. Crowhurst, Audio Sokal, Radio & TV News, issue unknown (Copies are available
Engineering, November 1952 from PEARL)
13 Audio Transformer Design, N.H. Crowhurst, Audio Engineering, 26 Balanced Output Amplifiers of Highly Accurate and Stable Bal-
February 1953 ance, E.M.I. Laboratories, Electronic Engineering, June 1946

AN 2.1 - Page 3
 The PEARL 280 Watt, Class AB Audio-Output Transformer

Specifications
0.375" NC (as of 6/94, subject improvement without notice)
n.
n o t show Output power at 20Hz: 280W into 6.
inals 6b. )
0.75" (Term See Fig. Primary impedance: 1040, centre-tapped.
Secondary impedance: 6, centre-tapped.
Ultra-linear winding: 40% of primary turns, centre-tapped.
Tertiary f/b winding: 2 x secondary turns, centre-tapped.
Turns ratio; pri. to sec: 13.16:1.
Square-wave response: See Fig. 3.
" Frequency response: See Fig. 4.
8.50
Primary inductance: 70H @ 1W delivered into 6
Tolerable DC-unbalance; 4 10mA
Leakage inductance: 685H, pri. to sec., referred to pri.
2750H, 12 primary to 12 primary.
Capacitances:
Pri. to sec: 6000pF.
1
2 pri. to 12 pri: 1500pF.
DC balance: From the center-tap to either end,
" "
5.16 8.28
any windings Rdc is within 0.1%.
AC balance: The 100kHz frequency response
Fig. 2. Overall from any half-section, 'A' to any
dimensions. other section is within .5dB of that
from the the other half-section, 'B,
to the same section.
First resonance: 190kHz.
100
Rise time: 1.3Sec.
90%
Winding configuration: Astatic, interleaved, cross-coupled.
Winding DC resistances:
Primary: 42 plate-to-plate.
Secondary: 0.16 load-to-load.
Screen: 34 screen-to-screen.
Tertiary: 8.8 end-to-end.
Winding rms current capacity @ 800 circular mils/amp:
Primary: 650 mA (325mA per 12 primary).
Secondary: 8A. (= 384Wrms. continuous into 6)
Screen: 250mA (125mA per 12 screen).
10%
Tertiary: 250mA (125mA per 12 tertiary).
0 Interlayer insulation: Low-loss, low dielectric constant.
Vert. - .2V/div. Insulation Voltage Rating:
Horiz. - 5Sec/div.
Primary to secondary 1500V
Fig. 3. 20kHz. Square-wave response @ 1.5 watts out- Primary to screen 1000V
put into a 6 resistor from a 600 push-pull source. Primary to tertiary 1500V

0dB

-3dB

-6dB
-8dB
1Hz. 2 5 10 20 50 100Hz. 10kHz. 20 50 100 200 500kHz.

Fig. 4. 1.5 watt primary-to-secondary frequency response from a 600 push-pull source into a 6 resistor.

AN 2.1 - Page 4
Impregnation: High-temperature, vacuum drying,
followed by vacuum/ pressure, .508 4.00 .508
microcrystalline-wax impregnation.
Potting: Microcrystalline wax .514
Core type: C-core; 4 mil, Silectron (3% Si.) .50 dia., 4 places
Core power-rating at 16kG:
20Hz: 280W.
60Hz: In excess of 2500W.
Enclosure: Deep drawn, aluminum can.
Finish: Textured, black powder-coat. Chassis cut-out area
Termination: Secondary: leads come directly off
the windings. All others: 21 AWG
Polythermaleze insulated, Ohno
Continously Cast, 99.99%-pure (4-
9s) copper. 8.28 6.00 7.00
Dimensions: 8.5H x 8.28W x 5.16D. See Fig. 2.
Mounting: 4 aluminum studs protrude from the
Potting can
bottom of the can. outline
Weight: 38 pounds, 17.3 kgs.
For example, the response from half-primary 'A' to half-secondary 'B' is within .5dB
.258
at 100kHz of that from the other half-primary 'B' to either half-secondary 'A' or half-sec-
ondary 'B'. The primary and secondary center taps must be connected together and
tied to the common of the measuring system used.
4.50

Fig. 5. Transformer schematic & winding polarities.

