Speaker-Builder-1980-02

A Three-cabinet suspended design | All about electrostatics
SPEAKER
2/1980 us $3.00
UK £1.25 I 15 Fr.
BUILDER
ITS
WIRE 11
SPEAKER CABLES:
Science or Snake Oil
A udiophiles recently began re¬
examining the performance of
every link in the audio playback chain,
by Nelson Pass
mercial sound distribution systems
have resorted to higher voltages, which
improve transmission much like the
and before long their attention turned that the cable can cause no power loss high voltage utility lines which carry
to the lowly loudspeaker cable. In or distortion. This corresponds to a power many miles.
response to demand, a number of com¬ wire having no resistance, inductance, Fig. 1 shows a fairly simple first
panies are producing or distributing or capacitance, which in real life order model of a loudspeaker cable.
new and exotic cables claimed to im¬ translates to an infinitely short cable. 1 The inductance, resistance and
prove audio power transmission from treat this premise as fundamental, be¬ capacitance are approximated as com¬
amplifier to speaker. Pointing to lower cause in general it results in the best ponents in a circuit, sectioned off per
resistance and inductance, proponents performance. (It may not do so in some unit of length. In this example, the
of the newer cables insist they sound specific situations; for example, one values are for simple 18 gauge "zip”
significantly better ("better than an ex¬ could imagine a special case where cord and one foot lengths, so that L is
pander!"); however, the subject is con¬ some resistance or inductance might the inductance per unit foot, R is the
troversial, and some hi-fi notables improve the sound.) resistance, and C is the capacitance,
claim performance increase is negligi¬ Regardless of the cable type, the ef¬ measured in Henries, Ohms, and
ble and the higher capacitance of some fects it introduces to a signal are pro¬ Farads respectively. As a practical mat¬
new cables can cause amplifier in¬ portional to its length: the shorter the ter, the values of these elements repre¬
stability and damage. 1' 4 cable, the more intimate the connec¬ sent tradeoffs against each other: for
Neither view is completely correct: tion between amplifier and loud¬ example, low inductance is easily
the new cables are neither panacea nor speaker. Subtle differences between achieved with high capacitance and
placebo, but components whose cha¬ cable types become more dramatic vice versa, and the ratios of these
racteristics must be evaluated in the with increasing length and shrink values give rise to the cable's
context of their usage. Hoping to shed toward zero as the cable gets shorter; characteristic impedance, as I shall
some light on the subject, I obtained thus the audiophile whose amplifiers discuss later.
samples of various cables, performed a sit close to his speakers need be less Researchers have chiefly concen¬
number of tests, and drew a few con¬ concerned than he whose cables are 40 trated on the cable's inductance and
clusions. feet long. To this end, some manufac¬ resistance, for they impede the flow of
Almost everyone seems to agree that turers have installed amplifiers within electrons between the amplifier and the
ideally the amplifier should be so in¬ their loudspeakers, exchanging speaker loudspeaker. Resistance causes loss at
timately coupled with the loudspeaker cable problems for preamp ones; com¬ all frequencies while inductance causes
FIG. 2
Cable Type il/ft. pF/ft. gH/ft.

24 ga. Zip .05 16 .24
18 ga. Zip .014 28 .21
Lucas .0055 20 .25
Monster .0034 24 .21
Fulton .001 28 .19
Polk .0075 500 .026
Mogami .0042 170 .023
Audio Source
Hi. Def. .013 280 .037
Audio Source
Ultra-Hi.
Def. .012 600 .029
Speaker Builder | 2/80 6

Fig. 3 Series Impedance Test
cables were virtually indentical at

higher frequencies.
I tested five different types of twin
lead cables: 18 and 24 gauge "zip" cord
and three specialty cables, "Monster",
Lucas cable, and Fulton wire (gold). I
bought two samples of each of 18 and
24 gauge wire off reels at a local Radio
Shack and a hardware store. All the
cables tested were 10 feet in length.
"Monster" cable, marketed by
Audio Sales Associates in San Fran¬
cisco, California, is an approximately
11 Vz gauge twin lead, similar in con¬
struction to very large lamp cord with
a thick clear plastic jacket, with large
spade lugs at each end for attachment
to large screw terminals on "five-way"
binding posts as commonly supplied
on loudspeakers and amplifiers. Lucas
loss proportional to the frequency. various gauges (lamp or ''zip'' cord be¬ cable is approximately 14 gauge,
Capacitance has not usually been con¬ ing an example) and low inductance- jacketed in green plastic with a ribbon
sidered significant because its values do high capacitance coaxial or interwoven shape, and is marketed by S.O.T.A.,
not impinge upon the audio band. types. Their measured performance Halifax, Canada. Fulton "gold" cable,
However, we will see later that it may also falls into two categories, 0-100kHz available from Fulton Musical In¬
sometimes important. effects and 100kHz - 40MHz effects, dustries with dual banana and other
The new kinds of cable seek to which for convenience I will treat connectors, is an extremely large gauge
reduce resistance and/or inductance separately. My analysis was greatly twin lead having by far the lowest
and thus improve the amplifier-speaker simplified by the fact that within the resistance of any cable tested; it is also
connection. They fall into two two cable categories performances useful for pulling up tree stumps or
categories: multistrand twin lead of were very similar; indeed, many of the jump starting locomotives. Fulton also
Fig. 4 Series Impedance
AUDIO SOURCE "HIGH DEFINITION"

AUDIO SOURCE "ULTRA HIGH DEFINITION’
LUCAS CABLE
I8G ZIP CORD
24G ZIP CORD 25Û —
FULTON CABLE
.i n PER FOOT POLK CABLE
MOGAMI WIRE
MONSTER CABLE
24G
AUDIO SOURCE "HIGH DEFINITION"

I8G
oin PER FOOT
AUDIO SOURCE ULTRA HIGH DEFINITION"
POLK
LUCAS
MOGAMI
MONSTER
FULTON
00IÛ PER FOOT 1-- ■ - ■ ' '. J ’ i
IHz 10Hz 100Hz IKHz lOKHz lOOKHz
7 2/80 I Speaker Builder

make a ''brown'' version similar to
Monster cable.
The four samples of low inductance
cable I tested boasted more exotic con¬
struction than zip cord. Two colorful
types, Polk Sound Wire and Audio
Source Ultra High Definition Wire, use
large numbers of separately insulated
strands closely interwoven in such a
way that the wires cross at an angle to
each other instead of running parallel.
This reduces the magnetic induction
between strands and lowers cable in¬
ductance, at the cost of higher
capacitance.
High definition cables, another
variety of low inductance cable from
Audio Source, consists of eight twisted
pairs of wire arranged into a flat rib¬
bon. Mogami wire is a large coaxial
cable consisting of a grey plastic hous¬
ing containing two concentric "shells" Hertz
of wire strands, the inner conductor
enclosing a plastic core. "Smog Fig. 5. Effects of 40ft 18 gauge cable on frequency response.
Lifters," another tested cable, is
distributed by Disc Washer. It bears a resistance; but each of these types has ping factor 4 5 and b. The performance
resemblance to Audio Source's high to all intents and purposes inductance context lies also in the amplifier's
definition cable, with loosely woven effects. The series impedance is a more source impedance and the
braids of conductor. or less linear function of the length of loudspeaker's load impedance. The
Fig. 2 shows the relative values of the cable, so a one foot length will have system's performance will depend on
resistance, capacitance and inductance one-tenth the series impedance shown the complex sum of the impedances in¬
of each of these cables. while 100 feet would have 10 times the volved:
amount. -1- 7 _1_ 7 -1- 7
Zsourct ^connections ^cable ^load
SERIES IMPEDANCE TEST
1 1 1
IMPLICATIONS OF LOW
I tested a 10 foot sample of each cable FREQUENCY TEST The speaker has generally been
type using the Fig. 3setup. I drove the designed to be driven by a voltage
cable by a high source impedance and The existing literature covers the sub¬ source, so our ideal premise requires
measured the voltage across it, show¬ ject quite well. However, I think you source and cable impedance to be very
ing its series impedance. This voltage, will find it useful if I briefly touch on small compared to speaker impedance.
referenced to a .112 non-inductive the effects of this series impedance as it In this case, the speaker's design
resistor and measured from DC to relates to frequency response and dam¬ dominates the performance as intend-
100kHz, clearly shows the cable im¬
pedance's resistive and inductive com¬
ponents (Fig. 4). For the twin lead
types, inductive and "skin effect" (an
additional high frequency resistance ef¬
fect) components begin to show up at
about 1kHz; they increase the im¬
pedance, causing high frequency loss in
addition to the cable's resistive losses.
Interestingly, all the twin lead types
have similar cable inductance values,
approximately 2^H per 10 feet, and in
the region just above the audio spec¬
trum they are nearly identical. Below
20kHz they fan out to their respective
resistance values. The lightest wire,
#24, clearly has the most loss, while
Fulton cable has the least.
The series impedance test differen¬
tiates the low inductance cables from
twin lead as they exhibit an order of
magnitude less impedance at 100kHz.
Of these, Mogami Wire had the lowest
series impedance, by virtue of its lower Fig. 6. Effects of 40ft 18 gauge cable on frequency response.

ed, so the variations in the advent of wide bandwidth power oscillator, it is surprising to discover
loudspeaker's impedance do not in¬ amplifiers has demonstrated other new amplitudes as high as 5V on the other
teract to produce frequency response effects; several amplifier designs (stable end —an apparent gain of 100! As
deviations. Fig. 5gives an example of with reactive loads such as capacitors) shown by the ringing in the pulse
such a case, where the aberration is oscillate into low inductance cables waveforms, this highly resonant condi¬
relatively minor; however, Fig. 6 with a variety of results. Threshold, tion occurs with every variety of cable,
shows the effects with one particular Stax, and Electro-Research designs but at different frequencies and "Q"
loudspeaker (which really exists) where behave violently, while others acquire factors depending on the source im¬
the deviation is dramatic. oscillation-caused colorations, usually pedance, the load impedance, and the
Just as the cables have an inductive either a hard, etched, high end or type and length of cable.
element, so do the amplifiers which warmth and thickness in the vocal The explanation for this resonance is
drive them. In tube amplifiers the out¬ range (due to low order intermodula¬ reasonably simple if we consider that it
put transformers provide the induc¬ tion sidebands and harmonics). takes a certain amount of time for the
tance whereas in most solid state Clearly things are happening above signal to travel down the cable. A
designs the designer has deliberately 100kHz; to display them I performed wave's velocity proportional to its
provided inductance in the form of a two tests on the cable samples. In the "characteristic impedance," Z Oi a value
coil for added circuit frequency stabili¬ first I swept the frequency output of a expressed in ohms and determined by
ty. The reason that damping factor Vtil resistive source from 100kHz to the inductance and capacitance:
(8í2/output Z) has been traditionally
quoted at low frequencies is not only
because much of our interest in damp¬
ing centers on the woofer, but also
because this coil destroys the damping
factor at high frequencies. In fact, ex¬
amples exist of solid state amplifiers
which quote damping factors of 500 or
greater at 20Hz but which have damp¬
ing factors on the order of 15 at 20kHz.
More recently a few designs
(Threshold, Audio Research, Yamaha)
have dispensed with output coils, giv¬
ing them more constant damping fac¬
tors across the audio band. The new
IHF test standards call for measure¬
ment of damping factor at all audio fre¬
quencies.
Again, this parameter must be
evaluated in the context of the system.
For example, most loudspeakers have a
considerable inductive component of
their own, which may easily provide
the desired case of amp Z + Cable Z <
speaker Z. Conversely, the source and
cable inductances become more impor¬
tant with loudspeakers whose load im¬ Fig. 7. High Frequency Test
pedance is resistive or capacitive, as Zo « - — <x velocity
yTC
especially with some electrostatic 40MHz while measuring the voltage at
designs. Our initial premise does not the other end of the cable with various When a wave travelling down a length
always hold either, for I have seen load impedances. In the second test, 1 of cable reaches the end of the cable, it
"poor quality" cables used to isolate an used the same source to send a 5/xS will do one of three things depending
amplifier from a reactive loudspeaker pulse down the cable for viewing at the on the impedance of the load. If there is
and thus improve the performance. In other end. (see Fig. 7). a high impedance load, so that Z¿, >
general, however, we are looking for Fig. 8 displays a condensation of Zc, the load will reflect energy
amplifiers with low output impedance over 50 photos showing the test positively back down the cable to reap¬
and cables having low series im¬ system's response to each variety of pear at the source (Fig. 9). If the load
pedance. cable and load. The unterminated "no impedance is less than the character¬
load" series shows the equivalent per¬ istic impedance of the cable, the wave
VERY HIGH FREQUENCY TESTS formance of loudspeakers with a is reflected back negatively; and if Z L
substantial series inductive compo¬ = Zo, then the wave is fully absorbed
Ordinarily a discussion of loudspeaker nent, while effects of resistive (8Q) and and none is reflected.
cables would stop here, at 100kHz, capacitive (,047^F) loading appear in This mismatch of load impedance to
where we could safely say that the per¬ the following two vertical columns. In cable impedance causes the resonance
formance is becoming negligible, if for the frequency response, we see large observed in Fig. 8, which we see
no other reason than that we cannot peaks occurring in the l-10MHz diminish in Polk and Mogami cables
hear this frequency (a concept disputed region. Remembering that we sent when they are loaded with 8 ohms, a
by some audiophiles). However, the 50mV down the cable from the value near their characteristic im-