1.14
.514

Anode 1 Sec 8
1 9 5.16

10 Fig. 6a. Chassis cut-out.

Sec 4
Half-
Fig. 6b. Lead-out configuration.
Primary 'A'
Half-
Secondary 'A'
11
Sec CT
2
Anode 1: B+ Sec 1 8 9
Anode 2: B+
Half- 1 Anode 1
3 Secondary 'B' Sec 1 4 10
12 2 Anode 1: B+
Sec 4 Sec: CT 11
Half- 3 Anode 2: B+
Primary 'B' Sec 2 4 12
13
Sec 8
4 Anode 2
Sec 2 8 13
Tertiary: +
4
14
8.28
Anode 2

Screen 1
5
Half- 5 Screen 1
Tertiary 'A' Tertiary: + 14
Half- 6 Screen 1: B+ Tertiary: CT
Twisted pair, the
Screen 'A' section-to-section 15
15 connection is
Tertiary: CT 7 Screen 2: B+ thereby external.
6 16
Tertiary:
Screen 1: B+
8 Screen 2
Screen 2: B+
7 Half-
Tertiary 'B'
Half-
Screen 'B'
16
Tertiary:
8 5.16
Screen 2

AN 2.1 - Page 5
 Verso Filler Page
 Web: http: // www. pearl - hifi . com #200, 2137 33 Ave, SW; Calgary, AB; T2T 1Z7
E-mail: custserv @ pearl - hifi . com Ph: + .1.403.244.4434 Fx: + .1.403.240.2851

Perkins Electro-Acoustic Research Laboratory. Hand-Builders of Fine Music-Reproduction Equipment

Update to Audio Note 2.1

The Regulated-Screen Ultra-Linear Output Stage

URTHER STUDY OF THE PROBLEMS relating to sibility. It is well known that pentode operation
F transformer-coupled, tube-type, audio-power-
output stages has yielded some interesting insights.
benefits greatly from good regulation of the screen
supply and only rarely does one see a modern design
While a detailed paper will have to await time to in which this supply isnt regulated. If a pentode is
write our Audio Note 17; Triode vs. Pentode, The Grass operated with its control grid-to-cathode voltage fixed
is Greener in the Middle; it is useful to present an and with its screen tied to an unvarying DC source,
overview here. Ultra-linear operation of the output variations in plate voltage will cause only slight
stage has long been the subject of controversy and changes in plate current. If, however, the plate voltage
some of the reasons for this have recently become is fixed and the screen voltage is varied, large changes
apparent to me. in plate current will be observed. It follows that if the
The basic operation of such stages has been screens are energized from the usually unregulated
investigated in a very thorough manner by several plate supply in a power amplifier, any variation in this
researchers, most notably Herbert Keroes and David voltage will cause a variation in plate current. See
Hafler early in the 1950s and by F. Langford-Smith & Figs. 1a & 1b. Due to the usually considerable internal
A.R. Chesterman6 a few years later. Broadly speak- impedance of an unregulated supply of any sort,
ing, the UL stage has been shown to exhibit lower changes in current flow through such a supply will
total harmonic distortion than either triode or fixed- cause the voltage dropped across its internal imped-
screen pentode/beam tetrode operation, very nearly ance to either increase or decrease. This causes varia-
the power gain of the pentode configuration and tions in the available output voltage as an inverse
much lower, more triode-like, source impedance. function of the current demanded from the supply.
While it would seem like an ideal solution to many The greater the need for current, the lower the voltage
problems, few designers have availed themselves of available to force that same current through the load;
its benefits. Fixed screen, pentode operation remains with a diminution in current flow, the available volt-
the most common output stage configuration despite age increases. This results in the dynamic compres-
its considerable disadvantages, particularly when sion of a music signal as follows; increasing signal
driven into Class B operation. voltage demands increased current flow though the
The sound of UL stages has often been described power supply, causing increased voltage drop across
as somewhat soft, veiled and lacking in dynamics. As the supplys internal impedance, resulting in lowered
all of these descriptions invoke visions of poor power available voltage at the output of the supply. Lowered
supply regulation it was decided to consider this pos- supply voltage means lowered screen voltage means
lowered plate current means lowered output power.
The terms pentode, tetrode, beam-pentode and beam-tetrode are often
In other words, dynamic compression is taking place
used interchangeably to describe the devices used in the output stages of
tube amplifiers. This can be a rather confusing state of affairs as these are dif- by means of the situation where the amplifiers output
ferent items. Because only a handful of beam-pentodes have ever been voltage decreases under load, due to lowered screen
designed, this configuration wont be further described.
Any tube designed for audio output service that uses a screen grid must
voltage, and increases with diminishing output, due
embody some method of controlling secondary emission. Three-grid pen- to increasing screen voltage.
todes and two-grid beam tetrodes both suppress secondary emission from If the output stage is connected in a conventional
the plate by the creation of a near-zero-potential region between the screen
grid and the plate. Tetrodes used for RF amplification arent generally UL manner, with the screens deriving their DC polar-
designed to suppress secondary emission but work around the problem by ization from taps on the primary of the output trans-
keeping the plate voltage well above the screen potential although this results
in a considerable loss in circuit efficiency. former, the mechanism of dynamic compression just
Providing it's not a triode and in the absence of the exact details of the inter- described is at work. The problem with convention-
nal construction of a given audio-output tube, the best way to describe it is as
a pentode/beam-tetrode. This being rather a mouthful, I usually abbreviate to
ally wound and tapped UL output transformers is
simply, pentode. that, short of regulation of the main B+, there is no