pedance. By contrast, twin lead con¬ Note that these effects exist with all ed in bandwidth beyond 5MHz, the
ductors have a higher characteristic im¬ cables. The fact that only the newer, two effects met and resulted in various
pedance and perform better at multi¬ low inductance cables appear to affect forms of sonic problems, fuse blowing,
megaHertz frequencies with about 50 amplifier stability brings us to a point and worse. At Threshold we first ran
ohms load impedance. The effect of which justifies our examination of a across the problem with the mating of
twin lead cable on a 5/iS pulse with an cable's performance in regions which our 400A and Polk sound cables which
8 ohm load shows the effect of a load are simply not audible. The lower reso¬ caused fuse blowing (due to oscillatory
impedance lower than Z o, where the nant frequencies of the cables having cross-current conduction) with great
cable inductance cable rolls off the low Z o enter into the output band¬ regularity.
edges of the pulse, but where the 8 ohm width of the amplifier as it approaches After a period of confusion, Matt
resistance is sufficient to damp the its unity loop gain, and by altering its Polk and I realized independently that
ringing which occurs with Z¿ = 0 or Z L phase response cause oscillation at the the lack of a characteristic termination
= e» . Not so for capacitive loads as resonant frequency. was causing the problem. Polk
shown in the fourth column where Earlier amplifiers as well as some developed and patented a "damper”
another resonance altogether has currently available, having slower out¬ consisting of a ,047gF capacitor and 6Í2
developed due to the inductance of the put stages (less than 1MHz), did not in¬ resistor in series placed across the
cable and the capacitance of the load teract with these resonances because loudspeaker, while I used the same net¬
where: they occurred above the cutoff fre¬ work but with ,1/xF and 5 ohms. The
quency of the active devices by an results of this network are seen in Fig. 8
order of magnitude. However, as the where the resonance in the pulsed
2% • freq = newer cables decreased impedances waveform is damped out, restoring
VLC“ and as amplifier output stages increas¬ stability to an otherwise oscillating
Fig- 8
Polk Cable
100kHz to 5p.Sec. pulse 5p.S pulse 5[iS pulse 5[kS pulse

40mHz swept without w/8Q w/ZL = . 04 7fiF w/ZL =
signal without termination ZL = 8^1 5Q, .l^F
termination ZL = oo
impedance
ZL = oo

amplifier. Since Polk's commercial in¬
troduction of the damper circuit we
have found it cures oscillation pro¬
blems caused by the other exotic low
inductance cables. It is necessary
whenever a reasonably long length
(>3 feet) of low inductance cable is
mated with any wide bandwidth
amplifier. It interacts unfavorably with
twin lead conductors (Fig. 8) which re¬
quire higher impedance values (say,
.Ol/zF, 60fi); however, twin lead's
higher characteristic impedance and
resonant frequencies are in any case
unlikely to induce oscillation in
amplifiers now available.
With this much information, we
might think we have the subject nailed
down. However, we could easily in¬
stall the finest amplifiers, cables, and
terminating impedances and achieve
100 times the distortion of the amplifier
alone. Loose, dirty, or oxidized con¬
nections can, while measuring well Fig. 9. Wave "pileup" results in resonance
with an ohmmeter, cause high amounts audiophiles who, evaluating any cable
at certain frequencies, where the current is
of harmonic and intermodulation in the context of program source,
high and the voltage is low at the source,
distortion. When high distortion oc¬ amplifier, speaker, and listening room,
and where voltage is high and current is low
curs during an amplifier checkout at at the load. decide they can hear the difference? A
Threshold one of the first things we do few guidelines have emerged here, but
Frankly, I found it difficult to assess the final judgment belongs to the user.
is replace or tighten the cable from the the results except at the extremes of
amplifier to the load; we have thus All the special cables mentioned work¬
performance. For 10 foot lengths with ed well on the test bench and, given the
cured many "defective" amplifiers. properly terminated cables and
Copper and aluminum oxidize assumption that series impedance
speakers with inductive high frequency should be minimized, all of them work
quickly and oils from our fingers find characteristics, the differences between
their way to the conductor surfaces, better than 16 gauge wire. If, like many
low inductance cable and twin conduc¬ audiophiles, you have spent a small (or
causing poor contact; so on more than tor are extremely subtle and subject to
one occasion the dramatic improve¬ large) fortune on your hi-fi system,
question. With a low output induc¬ money spent for high quality cables
ment provided by an exotic cable has tance amplifier and a Heil tweeter
merely demonstrated the extremely and connectors is a reasonable invest¬
(whose impedance is a nearly perfect ment. □
poor quality of the previous cable's 6Í2 resistive) the difference was discer¬
long neglected connections. Wire con¬ nible as a slightly but not unpleasant REFERENCES
nections can age, and anyone wishing softening of the highest frequencies.
to accurately evaluate the newer Fulton or Monster cables were a clear 1. Brochures/Pamphlets:
cable's improved quality should first a. Fulton Speaker Cable. Fulton Musical In¬
improvement over 24 or even 18 dustries, 4428 Zane Ave. North, Minneapolis,
renew the contacts on his current set. gauge, though a little less subtle than I MN 55422.
Banana plugs and five-way binding would have expected, leading me to b. Mogami Speaker Cable. At press time we are
posts make excellent connectors as believe that the effort associated with unable to locate a manufacturer or distributor
long as they are kept clean; however, for Mogami.
heavier cables pays off in bass response c. Polk Audio Sound Cable. Polk Audio, 1205
while the connector's plated surface and in apparent midrange definition, South Carey Street, Baltimore, MD 21230.
resists corrosion, the wire to the con¬ especially at crossover frequencies. d. Ultra High Definition Speaker Cable. Audio
nector interface can become bad and The worst case load, the modified Source, 1185 Chess Drive, Foster City, CA
should be periodically checked, Dayton Wright electrostatics, 94404.
especially if it is subject to motion. 2. Hiraga, Jean, "Can We Hear Connecting
presented some interesting paradoxes: Wires?", Hi Fi News and Record Review. August
the extremely low impedance involved 1977.
OPINIONS showed the greatest differences bet¬ 3. Retsoff, Alexander, "Retsoff's Remedies,"
ween all the types of cables. However, Stereo. Fall, 1978.
At this point many audiophiles are 4. Klipsch, Paul, "Speaker Wire," Dope From
the best sound cables were not Hope, Sept. 13, 1978.
wondering, "Where are the listening necessarily electrically the best because 5. Pass, Nelson, "Loudspeaker Damping," Audio
tests?" I have listened to these cables on several amplifiers preferred the highest Magazine,
a variety of amplifiers (mostly my resistance cable. In one case, I had to 6. Gailus, Mark, "Speaker Wire: What Size is
own) and loudspeakers, including use 24 gauge cable to prevent tripping Sufficient?", Audio Forum. Aug.-Sept.,
Magneplanar Tympani 1 D's, MG II 7. Greinier, R. A., "Another Look At Speaker
the amplifier's protection circuitry. Cables," Speaker, The Boston Audio Society,
A's, modified Dayton Wright XG 8
Dec. 1978.
MK Ill's (as shown in Fig. 6) Cabasses; CONCLUSIONS 8. Gross, T. A. G., "Multistrand Litz Wire Adds
I have also heard some examples on "Skin" To Cut AC Losses in Switching Power
Dahlquist and Snell loudspeakers. Who am I to dispute the feelings of Supplies," Electronic Design. Feb. 1, 1979.

A Three-Enclosure Loudspeaker System
with active delay and crossover: Part 1
T he system described evolved over

years and out of experimentation
with many different configurations and
by Siegfried Linkwitz
driver for example should only be used
up to 1.5kHz to maintain wide disper¬
sion. This is indeed a popular
types of drivers and enclosures. Many crossover frequency but it is also well
people have contributed their ideas. It This detailed description of a within the critical frequency range of
is not "the ultimate loudspeaker", but multiple-driver loudspeaker fundamentals and lower harmonics of
it reveals enough about microphone many musical instruments. It is
placement and recording practices to
design is in three parts and
unavoidable that some change in the
suggest that the recording studio is the covers driver selection, radiation pattern is introduced around
next weak link in the chain between enclosure design, the active the transition frequency from a larger
original and reproduction. The few crossover, equalization and to a smaller driver. In the design
recordings with good spatial definition described a 100mm diameter driver is
are proof that the full potential of positioning. Sufficient informa¬ chosen which crosses over to a 25mm
stereo has not been exploited. Possibly tion is given to duplicate the diameter unit at 3kHz. The radiation
this potential has gone unnoticed system or to improve existing pattern change occurs therefore an oc¬
because hardly any commercial systems by equalizing the low- tave higher in a relatively less critical
loudspeaker reproduces the depth frequency range, but still care has been
perspective adequately, giving either a frequency response or adding a taken in the design of the crossover cir¬
diffuse or thin-walled stereo image. separate woofer box. cuitry to minimize irregularities in the
Every driver becomes more direc¬ transition region.
tional as frequency increases. The
radiation pattern of a rigid piston
mounted at the end of a long tube 1is
omnidirectional at frequencies where
the ratio of piston diameter d to the
wavelength X of radiated sound is
small, Fig. I. As d/X increases the on-
axis pressure increases but the pressure
at 45° off-axis decreases relative to it.
Experience shows that wide dispersion
of sound is desirable for natural
reproduction. Allowing for a max¬
imum 6dB drop-off at 45° off-axis re¬
quires that a driver be only used over a
frequency range where its equivalent
piston diameter is less than one
wavelength. This is an idealized
assumption because real drivers do not
behave exactly like rigid pistons but
the general principle still holds that
uniform, wide dispersion can only be
expected for frequencies where d/X < 1.
In all loudspeaker designs the
physical dimensions of driver, box and
room have to be compared to the
wavelength of the radiated sound to
determine whether a dimension is
acoustically small, as when d/X <0.5,
or large, Fig. 2. A 200mm diameter Photo A. Author's system on location. The satellite/midrange tweeter
r-n
Revised from articles first published in Wireless World. London. enclosures hang from the ceiling by curtain cord, and remain oriented by the
©1978 by IPC Business Press. All rights reserved. signal cable at the rear of each.