AN 2.1.1- Page 1
way to stabilize the DC potential on the screens. Even cathode voltage, no sustainable amount of increase in
if such regulation was implemented, slight imperfec- plate potential will cause significant current flow to
tions in the regulators action would still act on the resume. With its high voltage gain, sufficient negative
screens and hence, detrimentally on the music. Con- control-grid-drive will abruptly terminate current
sequently, the SC280 transformer has been re- flow through one half of the pentode configured push-
designed to include a separate, unity-coupled screen pull output stage.
winding that allows the implementation of a dedi- The triode, with its lower voltage gain, does not
cated soft-starting, regulated screen supply. This exhibit this effect as plate current, for a given grid-to-
allows different voltages to be impressed upon the cathode voltage, is very much a function of plate
plate and screen making it possible to
recoup the loss in power when going Figs. 1a &1b. Plate current vs. plate voltage for two different screen volt-
from pentode to ultra-linear by ages is shown for the KT90 beam tetrode.
increasing the UL plate voltage to yield
greater available swings in plate volt- 500

age and hence greater power. 450 KT 90 Tetrode

Screen=200V
Additionally, the leakage induc-
400
tance between the two sections of A
n
each half-primary created by a screen o 350
d
tap is always high. With correct posi- e
C 300
tioning, a separate screen winding u 0V
r 250
can reduce this detrimental loss of r
e
coupling to nearly zero. n 200 -5V
t
In terms of secondary plate-emis-
- 150
sion, UL operation is not well optimized -10V
M
with the plate and screen energized a 100
from the same DC potential. By care- -15V
fully balancing the plate and screen 50 -20V
potentials and the screen-feedback ratio 0 -25V

further benefits can be obtained. 0 50 100 150 200 250 300 350 400 450 500 550 600 650
A n o d e Vo l t a g e - Vo l t s

MORE GOOD NEWS Fig. 1a. Observe that there is only a slight change in plate current when the plate
Regarding the previously men- voltage is increased from 250V to 350V with a negative biasgrid to cathode voltage
of 25V on the control grid. Also note the plate current at just a few mA.
tioned difficulties with Class A, fixed-
screen, pentode operation and the
notch distortion problem created 500
thereby, it can be shown that triode 450
0V KT 90A Tetrode
and UL stages do not suffer from this Screen = 300V
problem to any great extent. With a A 400 -5V
n
transformer-coupled, push-pull out- o 350
d
put stage the action of a tube being e
turned on causes its plate voltage to C 300 -10V
u
decrease, ie. its plate becomes less r 250
r
positive than the B+ supply. Due to the e -15V
n 200
autotransformer action of the output t
transformer, the opposing plate - 150 -20V
becomes more positive than the B+ M
a 100 -25V
supply. As a result, there can be a 80
75
peak-to-peak AC potential difference 50
-30V

of nearly twice the B+ voltage between -35V

0
opposing plates at full output. To 0 50 100 150 200 250 300 350 400 450 500 550 600 650
repeat an earlier statement; for a given A n o d e Vo l t a g e - Vo l t s
control-grid-to-cathode voltage, pen-
tode plate current is essentially a func- Fig. 1b. Note the increase in plate current due to the increase in screen voltage, from
+200V to +300V. The 25V bias line is highlighted for ease of comparison with Fig.1a.
tion of screen voltage. In other words, When the plate voltage is increased from 250V to 350V the plate current increases less
if a pentode is biased into cut-off by than 7%, from 75mA to 80mA. In both 1a & 1b observe the abruptness with which plate
sufficiently negative control grid-to- current drops to zero (cut-off) with increasingly negative control grid voltage.