MULTI-DRIVER HAZARDS
Some designers try to obtain wide
dispersion or "omni-directionality" by
using multiple drivers, covering the
same frequency range. The fallacy in
this approach can be seen by compar¬
ing the radiation pattern of a single
driver to the resulting dispersion when
two of these units radiate together, Fig.
3. If the distance d between the driver
is greater than half a wavelength signal
cancellation can occur. The two out¬
puts will be 180° out of phase
whenever the path lengths from each of
the drivers to the listener differ by an
odd multiple of a half wavelength. As
frequency increases the two units move
relatively further apart (Fig. 2) and the
locations for which the outputs cancel
become more frequent. Such a system
can only be described as multi¬
directional. Additional drive units fur¬
ther destroy the phase coherence of the
direct sound output from the speaker
system. This imparts the illusion of
wide dispersion to all program material
but lacks the accuracy in sound
perspective which can be obtained Fig. 1. Directivity pattern for a rigid circular piston in the end of a long tube as func¬
from a single drive unit. tion of d/X (d is piston diameter, X is radiated sound wavelength). Wide dispersion
After establishing from the cone can only be obtained for frequencies where d/\ = 1.
diameter the highest frequency up to
which a driver can be used with good
dispersion, the lower frequency limit
will be determined from the cone ex¬
cursion capability of the drive unit and
the desired sound pressure level.
The radiation from the piston in a
long tube was found to be omnidirec¬
tional for low frequencies where
— < 1 Fig. 1. If the piston moves with
a peak-to-peak excursion a pp at fre¬
quency f and radiates into free space,
then pressure p at a distance r from the
source is:
p = ir 2po a
pp f
2d2
8^2 r
Normalizing the pressure with
respect to the reference pressure
p = 2X10' 4 /xbar yields an expression
for the more familiar sound pressure
level (d and app expressed in mm)
Fig. 2. Dimensions (d) of driver, box or room must be compared to wavelength of

SPL= sound to determine whether a dimension is acoustically small.
201og(p/p o) = — 86 + 401og f — 201og
r + 401og d + 201 og a pp
Assuming a pp is 6mm and f 70Hz a

direct pressure level of 83dB at Im can seem very high but crossing over to a such low frequency, a further increase
be obtained from the 100mm driver woofer at 70Hz will double the max¬ of approximately 6dB can be expected.
and 95dB from a 200mm unit. imally obtainable sound pressure Therefore from a stereo system with
( + 6dB) and because signals from the 100mm drivers a direct free-field s.p.l.
MAXIMUM SPL'S left and right channels of a stereo of about 95dB can be expected. Fur¬
These sound pressure levels may not system are predominantly in phase at thermore the normal listening environ-

ment is a semi-reverberant room where levels and low distortion. This moves be placed separately from the relatively
sound is reinforced by reflections from the crossover to the woofer to a less small midrange and tweeter enclosures
walls and objects. critical frequency range and a single and still be acoustically close because
Practical experience confirms that a large woofer can be used to cover the d/X is small. Further consideration is
100mm unit can handle program remaining frequency range below given to this aspect of the system
material down to 70Hz at adequate 70Hz. The large woofer enclosure can design later.
The frequency range below 70Hz
could be covered by two 200mm units
which will generate 90dB of direct
s.p.l. at 35Hz and Im or two 250mm
diameter drivers with 94dB s.p.l.
assuming 6mm peak-to-peak excursion
capability.
DRIVER CHOSEN
The particular drivers chosen for this
design are the 100mm KEF B110 low
frequency/midrange unit, the 25mm
KEF T27 tweeter and the KEF B139
woofer. A different unit like the KEF
B200 or some other make with ade¬
quate excursion capability and lineari¬
ty could be substituted for the B139.
There are of course considerations
other than dispersion and cone excur¬
sion which must enter into the selection
of a drive unit, such as smoothness of
frequency response, freedom from high
Q resonances, minimum phase
behavior, and low non-linear distor¬
tion. Unfortunately, manufacturers
publish scant useful data. Knowing the
magnet weight and flux density is of lit¬
tle help. With some training though the
ear can sort out those drive units that
seem worth further investigation and
the units chosen for this design proved
to be very satisfactory.
I was primarily guided in the selec¬
tion of drivers by the desire for wide,
uniform dispersion and crossover fre¬
quencies as high and low as possible.
Had I chosen to put emphasis on high
power output capability or lower non¬
linear and Doppler distortion then I
would have chosen larger diameter
drivers, or shifted crossover frequen¬
cies to a more critical frequency range.
Wide dispersion can only be obtained
from a small drive unit which will also
have higher distortion than a larger
unit. It appears though that
psychoacoustically the increased
distortion is outweighed by an improv¬
ed sound perspective which gives a
greater sense of realism. Some further
investigation of this subject is needed.
SPEAKER ENCLOSURES
Usually the size of a loudspeaker
enclosure is dictated by the required
low frequency response and efficiency.
Fig. 3. Use of multiple drivers to give wide dispersion produces interference effects I took a different approach here where
for d>0 (a) single driver (b) two drivers one wavelength apart, (c) two drivers four the enclosure is optimized for
wavelengths apart, all with same signal. minimum secondary radiation over as
Continued on page 16

A THREE-ENCLOSURE
SPEAKER SYSTEM
Continued from page 13
wide a frequency range as possible.
The low frequency output capability is
treated as a separate problem
My enclosure's purpose is to control
the radiation from the back of the
cone. A closed box design is the
simplest form of enclosure. If the
‘'largest box dimension is less than a
quarter wavelength at the highest fre¬
quency from the driver then the box is
acoustically small and the air volume
inside the enclosure will act like a
uniform spring. The box walls must be
made sufficiently stiff so the internal
air pressure changes will not deflect the
walls and cause secondary radiation.
The woofer enclosure can be made
small relative to the 70Hz maximum
frequency. It will therefore have no in¬
ternal air volume resonances and box
panel resonances can be pushed above
70Hz by crossbracing.
The BllO's enclosure presents
greater difficulties because of the wider
frequency range covered. The volume
inside the enclosure will exhibit cavity
resonances which have to be elimin¬
ated. Acoustic energy is stored
whenever one of these resonances is ex¬
cited and gradually released after the
excitation has been removed. Most of
this acoustic energy exits through the
cone: the speaker regurgitates its own
characteristic box sound. Fortunately it
is not difficult to dampen cavity
resonances. The technique of filling the
enclosure with long fibre wool is well cited to vibrate by the internal air Fig. 4. Loudspeaker enclosure dimensions.
established and very effective. 2,3 As there is little stereo information below
volume pressure changes, but more
Another form of undesired second¬ serious is the direct transmission of the the 70Hz limit of the enclosures (above), a
ary radiation comes from the enclosure mechanical vibration of the driver's center woofer covers the remaining range
walls themselves. The walls can be ex¬ down to 25Hz (below). The revised
cone to the enclosure. The walls then
enclosure (above left) has minimum width
radiate the transmitted mechanical and optimum tweeter position for reduced
energy as sound, particularly when its diffraction from cabinet edges. The gum
Enclosure Design Objectives frequency coincides with a panel plywood construction allows only
resonance. minimum bevelling.
' ’Narrow frontal area for op¬ WALLS & MOUNTS
timum horizontal dispersion;
tweeter mounted directly above It is not unusual for more energy to b.e driver to the enclosure. Vibra tion-_
, the midrange unit. radiated~Hirectly from the enclosure mounting the driver to the enclosure
' ’Box edges rounded to reduce grails than from the cone at resonance with some form of compliant suspen¬
scattering. frequencies. If, for example, the sion will significantly reduce the wall
7 ’Drivers mounted to minimize vibrating enclosure surface has ten excitation, but it poses some difficult
direct transmission of vibration times the area of the cone then its mechanical design problems. The
to the panels. acoustic output will already equal that natural frequency of the driver moun¬
v ’Panel resonances attenuated of the cone if it has only one tenth of ting should be well below the
with damping layers of sufficient the cone excursion. The output of most acoustical output frequency to avoid
stiffness and loss. loudspeakers is colored by the radia¬ frequency response irregularities. The
•Air cavity resonances at¬ tion from the enclosure walls. mount must seal the enclosure air tight
tenuated with filling materials to It has been verified experimentally and provide sufficient mechanical sup¬
eliminate delayed re-transmission that vibration coupling between the port for the driver. Another approach
through cone. driver and the walls occurs primarily might be to enclose the box to which
through the rigid mounting of the the driver is mounted by a second box

Siegfried Linkwitz became interested
in sound reproduction as a hobby,
recognizing the many dualities be¬
tween the propagation of microwave
and acoustic fields. “A 1GHz elec¬
trical wave and a 1kHz acoustical
wave have about the same
wavelength. " After modifying and
equalizing several commercial
loudspeakers, he set about designing
his own system —out of frustration
with available units.
He is section manager for signal
analyzer developments with Hewlett-
Packard. He has been program
manager for the 8566A high perfor¬
mance microwave spectrum analyzer
and he's been involved with the
Fig. 5. Sound scattering from the sharp comers of a loudspeaker enclosure design of signal generators (608E), a
producing a smeared out transient behavior of the system. vector-voltmenter (8405A) and was
project leader for the 8554L RF
analyzer. He joined Hewlett-Packard
Company in California as a develop¬
ment engineer for r.f. test equipment
following graduation in 1961 from
Darmstadt University.
Fig. 6. On-axis frequency response of a point source mounted in different

enclosures shows smoother response of sphere (after ref. 6).
avoiding all rigid coupling between the

two. filling. The lowest cavity resonance oc¬ original enclosures had panels which
For the design described here I chose curs at 600Hz, the next ones at 800, were quite stiff. The lowest panel
a single, totally enclosed box, Fig. 4. 1000, 1200Hz etc. the resonances are resonance was observed at 430Hz using
The B110 driver was attached to it with measured easily with a small om¬ a magnetic phono pickup in a
soft rubber grommets in the four nidirectional electret microphone pro¬ makeshift tonearm as vibration
mounting KóIês~oF the basket which truding into the box and applying a transducer. A 430Hz tone was
slide on the mounting bolts. The seal¬ sweep signal to the B110. Filling the measured to decay b y 40dB in the
ing foam gasket is barely compressed. boxes rather tightly with long-fibre rather long time of 120msec indicating
Comparing this to a directly mounted wool attenuates all the resonances to a a Q of 36 which is typical of undamped
driver by tapping on either basket in¬ smooth frequency response inside the wood panels. By applying damping
dicates a significant reduction in coupl¬ box and at the outside cone surface. material the decay time could be reduc¬
ing to the box. Some further investiga¬ My original enclosures were con¬ ed to 40msec corresponding to a Q of
tion of this subject is in progress. structed of 20mm mahogany board 8.4 at 300Hz. This is a low Q for a
with the internal surfaces coated with mechanical structure.
WOOL & TAR tar-based damping layers. I have since The basic difficulty consists in fin¬
The relatively small size for the buiTFnew versions using 6mm plywood ding a damping material which has
B110/T27 assembly has the advantage with a 15mm damping layer consisting adequate stiffness to control the mo¬
that the internal air volume resonances of a 3:1 mixture of water based asphalt tion of the panel and which at the same
occur at high frequencies where they émulsion (Henry's 107) and sand, time has sufficient loss to dissipate the
can be damped effectively with wool which gave optimum results. 4 The mechanical vibration energy as heat. A

Photo B. Overall view of the author's system. The comfortable chair evidently
gets moved to a sonically less active position during listening sessions.
better match between available cone will initially be blocked by the For the given design the frequency
materials is achieved by using thinner panel and then encounter an abrupt range extends down to 70Hz where the
walls of correspondingly less stiffness transition where it ends, Fig. 5. A se¬ B110 has its enclosed resonance. As
so that a softer damping material can cond wave is generated at the cabinet there is little useful stereo information
reduce wall flexures. A simple and edge which will interfere with the below this frequency a single center¬
quite revealing test is to knock on any original wave. If a pulse is radiated channel woofer box can cover the re¬
box to hear how dead it is acoustically. from the T27 then a secondary pulse of maining range down to 25Hz, Fig. 4.
A small box presents a small obstacle lower amplitude is generated at a time This is built with internal bracing to
to omnidirectional sound propagation. t = d/c = 260gS later, the original stiffen it and to push panel resonances
This is a clearly audible advantage pulse is smeared out. This scattering of to frequencies above 70Hz. In addition
when properly placed in the room. As sound should be avoided by 25mm-thick heavy felt is glued to all
the box is only marginally wider than eliminating sharp discontinuities panels to reduce direct transmission.
the B110 driver it can be assumed that through bevelling of the cabinet edges. 5 The box is loosely fitted with long-fibre
the radiation pattern for a piston at the Fig. 6 shows the on-axis amplitude wool. □
end of a long tube is an adequate first response of a small driver mounted in
order approximation to its sound the center of a cube and of a sphere. 6
dispersion, Fig.l. The T27 tweeter is Clearly the sphere with its surface
REFERENCES
mounted as close as possible to the gradually receding from the source
1. Beranek, L. L., Acoustics, McGraw-Hill 1954.
B110. At the crossover frequency of produces a much smoother response 2. Bailey, A. R., Non-resonant loudspeaker
3kHz the spacing corresponds to a than the cube with its sharp edges. enclosure design, Wireless World, October 1965,
distance of one wavelength. In the ver¬ Therefore the larger the box relative to p. 483.
tical plane therefore the radiation pat¬ the driver the more closely should it 3. Bradbury, L. J. S., Use of fibrous materials in
loudspeaker enclosures, JAES, Vol. 24, April
tern at the crossover frequency should approach the shape of a sphere. It 1976, p. 162.
follow the previously discussed follows that the midrange/tweeter 4. Harris, C. M. and Crede C. E., Shock and
behavior of two drivers contributing enclosures might be further improved Vibration Handbook, 2nd edition. McGraw-Hill
equally, Fig. 3b. by reducing their size and constructing 1976.
a more curved driver mounting area. 5. Heyser, R. C., Energy-time test. Audio, June
RESPONSE RESULTS 1976, p. 74.
Papier mache, cardboard or epoxy 6. Muller, G. G., Black, R. & Davis, T. E., Dif¬
Ideally the sound from the T27 should fiberglass with damping materials ap¬ fraction produced by cylindrical and cubical
be able to disperse freely in all direc¬ plied to it rather than wood might be obstacles and by circular and square plates,
tions, but because of the large width of more suitable materials for the un¬ JASA, vol. 10, July 1938, p. 6.
the front panel relative to the cone conventional contours of such an
diameter a wave emanating from the enclosure.

An Electrostatic
Speaker System
Part 1
S ince the publication of my speaker

design in The Audio Amateur
(Issue #4, 1975, p.18), hundreds of
by Roger R. Sanders
(hereafter called "charge") can be easily
placed on it or removed from it by an
audio amplifier. If the non-conductive
readers have contacted me and scores rather than at just one point as in a film (hereafter called diaphragm) has
have built speakers. From this and dynamic speaker; therefore the driving 2000 volts (2kV) positive on it and the
from continued investigation on my surface is scarcely distorted. Because of amplifier is causing the perforated
part several new designs have been the lack of moving mass, the driven metal plate (hereafter called "stator")
built and tested and many small bits of surface response is exceedingly linear to have 3kV negative charge, then the
worthwhile information have been and therefore essentially perfect fre¬ diaphragm is going to be strongly at¬
learned. The purpose of this series of quency response should be available tracted to the stator. It will be more at¬
articles is to make it possible for the from the system. tracted to a 3kV negative stator than to
average audiophile to construct his a zero volt one. If our stator is chang¬
own speaker that will exceed the per¬ DRAWBACKS
ing voltage as directed by our audio
formance of the finest commercial Well, it certainly sounds good; there system amplifier, then the diaphragm
systems. must be a catch or else everybody will move back and forth and you will
I shall discuss theory, construction, would be using it. There is a catch—in hear music as the air moves through
and custom tailoring. If you build the fact several. To understand one must the holes in the stator.
proven design without modification, I understand the speaker's working prin¬
can virtually guarantee you superb ciple. Unlike the magnetic forces that HIGH VOLTAGE
results. I'll present various modifica¬ drive a dynamic speaker, the ESL is The principle is simple, but the solu¬
tions to allow for individual prefer¬ driven with electrostatic attraction. tion to the problems are not. Let's ex¬
ences, but all modifications have been This is the force you feel when you run amine the problems and possible solu¬
tried and studied and I have chosen the a comb through your hair on a dry day tions.
best combination of compromises for and feel the comb attract the hair on The first problem is that we need
the "ultimate" system; modifications your arm or have little bits of paper high voltages. This means the speakers
will degrade performance in one area stick to it. have to be built with insulation in mind
or another. The force is developed by high and some safety precautions are need¬
Electrostatic loudspeakers (hence¬ voltages (problem #1), and it is very ed. Conventional audio amplifiers
forth called ESL's) offer improved weak (problem #2), it is only available generate only approximately 100 volts
detail and clarity compared to conven¬ across very short distances (problem of output swing (peak to peak), and we
tional dynamic loudspeakers. The #3), and tends to migrate across the need several thousand volts. The
reasons for this are clear. Essentially, surface of the charged object (problem diaphragm voltage is reasonably easy
the dynamic loudspeaker has a great #4). Electrostatic force decreases with to develop: we used about 2kV and
deal of mass compared to the air it is the square of the distance which causes essentially no current is required. An
driving, while the ESL has essentially non-linear motion (problem #5). easy way to get this voltage is to con¬
no mass compared to the air it is driv¬ We can take a thin piece of non- nect a surplus copy machine power
ing. This offers much better transient conductive film such as the cellophane supply to the diaphragm.
response; more important, no reso¬ found around cigarette packs, and The amplifier voltage problem is
nances are present above the ESL's fun¬ place several thousand volts on it. It usually solved by driving the speaker
damental resonance because the will have an electrostatic force if placed through a transformer with a high
speaker is air damped. within a few thousandths of an inch turns ratio in order to step up the
Sonically this means the sound is (hereafter called "mils") of another ob¬ voltage. Transformers don't really like
very smooth as well as detailed. The ject. If that other object is conductive, this because the speaker does not draw
ESL is driven over its entire surface, like a perforated metal plate, a voltage current: electrically the transformer

sees the speaker as a capacitor. This the sound pressure levels (hereafter To compensate one can equalize the
makes for problems with transformer called SPL's) to decrease even though bass and make the speaker go through
resonances, goofed-up frequency the diaphragm motion is linear. This is large excursions, which is difficult to
response, and instability in some caused by "leakage" of the pressure do because of the weak forces, small
amplifiers. Good transformers of this waves from one side of the speaker to distances, and large voltages required.
type are also rather expensive. the other. Essentially the compromise here is that
Several people have tried to build a To demonstrate, pretend a canoe you get bass at the expense of SPL's.
high voltage amplifier directly con¬ paddle is the diaphragm and put it into Alternatively, we can increase the
nected to the ESL stator. These are not a tub of water and pretend that it is air. speaker area. This works well up to a
without problems, the biggest one be¬ Move the paddle rapidly back and point; after all, if the speaker complete¬
ing instability (see my article in The forth in the water and you get big ly seals off the end of the room, no air
Audio Amateur, issue #1, 1976, p. 12, waves (loud high frequency sound). can get around it. In free space, the
and David Hermeyer's articles in the Now move it very slowly the same speaker would have to be at least 30
same publication, issues 2 and 3 of the distance back and forth: you get little feet across the smallest dimension to
1977 series). They are also costly to waves or none (weak bass). You can reproduce 30Hz linearly—see the sizes
operate because they are generally see the water has more time to move required in Fig. 1.
class A devices, and they are quite an around to the other side of the paddle Not only is it impractical to make
undertaking to build. With the correct and cancel out the pressure waves at such large speakers, but they are also
transformer, I believe a transformer low frequencies. The same is true of an very difficult to drive: the load they
coupled system with a conventional ESL and air. present to the amplifier and
transformers (if used) becomes un¬
manageable. Imaging also becomes a
problem with the sound arriving from
such large areas. The most practical
way to get the bass response required is
to use conventional dynamic woofers.
A good woofer in a good enclosure
(either a transmission line or a large
horn system) can be made to sound
clearly superior to a full range ESL in
direct A-B tests.
WHICH IS BETTER?
Now, I realize that comment is going to
make the "techno-freaks" say I am
crazy: a crossoverless ESL just has to
be superior to a hybrid system. Well,
"techno-freaks" generally don't have
hands-on experience and so aren't
realists; I know, I used to be one. The
point is, just because theory says
something should be better doesn't
PHOTO BY REG WILLIAMSON. mean it is better.
Photo A. The author's system including two transmission line woofers with KEF drivers I can't tell you how much time I have
and four 2x3 electrostatic panels. The speakers are driven by two stereo amplifiers with spent with readers who insist on
electronic crossovers and equalizers. operating the ESL's full range or at
least down to 100Hz because "it should
amplifier is still the best compromise. FIG.I sound better." It is not better. I have
More discussion of this later will make demonstrated in A-B tests on several
my reasoning clear. occasions that the full range system is
inferior to the hybrid system. Tests
SHORT FORCES
were done with the ESL equalized to
The second problem: the forces are achieve reasonably linear bass re¬
small; the third: they are available only sponse to about 70Hz (one just can't
across very small distances. It therefore get a dipole of this design to produce
becomes impractical to drive the dia¬ really deep bass). I used transmission
phragm over large excursions, which line woofers rolled off at 70Hz to ap¬
severely limits how loudly the speakers proximate the ESL response, alone
will play. The frequency spectrum of with 18dB/octave crossovers with an
music is such that a great deal of effective crossover point at 400Hz. 1
energy is needed in the midbass and matched levels with an SPL meter.
lower midrange (100-500Hz by my All listeners were clearly biased in
definition). This is particularly unfor¬ that they expected the full range ESL to
tunate because also in the lower fre¬ FREQUENCY(Hz) easily outperform the hybrid system,
quencies we run into problem #6, Fig. 1. The wavelength of sound versus fre¬ particularly with regard to mid-bass
phase cancellation or the tendency for quency. detail and clarity. At no time during