AN 2.1.1 - Page 2
potential. For a given plate voltage, the control grid A properly executed ultra-linear stage offers the
can be made sufficiently negative to cut-off plate cur- advantages of both triode and pentode operation
rent flow. However current will flow once again with with shortcomings of neither.
an increase in plate voltage and such an increase
occurs as a natural consequence of
the autotransformer action of the out-
put transformer. The greater-than-B+ 500
plate voltage induced thereby acts to 450 0V KT 90A Triode
keep current flowing through the tube
400
being turned off despite increasingly A -10V
n
negative control grid drive. See Fig. 2. o 350
d
As it is the abrupt cessation of e -20V
300
current flow in a half-primary that C
u
sets its leakage inductance into reso- r 250 -30V
r
nance with its stray capacitance, it e 225
n 200
t -40V
can be seen that if this sudden
- 150
switching action can be alleviated or -50V
M
eliminated so can the notch-distor- a 100
-60V
tion problem. Being that this distor- 70
tion is more-or-less a given with pen- 50 -70V

tode operationin that feedback has 0

no beneficial effect and in fact aggra- 0 50 100 150 200 250 300 350 400 450 500 550 600 650
vates the situationit becomes all A n o d e Vo l t a g e - Vo l t s

the more apparent why triode stages

Fig. 2. Plate current vs. plate voltage for a triode-connected (screen tied to plate) KT90.
with their softer turn-off characteris- Note the substantial effect of plate voltage upon plate current; an increase in plate volt-
tic are enjoying renewed interest. age from 250V to 350V results in a 320% increase in plate currentfrom 70mA to 225mA.
Because the UL stage makes the Following the highlighted 350V line down from maximum to zero plate-current, note that
transition from Class A to Class B in the amount of negativegrid-drive required to cut off current flow through the tube is
50V and that current flow is gradually turned off with increasing negative control grid
a similarly gentle manner, it exhibits bias. Follow the highlighted 50V line over to the 500 plate-volt line, note that plate current
the same freedom from the nasty has risen from zero to 80mA. Compare this result with Fig. 1b to see that, in the tetrode,
sounding notch-distortion problem 50 control-grid volts will cut off current flow almost completely.
as the triode stage. See Fig. 3.
A d d i t i o n a l l y, t h e U L s t a g e
500
exhibits a very low input capacitance 0V KT 90A Ultra-Linear
which is something that must be for- 450 300v Screen - 25%
Screen Feedback
saken with triode operation. Due to 400
A -5V
the Miller Effect, this value is typi- n
o
cally 3-5 times greater with a triode d 350
e
connected output stage than with UL -10V
C 300
or pentode operation. The stage dri- u
r 250
ving triodes must therefore source r -15V
e
current to and sink it from the output n 200
t
stage at a far greater rate than when - 150
-20V

driving a pentode or UL output stage. M -25V

This must be done in order to retain a 100
90
equivalent output bandwidth, feed- 60
-30V
50
back loop stability and consequent -35V
-40V
immunity to the reactive loading pre- 0
sented by most loudspeakers. 0 50 100 150 200 250 300 350 400 450 500 550 600 650
B+
A n o d e Vo l t a g e - Vo l t s
Thus we see that UL operation
175 200 225 250 275 300 325
can yield the much appreciated sweet- Screen Voltage - Volts
B+

ness and load tolerance of the triode

stage with the power output capacity Fig. 3. Plate current vs. plate voltage for a KT90, ultra-linearly driven. The screen grid
is connected to a dedicated winding of 25% the number of turns on the primary. Note
of the pentode stage all for the price of the triode-like plate-current-as-a-function-of-plate-voltage and the gradual cut off of
a well-placed screen-winding within plate current with increasingly negative grid drive. Note also the increase from zero
the output transformer. plate current at cut-off to some significant value with increased plate voltage.

AN 2.1.1- Page 3