the test were the ESL levels so high as parent distortion, in a system I can af¬
to produce obvious distortion. The test ford and would be able to put in my
results converted the listeners without living room. I require SPL's in excess of
question to thinking the hybrid system 103dB at my listening location four
was clearer and more at ease than the meters from the speakers. Frequency
full range system. Of particular interest response must be essentially linear
was the fact that the critical mid- from 20Hz to 20kHz. I find my system
bass/low midrange was actually is adequate to these tasks; if you wish
clearer with the hybrid system than to make modifications, please do, but
■■■■■■■■
with the ESL operated full range. The the suggested system is proven.
logical question is, "Why?"
It's difficult to be certain, but my BOXED ESL'S?
best guess is the ESL diaphragms are I tried one other way to get full range
being driven so far that the diaphragm from an ESL: I enlcosed it in a box to
is changing tension significantly and in isolate the front from the back and pre¬
a non-linear fashion, and also that the
amplifier is being asked to deliver
vent low frequency cancellation. This
did not work. The enclosure produced mURNCÜMIMII
tremendous voltages which it cannot all kinds of colorations from resonance
nmim
easily do. This is not to say the full effects, but worse still, the bass was not
range ESL was not good; it was very there. After some head scratching and
clear and detailed, considerably more measurements 1 realized the problem
so than conventional box speakers. was the ESL's fundamental resonance.
The point is that if the woofer system is Because the ESL was rather small Fig. 2. Top photo, a storage oscilloscope
done right, you can have your cake (about 18" X 34"), the resonance was at trace of what ‘/s-octave pink noise looks
and eat it too—you can have clear bass about 300Hz. Without the mass of a like fed into a real time analyzer measuring
and high SPL's using a hybrid system. the unequalized response of Sanders' ESL
conventional dynamic woofer, the
panels indoors with a mike at five meters.
system was simply reaching fundamen¬
BIG PICTURE The lower photo is the effect of an equalizer
tal resonance and the output was fall¬ on the panels with a crossover at 120Hz.
I must emphasize the term "system." I ing rapidly below that. The large baseline marker in both photos is
have spent nearly as much time work¬ It appears it is not possible to 1kHz, and the horizontal markers represent
ing with woofer systems as I have with achieve bass in a reasonably sized ESL 5dB.
ESL's. They are equally important. with the use of an enclosure. This does
Crossovers are also important. You not mean this is not a useful technique, (a 20kHz wave is about one quarter of
can't take an iron core inductor, and readers who wish to pursue it are an inch long). Most audiophiles believe
passive, high level, 12dB/octave, con¬ referred to Peter Walker's famous ar¬ this directionality is undesirable and
ventional crossover and make a clean ticles, "Wide Range Electrostatic therefore a problem; this point is
hybrid system: you must use "elec¬ Loudspeakers," in Wireless World, highly debatable. There are those who
tronic crossovers" at 12dB/octave or, May, June, August, 1955. The first ar¬ believe a highly directional loud¬
better yet, 18dB/octave and bi-amp the ticle deals with the concept of constant speaker is superior to wide dispersion
system. You can't take a little, boomy, charge operation which solves our pro¬ types. [See: J. Newman, "Dipole
cheap "woofer in a box" and get a good blem #4 involving charge migration. Radiator Systems," Journal Audio
hybrid system, either; it really takes The second covers enclosures for Engineering Society, Jan/Feb 1980;
transmission lines or horns to do the ESL's, and in the third, on Vol. 28, Nos. 1 & 2, pp. 35-39.)
job. loudspeaker/room relationships, Personally, I prefer the highly direc¬
Now, this doesn't mean that none of Walker makes a very strong case for tional types: they minimize room
the above will work. I have had using the dipole radiator. resonance effects, image better, and are
readers who listened only to chamber Problem #5, non-linearity of the easier to build. Even with wide disper¬
music or gentle popular music and diaphragm motion is solved by what is sion speakers one needs to be an equal
were ecstatic over the sound of their commonly called the "push-pull" ESL. distance from both, so one must listen
full range systems, but they didn't need This requires placing a second stator to music seriously at only one location
deep bass or high SPL's. Others are on the other side of the diaphragm. in the room anyway. Background
finding so much improvement over Now when we place voltage on one music listening off-axis is no problem
their conventional systems even when stator we place an opposite charge on because the sound bounces around the
using poor woofer systems that they the other. As the diaphragm moves room.
are quite satisfied to continue. But toward one stator and is attracted non- The image from a conventional box
most readers are interested in the best linearly, it moves away from the other loudspeaker can be described as com¬
possible reproduction, and they will whose force is decreasing non-linearly ing from the speaker, while the image
get such reproduction if they are will¬ in a mirror image fashion, the forces on from really superb conventional
ing to listen to experience, do what has the diaphragm are then linear. dynamic loudspeakers and from wide
been proven to work, and not let their dispersion ESL's can be described as
"techno-freak" ideas stand in their BIPOLAR BEHAVIOR coming through the speaker: it has
way. Large planar speakers are highly direc¬ "transparency." However, the sound
My personal standards are severe: I tional. This is a physical characteristic fron highly directional ESL's is dif¬
want to be able to play my live concert of large areas whose smallest dimen¬ ferent from all the others. It appears
recordings from uncompressed master sion is as large or larger than the wave¬ not to be associated with the speakers
tapes at "row A" levels, with no ap¬ length of the sound being reproduced at all. The image seems to float in the

room; it is three dimensional, and its This is not completely satisfactory in eventually burns pinholes in the
exact location depends upon recording that the dispersion is not uniform: each diaphragms. I have probably several
techniques used in producing the narrow cell has its own narrow beam hundred pinholes, but the speakers still
source material. and moving about in front of the sound fine, and 1 cannot detect this has
Some very unnatural effects are speaker causes a "Venetian blind" ef¬ harmed their performance in any way.
created this way. For example, a pop fect. You can minimize this by using 1 have since found some grille cloth
vocalist recorded with a great deal of small angles between the strips; I that doesn't mess up the sound, and
artificial reverberation and a single recommend less than eight degrees. would use it in the future. It is a plastic
microphone (a monophonic recording Robert Unterbrink and I have fiber cloth marketed by Mellotone.
passed off as stereo) is "panned" to the developed a method of forming a While it is quite expensive, the smooth
center of the sound stage. The singer smoothly curved ESL cell with perfect¬ fibers do not harm the high frequency
will appear to be singing slightly inside ly uniform dispersion. Patents are pen¬ response and the weave is fine enough
a very deep 30' diameter pipe; the im¬ ding on the design, and we expect com¬ to keep out bugs. There is absolutely
age will be huge and very unrealistic. mercial development. The design of no need to use dust covers. 1 even sand
On the other hand, a singer recorded in this speaker will be presented at a time balsa wood in my listening room and it
a concert hall with a pair of micro¬ when I can devise good fabrication gets into everything, but the speakers
phones in proper stereo will appear to techniques for the home builder. do not seem to care. Occasionally I
be of normal size and sound entirely Reliability has been a problem with blow out dust with compressed air or
natural. commercial ESL's. I am pleased to say gently vacuum them, but I do this only
IMAGE, SPEAKERS & MIKES my original speakers, built in 1974, are for appearance, not for performance
still running fine and I have never reasons.
A common complaint is that planar replaced a diaphragm since working
speakers "cause" soloists to sound 10 out their "bugs." I attribute this to the LINEARITY PROBLEMS
feet wide. This is simply not true. availability of better materials such as The typical commercial ESL does not
Planar speakers accurately reproduce polyester film. I did not put grille cloth have linear frequency response even
the poor recording techniques com¬ on my speakers because tests indicated though theory indicates the diaphragm
monly used. Naturally recorded music this ruined their high frequency should be very linear. The reasons for
sounds natural through them. If you response. Foam grille didn't hurt the this are numerous. First of all, the high
insist on wide dispersion speakers, you frequency response but messed up the frequencies have a narrower dispersion
can easily make them by angling phase response. I have therefore had than does the midrange, which effec¬
several ESL strips to each other around some problems with insects getting into tively makes the speaker sound
the imaginary surface of a cylinder. the speakers and causing arcing, which Continued on page 24
audio
Loudspeaker
Components
HORNS • WOOFERS • MIDS • TWEETERS Q

3238 Towerwood Dr. Dallas, TX 75234
AT TREMENDOUS OEM SAVINGS. ALTEC. ELECTRO¬ 05
VOICE. PHILIPS-AMPEREX, PEERLESS. POLYDAX. ENCLOSED IS $1.00, PLEASE SEND CATALOG
ISOPHON. AND MANY OTHERS. HUGE SELECTION OF
X-OVER COMPONENTS. LARGE INVENTORY FOR IM¬ name
MEDIATE DELIVERY. GOOD SELECTION OF MUSICAL IN¬ address

STRUMENT LOUDSPEAKERS. CATALOG CONTAINS VAL¬
UABLE X-OVER INFORMATION. AND TECHNICAL DATA ON city
SPEAKERS. SEND $1“’ TO: SRC AUDIO DEPT.SB state zip code
AN ELECTROSTATIC
SPEAKER SYSTEM: Part 1
"bright." The low frequency phase
cancellation previously mentioned
begins at a considerably higher fre¬
quency than you might expect. For ex¬
ample, the minimum dimension of a
cell of my design is 24”. Looking at Fig.
1, it would appear the wave length of a
500Hz tone would equal the speaker's
minimum dimension and that the low
frequency rolloff would begin there.
Well, it obviously begins much higher
than that: it becomes noticeable at
2kHz.
Fig. 2 demonstrates early tests with a
spectrum analyzer: the frequency
response (1 cm. from diaphragm
center). Fig. 4 is a composite of theory
and objective measurements that I feel
accurately reflects the true behavior of
the ESL alone at far field (my listening
location).
Fig. 5 reveals the techniques I used to
flatten the frequency response. The
ESL's bass is severely attenuated and is
highly irregular because of phase
cancellation and fundamental
resonance. The best way to correct this
is to simply avoid using the ESL in that
region. This also makes high SPL's
possible, as I discussed above.
The ESL response above 400Hz is
still rising, and the system will still
sound bright and thin unless something
is done to correct this. With modern
electronics it is possible to make a high
quality equalizer that presents a mirror
image depression to the high frequen¬
cies. I described this in my 1975 article,
and I am sorry to say many readers felt
this would degrade the high frequen¬
cies and so did not use it. The fact is the
equalizer attenuates the high frequen¬
cies, but then the entire ESL section
level is raised until the highs are back Fig. 5. Composite system response.
to normal balance. The net effect of
this is that the midrange is elevated to
FIG 6
its proper level and the sound is now
properly rich and full.
EQUALIZATION: YES OR NO
"Techno-freaks" will be loth to put an
equalized signal into the speakers
because they believe it degrades the
sound. However, there is simply no
question that the system is far more
natural with the equalizer, and once
one hears the system with the
equalizer, no listener has been willing
to do without it. It adds no detectable
sonic distortion.
I went to considerable effort to
minimize the "modern” distortions Fig. 6. Schematic for one channel of an equalizer with gain for the electrostatic
such as TIM and SID. I discarded the panels.

earlier equalizer in favor of a much in order to isolate it from the im¬ 20dB, so you will probably have
simpler one utilizing the very high per¬ pedance loading effects of the enough spare gain to use it unless you
formance LM318 IC, and placed the associated equipment, which had been listen only at low levels. I personally
equalization in the ICs feedback loop a problem in some installations with don't use a preamp at all, just a passive
the earlier passive system. In addition, attenuator for level control of high
FIG 7 the equalization stage has 18dB of gain, level circuits, so I had to have addi¬
IOK
making the system's sensitivity equal tional gain in my system offered by the
IN °-^AAr to a typical loudspeaker system. gain section of the equalizer. Fig. 8
Fig. 6 is the revised schematic of the shows the frequency response of the
zyx 04 m F eq/gain stage. For those of you who eq/gain section alone.
have preamp gain to spare and are able
j 5K
I f HI FREQUENCY to measure electrical frequency CROSSOVERS
I CORNER ADJUSTMENT response, I have included the schematic Electronic crossovers are a must.
of a passive equalizer (Fig. 7) with an Beside the fact that they are audibly
adjustable high frequency corner superior to passive high level
Fig. 7. A passive equalizer with 20dB inser¬ which will allow you to "tweak" it to crossovers, the ESLs capacitive
tion loss for use before the power amp driv¬ the correct frequency response curve. characteristic makes it very difficult to
ing the ESL panels. The insertion loss of this is nearly get proper response from passive
crossovers. You have several crossover
choices. You can make passive 6dB/oc-
tave crossovers, but the slopes are real¬
ly not steep enough even though in
theory they are superior to any other
type. I have used them and they work,
and I have included a suitable
schematic in Fig. 9. This, like the
passive equalizer, has a large insertion
loss so if you use it you had best have
an extra 20dB of gain available. You
will also have to "tweak" to the correct
Fig- 8- Response curve of the effect of the active equalizer/gain circuit shown in Fig. 6. crossover points based on the im-
BUILD THE WORLD’S BEST

Fried model c SATELLITE
No gluing,
drilling,
cabinet assembly
or finishing
STUDIO BLACK OILED WALNUT
For as little as $570 a pair
$570.00 a pair in black The rounded pyramidal cabinet of the Fried MODEL C approaches
the classical expanding sound source ideal. The cabinet design
minimizes internal standing waves and allows for minimum dif¬
$695.00 a pair in oiled walnut fraction effects thus providing excellent depth and localization of
sound. The HF driver is an ultra advanced, ferro fluid cooled, soft
Do not send cash. dome, capable of sustaining 900 watts at a 10msec burst. The MID/
Money order or cashier’s check allows for BASS driver has a new, low profile, bextrene cone to reduce
immediate delivery. Check delays shipment. Freight colorations normally found in paper.
■
collect. Send your name, address with check, This uniquely designed speaker may be used alone as a mini-monitor
money order, or cashier’s check to: or crossed over as low as 75hz to a sub woofer.
These cabinets have full rabbit joints for rigidity, stength and air
ICG 8489 West Third Street
Suite 54
Los Angeles, CA 90048
tightness; high-density 3/4” particle board; screw holes pretapped
for crossover mounting; fully routed and drilled with installed T nuts
for driver mounting; black "full dress” screws; and come in either
Interface Communications Group professional “Studio Series" black or oiled walnut finish.

overheat. The solution is simple: install
AN ELECTROSTATIC
a small capacitor between the
crossover and the amplifier. For the
ESL amp, .OlgF is adequate. The
pedances you are driving with it. capacitor should be a non-polarized
The Old Colony crossover kit is type such as polycarbonate, polyester,
12dB/octave and active, works very mica, or glass; do not use electrolytics
well and has no insertion loss. It is also for coupling. If the problem also exists
very cheap, and being quite small you in the woofer amp (usually noted
can build it and the eq/gain stage on because the woofer moves away from
the same small chassis and drive them C. C2
500Hz 0.0039gF 0.0056gF
its center and stays there when the amp
by the same power supply. Several is turned on), then use a .5gF capacitor.
authors have tested crossover slopes 1kHz 0.0022^F 0.0022gF
2.5kHz 560pF O.OOlgF TRANSFORMER MATCH
and the general consensus of opinion is
that odd order Butterworth filters are Fig. 9. A passive 6dB/octave crossover Finding suitable matching transformers
better than even order ones —in other with adjustable high pass gain control. is difficult. The problems are two: first,
words, each "order" in the filter is 6dB, Values in the chart set crossover points at finding high enough turns ratio
and therefore a 6dB filter is 1st order, various frequencies. transformers is not easy, and second,
12dB is 2nd order, 18dB is 3rd order, high frequency response is usually
24dB is 4th order, etc. In theory the poor. The problem with the frequency
Old Colony crossover should not be and detail in the bass and lower response seems to be leakage induc¬
best because it is even order. However, midrange are definitely not improved. tance in the transformer resonating
I could detect no obvious problems Fig. 11 is a block diagram of the
system as well as a schematic diagram with the capacitance of the ESL. This
with it with one exception: I could not
of the correct connections for puts a 3 or 4dB peak in the response
determine the correct phase between
transformer coupling the ESL to a con¬ with a rapid rolloff above that. I have
woofer and ESL.
ventional amp. If you want high SPL's, tested many transformers and find few
Clearly, there should be a difference,
of them will perform above 8kHz with
and with 18dB/octave filters there was use a modern "super amp"—not that
ESL loads. This problem worsens with
one. The upper bass/lower midrange the speakers require a lot of power, but
larger speakers, as the turns ratio in¬
sounded properly full in one phase and rather that they need a lot of voltage
creases, and as the transformer's power
thin and incoherent in the other phase. and large amplifiers have higher
rating increases.
One phase was obviously correct. I voltage power supplies than low
What you need is a transformer with
therefore feel 18dB/octave crossovers powered ones. The amplifier should
both a turns ratio and frequency
are superior to 12dB/octave cross¬ run no warmer when driving the
response as high as possible. In two
overs. However, the fact remains that speakers at high levels than it does
years of searching 1 finally found
unless you make a direct comparison when it idles. If it gets hot, that in¬ transformers with 44:1 turns ratio with
the Old Colony units have no detec¬ dicates you are getting DC in the out¬ a frequency response of 20-50kHz.
table problem. A complete article con¬ put, usually because the amp is one of They will drive an ESL with a capacity
the DC coupled types and is getting a
cerning these crossovers appears in The of 2400pF essentially linearly to 20kHz.
Audio Amateur, issue #2, 1972, p. 10. little bit of DC at the input from one of You can buy them through me —see the
I personally use crossovers by the crossover IC's.
Most IC's have a small DC offset, end of this article.
DeCoursey, who offer them in various You may be able to use audio output
forms and slopes. I simply bought usually around .05 volts. When this is
transformers from a good tube
completed boards and installed them in amplified, you may have a volt or so of
amplifier. Ideally they should have a
my eq/gain stage. They are about two offset in your power amp. Into essen¬
very high primary impedance —8kQ if
inches square and cost about $40 per tially a dead short such as the primary possible. Install them backwards: con¬
channel. I did not make my own of a low impedance transformer, lots
nect the transformer's 4Í2 taps to your
because the precision components re¬ of current will flow and the amp will
quired are not readily available to the
amateur constructor. (DeCoursey
Engineering Laboratory, 11828 Jeffer¬
son Blvd., Culver City, CA 90230.)
PERFORMANCE
Fig. 10 is the total system electrical
response with the eq/gain and
18dB/octave crossovers. Note that the
correct crossover point is 460Hz. When
this is combined with the equalization
the actual crossover point is reduced to
400Hz which I have tested extensively
and found to be optimum for this
system. Please resist the urge to cross FREQUENCY(Hz)
over lower: this causes a large depres¬
sion in the frequency response in the Fig. 10. Total system electrical response with the author's equalizer/gain control in
200 to 500Hz region, and the clarity place and using 18dB/octave crossovers.

amplifier and the 8kQ taps to your
stators. Test the frequency response by
connecting a high impedance voltmeter Refresh in sonic purity with these innovative drivers and system
(at least 10MQ) to the 8kQ leads and designs. If musical accuracy, high signal resolving ability, with
drive your amplifier with an audio phase coherency is what you seek in your dream system, then look
generator so it puts out between 1 and into the JORDAN 50 mm Module and Ted Jordan system designs.
10 volts. Be careful not to get the
voltage too high or you can arc your Now you can be ready for the
meter's guts. Run a frequency response new generation of SOTA low
sweep. You may see some irregularities distortion, wide dynamic range
in the area from 10kHz to 20kHz, but recordings that are coming,
so long as this does not result in severe some forerun-
rolloff of the highs. You can consider it
ners of which
are already in our collections.
superb performance.
The transformer's power rating
should be about 15 watts. This will Satellite sys- terns can use
take all the amplifier power you want enclosures as small as 96 cu.
to feed it and is small enough to have in. net volume, with woofer
good frequency response with commode using twin bextrene drivers per channel in push-pull,
minimum leakage inductance. Larger cancelling second harmonic distortion. Crossover network parts
transformers may work, and lower kits for the 150 Hz point with your choice of iron core or Super Air
turns ratios will work, but achieving Core (see picture) inductors are available.
good frequency response and high
SPL's becomes progressively more dif¬ Send $1.00 for some interesting and surprising literature, which will
ficult. If you are using tubed be in the mail to you promptly.
T » Transcendental Audio
amplifiers, try connecting the 16Í2
amplifier taps to the 40 transformer
taps; this will give you the highest
voltages but sacrifice good power INNOVATIVE COMPONENTS AND DESIGNS IN HIGH-END AUDIO
transfer. Since the speakers do not use 6796 Arbutus St., Arvada, CO 80004 303-420-7356
power, this is no problem.
E. J. Jordan • Polydax • Decca • TA Unbox Enclosure Systems
CAPACITIVE AMP LOADS
Some amplifiers may be unstable with
capacitive loads. I have found pro¬
blems only with a Leach low TIM amp;
Crown, Williamson, Perkins, Dyna,
Phase Linear, and Quad amps have all
worked without difficulty. If you
should have a stability problem or
your amp blows fuses, try adding some
series resistance: 1-10Q seems to work
OK. You may also want to try using
the transformer's 80 or 160 taps;
however, this will decrease your SPL's
LOUDSPEAKERS
to some extent.
When deciding what amplifier to
use, please ignore all the
"techno-freak" ideas you may harbor
about amplifiers. Several readers and I A Speaker
have discovered most amplifiers sound
identical on the ESL's even though the Manufacturer
same amplifiers may sound markedly
different on conventional dynamic
With A Sound oa
speakers. The "tube vs. transistor"

controversy also becomes meaningless:
Reputation
transistor amplifiers do not sound
"harsh" or "edgy" with these speakers.
This puzzles me because generally o O
ESL's seem to place more stringent
demands on an amplifier and commer¬ ¿y
cial ESL's are very amplifier sensitive.
My only explanation is that these J
speakers present "pure" loads to the Z,-.*?
amplifiers rather than the complex


AN ELECTROSTATIC watt unit designed to do the major part You need only one power supply for
SPEAKER SYSTEM: Part 1 of the voltage reduction, you can place several speakers.
Continued from page 21 a x/i watt pot across the input to allow High voltage test prod wire makes
loads presented by the commercial a small amount of trimming; if it burns the best connections from the power
units with their passive crossovers and up, you need more resistance in Rj. supply and transformers to the
woofer systems, often of the dynamic If you can get your hands on a small speakers. Do not tape or otherwise
type. The combination of capacity, in¬ high voltage transformer you can place hold the wires close to each other, as
ductance, and resistance of such a load a lOkV diode in series with it to get this increases stray capacitance; just
under dynamic conditions must be just DC. It doesn't matter what polarity lay them loosely on the floor. I have
awesomely complex in comparison to a you feed the diaphragm if you are us¬ used common 18 gauge "zip cord" for
simple 2400pF capacitor. The ESL also ing transformers. The speakers will connections and have had no arcing
essentially requires no current, so the usually take about 3kV, but they are problems, but I don't recommend it
amplifier is working only with voltage marginally stable at that voltage and since it is only rated for 600 volts. I like
rather than with large amounts of will sometimes arc, or the diaphragms using banana plugs for connectors.
both. This is quite different from a con¬ will "cave in." If this happens simply Keep plenty of space between them as
ventional speaker, and is also the turn off the power supply, go over to the voltages are high enough to cause
reason a 200 watt amplifier doesn't the speaker, and blow the diaphragm arcing over at least a quarter of an
overload a 15 watt matching away from the stator it is stuck to. This inch. Getting shocked by the system is
FIG II
WOOFER
LO PASS POWER AMP
WOOFER
EQ/GAIN
AUDIO
MATCHING
TRANSFORMER
HI PASS ESL AMP STATOR
DIAPHRAGM
STATOR
VARIOUS HIGH VOLTAGE TO OTHER

POWER SUPPLIES CHANNEL
120 VAC RI TRANSFORMER
5-2OK IOKV
CENTER TAP
I0W
R2 05uF 22MEG
IK W
I/2W POT 22 MEG TO OTHER
CHANNEL
DIAPHRAGM
TRANSFORMER 1700V OR HIGHER
0.1mA CURRENT CAPACITY OR HIGHER
Fig. 11. Block diagram of system interconnections as well as a hookup guide for both an output transformer and a high voltage
power supply. Note that each panel requires a 22mQ resistor in series with the supply. Ri may also be a 10 to 20W unit. R 2 should
probably be a wirewound unit as well.
transformer even when the amplifier is may or may not pop a pinhole in the not lethal, but I can assure you you'll
clipping. diaphragm; at any rate it doesn't know when it happens!
destroy the speaker.
POWER SUPPLY To be on the safe side, don't run the DIRECT COUPLED AMPS
Fig. 11 includes the schematic of a voltage at the "ragged edge." I find no
typical high voltage power supply. detectable difference in SPL between 2 You may wish to build a direct coupled
There are several ways to approach and 3kV even though in theory there high voltage amplifier to drive the
this device. Probably the easiest is to should be. You may find you can use a speakers. If so, build the high voltage
buy a surplus electrostatic copier flyback transformer from a television diaphragm supply into the amplifier. I
power supply for around $10. These set for the supply. A small capacitor is used to use such an amplifier but have
are usually rated around 5-10kV and usual at the power supply output, but I since returned to transformers; chiefly
you need only 2kV, so to reduce the think it's unnecessary since the speaker because I finally found good ones. The
voltage you will have to pad the input. is a capacitor. Do not omit the 22MQ DC amp is a class A device which
Generally this is done with a voltage resistor in the line to each diaphragm, generates a lot of heat and wastes elec¬
divider at the input. If Ri is a 10 or 20 as this helps prevent charge migration. tricity. You can achieve high SPL's

with the right transformers and a con¬
ventional amplifier. Build Your Own Speakers!
(SEAS)
Stability is a problem with a DC
amp, whose only advantage is that ob¬ low DCR inductors
jectively it is considerably better than a metal film resistors
transformer. However, the trans¬ electrolytic
formers I now use sound every bit as
detailed as my DC amp; in fact other film capacitors,
ONE OF EUROPE'S LARGEST
users have told me the inner detail SUPPLIERS OF SPEAKERS driver components
seems better with transformers than NOW MANUFACTURES A VENEER SPEAKER CABINETS
with an amp! I feel this is due to subtle QUALITY LINE of epoxy printed circuit boards
differences in frequency response SPEAKER KITS
rather than to any inherent trans¬ AND DRIVERS. FOR THE 10mfd50vNP cap. $1.00
former superiority, but a transformer KIT BUILDER 16mfd50vNP cap. 1.20
coupled system can be just as good as a Imh inductor air c.l.LO
DC amp system. AT A GREAT SAVING,
3.5mh " PIC 3.50
EASY TO ASSEMBLE,
NO SPECIAL TOOLS NEEDED, Uïg"8ohmMid/potCov .10.00
THE COMPLETE SYSTEM
EVERYTHING INCLUDED. 10" " Woofer 50W 21.00
My recommended and proven system iV'HornSuperTweetr U.50
consists of:
1. A pair of ESL's 2'x6' in radiating WE HAVE A TOP QUALITY Partial List/Much More!
area. SERVICE DEPARTMENT. FOR
2. A pair of transmission line woofers. RESTORING YOUR DYNACO FREE CATALOG & PRICE SHEET
STEREO TO FACTORY SPECS. Write to:
3. A pair of stereo power amplifiers,
one for the woofers and one for the PHASE MATRIX GROUP
ESL's. Box 101
4. A pair of equalization/gain stages. ; J. C. ELECTRONICS, Inc. Bethlehem, CT O6?51
5. A pair of electronic crossovers. 2001 SPRINGFIELD AVENUE (203)266-5585
6. A pair of matching transformers. MAPLEWOOD, NJ 07040
7. A high voltage power supply. (201) 763-4970
! MINIMUM ORDER $5.00
I constructed the electrostatic cells as All orders odd SI.00 Postage and Handling.
large cells rather than small ones

because I didn't want to have to build a
bunch of little cells and wire them
together. Two 2'x3' cells are used for
each channel, stacked and mounted in¬
to a narrow wooden frame with both British Components and designs for
sides completely open. The system in
the photographs has wire stators, but the high quality speaker builder!
for general construction I recommend
perforated aluminum. 1'11 discuss both Our range runs from transmission lines to a system similar in
construction methods later in this
series. all essential respects to the LS3/5A. Audiogram have twice
recommended our Webb kit alongside the KEF RI05 and the
FOR THE LOWS Linn Isobarik (issues 12/78 and 12/79 refer).
The woofers are KEF B-139 drivers
mounted in 10' transmission lines. I All our systems are comparable in performance with the best
have tried many other types of woofer similar commercial types and usually better—we do not con¬
systems, including bass reflex, acoustic fuse things by handling radio/T.V., group, disco, PA or junk
suspension, infinite baffle, semi¬ designs or components. Full catalogue, data and export rates
infinite baffle, and horn. In my opinion
$5.00 air mail. Stockists of Crimson Elektrik amplifier
only the transmission line and the horn
systems were able to produce adequate modules.
bass; only they could make the sound
coherent so you could not detect the
woofer's presence in the system. The
entire sound appeared to have the Badger Sound Services Ltd.
quality of the electrostatic when the
woofer system was operating correct¬ 46 Wood Street
ly- Lytham St. Annes, FY8 1QG Lancashire,
An often overlooked point is that the ENGLAND
woofer must be in the same plane as
the electrostatic. Because the crossover
point is above 100Hz, one needs two
Craftsman's Corner
PHASE CONCEPT LTD. PCL— 101A.
distance for phase alignment.

are due to differences in data, a subject I form I received new data from KEF that
SB Mailbox shall return to in a moment. The differences
in the results of equations 2 and 3 are of
considerably greater importance.
were substantially different, and pending
clarification I decided to incorporate the
new data into my Tables 3 and 4.
ST AMLER QUERIED The explanation is that the equations as I have since talked with Mr. Hoge, and
printed in the magazine are pure hash; he explained that the data in his article were
my appreciation to you for Sneaker Builder somewhere in the works a couple of ex¬ the results of tests by Pro. Ashley of the
magazine should be expressed; and I’m hap¬ ponents got translated into multiplying fac¬ University of Colorado, a highly respected
py to do just that now. I have been de¬ tors instead, and although these errors were figure in the world of speaker design, he
vouring the contents of its first issue, and corrected on the galleys, in the rush of also said that the only major discrepancies
enjoying every minute of it. deadline pressure they remained in their between Prof. Ashley's figures and the
However, I am also writing in regards to incorrect form. The correct equations are: manufacturers' figures were for the KEF
Paul J. Stamler's article, which has proven drivers, and that they had corresponded
to be of most interest to me. I have found it (Eq. 2) Ripple (dB) = about it without resolving the differences.
most frustrating to use his calculations and Which leaves us a bit up in the air. The per¬
to come up with substantially different 20 log [ 2.6 Q„ ( )° ” ] formance of my speakers in the lower bass
results than his. region tends to bear out KEF's figures to lit¬
I have no quarrels with equations 4 and tle better than Prof. Ashley's, but it seems
5; indeed I obtained the same final figures to me that another set of tests is in order. I
(Eq. 3) fB = f. ( Vx5 )° ” plan to undertake a limited series of tests,
as he. But equations 1, 2 and 3 just do not
compute using the figures he provides both VB and will report results as they emerge. Stay
in the article and in table 3. I nave enclosed Note that in both of these equations the tuned.
my calculations which show a small devia¬ quantity: Mr. Hoge, incidentally, expressed confi¬
tion for equation 1, and absurd deviations Vas dence in the manufacturers' data from JBL,
for equations 2 and 3. vB CTS and Electro-Voice. For the most part,
It may well be that Mr. Stamler's design the data in my tables for these manufac¬
is based upon W.J.J. Hoge's calculations (also known as a) is raised to a power turers came directly from their literature;
(which I do not have access to, and which (0.35 or 0.32), not multiplied by 0.35 or the Altec, Philips and Polydax data came
are not provided within the text of Mr. 0.32, as given in the magazine. from tests performed for SRC Audio in
Stamler's article). However, if this is the I am very grateful to Mr. Page for cat¬ Texas.
case, clearly, the system that Mr. Stamler ching these goofs, as otherwise someone Paul J. Stamler
designed is not even close to being optimiz¬ might have attempted to apply the equa¬ St. Louis MO 63130
ed to his own figures, (those in table 3). tions, obtained some hair-raisingly bad
I would hope that you might provide speakers, and come after me muttering im¬ STAMLER TIPS
some clarification of this disparity in a precations. In all seriousness, the moral is
future issue. clear: before starting a project based on an congratulations on your first issue of
In any case, continued thanks and best article like mine, run through the author's Speaker Builder.
wishes to and for what should be a fine and own calculations using his or her data, and There are a couple of small problems in
much needed magazine. if you get hash, write a letter. Paul J. Stamler's article, "How to Improve
Gilbert Owen Page Now for those differences in data. Mr. that Small Cheap Speaker" (SB 1/80, p.
Ansonia CT 06401 Page is right; the original design used 18), that I feel need to be corrected. In Fig. 1
numbers supplied by Mr. Hoge in his article Stamler shows the schematic for KEF's
Mr. Stamler replies: in Audio; specifically, his numbers for the DN-13 crossover network. However, he
Mr. Page is correct in every respect. The KEF B110 are: P = 31, Q" = .39, and Vxs failed to point out that the 2mH inductor
small differences in the results of equation 1 = 27. While the article was in the galley must have a DC resistance of 10 ohms. This
BOXED BOOKS BASS

If he wishes to have a flat response, i.e. OdB ripple, he can
I have read Mr. Stamler's article in SB's first issue (p. 18) and it's choose VB = 28.1 liters giving an f3 = 30.4Hz with fB = 30.6.
nice to know that others are as enthusiastic as I am about the 1 would also like to comment on the L-pad formulas. First,
Thiele-Small alignment procedure. there is a subscript typo in the definition of R P. It should be R P
In his article, Mr. Stamler expressed concern about the effi¬ = R 2II8. Second, this circuit is nice to use for other impedances.
ciency of his system because of the low cutoff frequency. His If the tweeter impedance is R r then:
numbers were VB = 36 liters, fB = 28.4Hz, f 3 = 27Hz. I suggest
he decrease the box volume to 27 liters. This is easily ac¬ R, = (l-A)Rr and R 2 = — R r.
complished by storing some old books in the box. About four 7" 1-A
X 91/2" books each 2" thick should do the job. Assuming he used
b = 31Hz, Qts = .39 and VAS = 27 liters, the box should then From A = —R, + R„ = Rr, A = 5iz51- 51 =i-A.
be tuned to f B = 31Hz giving an f 3 = 31Hz. With this alignment
Ri + R b Rr Rr
there will be a 14% improvement in efficiency.
From A = ——— and Ri =(l-A)R r it follows that RB= AR T.
V"" = Kw(31) (27) _ j _ 14% ¡ ncrease Ri + R P
K„(27P(36)
„ 1 1
Hence, 1—
As an additional benefit, the ripple will now be +.12. R: ARr Rr
Ripple(dB) = 20 log
Robert M. Bullock
20 log [2.6(.39)1 = +.12) Oxford OH 45056

is not mentioned in any of KEF's published persuading people to clutter up their rooms
literature, but they do use a high-impe¬ with additional loudspeakers and
dance inductor in their pre-assembled cross¬ associated wiring) made it necessary to
overs. If someone attempts to build this design and promote time-delay systems
crossover from scratch, serious frequency OLD COLONY’S ’
primarily as quad systems. But it became
response deviations will result if the proper
resistance is not used. I cannot stress this
strongly enough. If a lower-resistance in¬
clear early on that with six or eight speakers
the listening area is enlarged, the setting of BARGAIN
ductor is used, then the builder must add
enough resistance in series with the induc¬
tor to equal 10 ohms.
delayed-channel levels is less critical, and
false localization effects seldom occur since
each speaker is operating at a very low
BASEMENT
In Fig. 3 Stamler fails to mention that the
0.3mH inductor must have several times a
DC resistance between 0.55 and 0.6 ohms.
volume level.
Two more products should be included in
your list of time-delay devices capable of
BONANZA
This is mentioned in the KEF literature on
the acoustic Butterworth filter. Fortunately providing high-quality delay without elec¬
most commercially available 0.3mH induc¬ tronic recirculation: the ADS 10-01 (Analog A once-in-a lifetime purchase allows Old
tors will fall within this range, but builders & Digital Systems, Wilmington MA 01887) Colony to offer Audio Amateur's readers
should measure their inductors to be sure. and the Sound Concepts SD550 (Sound some unusual value in electronic parts All
are brand new devices. Quantities are
Again, if the measured resistance of the in¬ Concepts, P.O. Box 135, Brookline MA strictly limited and are offered on a first-
ductor is less than 0.55 ohms, appropriate 02146). Each retails for $700. As for wide¬ come, first-served basis. All prices are
series resistance must be added for the filter band professional delay lines, DeltaLab postpaid in the 50 US states. Outside the
to work properly. makes two models which would be more USA please include 25% over the listed
I hope these suggestions are printed be¬ suitable than the DL-1 listed: the DL-2 price for postage. Excess will be refunded
fore too many readers build crossovers
from the published plans and end up with ($1750), a stereo delay system with BB-1 ELECTROLYTIC FILTER CAPACITOR:
bad-sounding speakers. I wish you the best elaborate signal-processing options, and the 10,000uF @ 75V (85V surge), Sangamo in¬
DL-3 ($750), a minimum-cost highest- sulated. No hardware included. Excellent
of luck with your new publication. for expanding amplifier • power supplies
David Barnett quality mono delay. Their correct address: such as Dyna 400. Each $4.50;
Little Rock AR 72207 DeltaLab Research, 27 Industrial Ave., Four for $1 5.00.
Chelmsford MA 01824.
OF SHROUDS & SAVINGS Peter W. Mitchell BB-2 ELECTROLYTIC FILTER CAPACITOR:
3,300uF @ 60V, (70V surge), Sangamo in¬
West Medford, MA 02155 sulated. No hardware. Each $2.00;
i have just finished enjoying the first issue Four for $7.00.
of Speaker Builder. A magazine of this sort
has been sorely needed for many years. It is SAFFRAN, TLS & DALINE BB-3 ELECTROLYTIC FILTER CAPACITOR:
about time that speaker building came out 5,000uF @ 35V, (45V surge), Sangamo. Low
of the dark ages. Its shroud of mystery, that ESR. Insulated. No hardware. Excellent for
Your first issue is better than I expected, preamplifier power supplies. Each $2.50;
some manufacturers would be all too happy since you have articles of various themes Four for $9.00.
to continue, should be removed. Several which strike a nice balance. It would have
years ago I decided to build my own been onerous reading to have only con¬ BB-4 SLIDE POT: Mono, 10k linear. Travel
speakers, long after I had built my own struction projects and crossover design ar¬ 2”. Five for $4.00; Ten for $6.00.
electronics. It was surprising to find out ticles. After all, the final aim of the
magazine should be twofold: a) speaker BB-5 PAT-5 HIGH LEVEL: One pair of
how many $200 list price speakers have production per se, and b) how do speakers boards, all parts soldered in place including
$30.00 worth of components in them. After fit into the general scheme of music ap¬ socketed FET IC's, mounted on U-brackets.
several speaker projects and experimenta¬ Very limited quantities. Pair $35.00.
preciation in the home. Congratulations,
tion, I am thoroughly convinced that and keep it coming! BB-6 Sold Out
building your own is always rewarding, less Richard Saffran's letter is extremely in¬
costly and can provide a tailored level of teresting to me, since he mentions his Thiele BB-7 PAT-5 High level boards with all parts,
performance not available or affordable alignment of the Webb design... Please original non-FET version. Brackets sup¬
publish the dimensions of his enclosure. I plied. Limited quantities. Pair $1 5.00
through purchase of ready made units.
I am a graduate mechanical engineer and am eager to modify mine in order to get
that 20Hz cutoff point! QUALITY CAPACITORS
have worked in a manufacturing engineer¬ Transmission line speakers are strongly
ing capacity for 12 years. BC-1 TRW, 5uF/200V ±20% Polypropylene
defended by some authors, but strongly at¬ units. 1 % X’A”. Pair $19.00
Robert Galler tacked by others. Will somebody give us (Available after May 1, 19801
Altoona, PA 16601 the real story on this?
I am presently constructing one pair of BC-2 Electrocube W950DIC505K 5uF
the Daline + B-110 design by R. Fris. In¬ ± 10% @ 135VAC Polypropylene 2'/, x 'A".
itial testing with the B-110 alone (no Pair $9.00
AMBIENCE APPLAUSE
tweeters) has given me a bass response that BC-3 Sangamo 500R 5700 @ 30VDC Elec¬
As a member of the design team which is, subjectively, on a par with the Webb trolytic Low ESR. Insulated can.
developed the Audio/Pulse, ADS, and design, on musical material. The Webbs Pair $16.50
come out better when the source is an
DeltaLab digital time-delay systems and as oscillator. Still, this response from such a UNUSED SURPLUS
the author of their instruction manuals, I small system is phenomenal! It is hard to
applaud Clark & Muller's article on am¬ believe, but it's true. The Daline is very BC-4 1500uF/80V Mallory Electrolytic.
bience speakers (Speaker Builder #1 ). It cor¬ strong competition for the Webbs, and it is Each $2.50
rectly focused on two essential aspects of much less expensive. BC-5 2500uF/80VDC, 105V Surge Electro
ambience reproduction which are all too Carlos F. Bauza lytic. Each $3.50
frequently ignored—(1) the desirability of San Juan, P.R. 00936
pointing the ambience speakers upward or Send all orders to
away so that their sound sprays off the OLD COLONY SOUND LAB
walls and ceiling, and (2) the dramatically BOX 243 PETERBOROUGH NH 03458
more subtle and naturalistic ambience (603) 924-6526—9-4 M-F.
which is obtained by using six or eight
speakers instead of the usual quad array.
Commercial realities (i.e. the difficulty of

AN ELECTROSTATIC
Continued from page 29 Classified
woofers for best imaging. I have used ESL or the bass. If your amp is suitable Advertising
very large woofers, including 24" for the ESL, then you can get by rather
Hartleys, but smaller units offer cheaply because a bass amp is not such
distinct advantages. First, the physical a problem—the Williamson amps cost PRIVATE CLASSIFIED ADVERTISING
size and expense of large ones is pro¬ under $100 from Old Colony. On the SPACE up to 50 words in length is open
hibitive. Second, smaller units have other hand, if you have something like to Speaker Builder’s subscribers
better upper bass response because of a medium power receiver, you will without charge for personal, non¬
smaller mass. Third, the T.L. and horn probably want to use its amp for the commercial sales and for seeking in¬
systems will allow the small units to go woofers and buy a new or used "super formation or assistance. The
publishers reserve the right to omit any
awesomely deep. Fourth, I suspect the amp" for the ESL's. A used Crown DC-
ad. Any words beyond 50 are 15 cents
smaller units have less distortion 300a will go for around $500, and per word. Please type or neatly print ad
because of relatively stiffer piston with everything goes up from there. copy on a separate sheet or card with
less mass. You will see this is no small project; your full name and address.
Finally, having used "sub-woofers" I it will cost anywhere from a few hun¬
feel that there is such a thing as too dred to a few thousand dollars depen¬ TRADE CLASSIFIED ADVERTISING
much deep bass. Even on master tapes ding upon your innovativeness, parts RATES: 25 cents per word including
essentially no musical information ex¬ on hand, and general financial condi¬ name, address, and zip code—prepaid
ists below 30Hz, but you sure can hear tion. You can also cut a lot of corners only. 1O°/o discount for four insertions.
to save considerable money, space, Speaker Builder cannot accept respon¬
air conditioning noise, disc rumble,
sibility for the claims of either the
trucks shaking the ground on master and time. For example, you can get by buyer or the seller.
tapes, etc. The KEF T.L. system goes with a single ESL cell per channel. You
well below 30Hz, and I often find can probably use your present speaker
myself using a 30Hz garbage filter to system for woofers. You may be able
clean up the sound. Another advantage to build a DC amp cheaper than you TRADE
to the smaller T.L. system is it is highly can buy a new "super amp." You may
efficient; a horn system is even better. I find you can be satisfied with an ESL
am using a pair of Williamson 20 watt operated full range driven by your pre¬
amplifiers on the woofers and a 100 sent system amplifier. Perhaps you can
watt Perkins amplifier on the ESL's: the be satisfied with lower SPL's and save HOW TO BUILD CROSSOVERS.
ESL amp distorts before the woofer yourself the cost of a "super amp" by Send S.A.S.E. for catalog of
amp does. I'll present a suitable T.L. using Williamson amps throughout the crossover information publica¬
tions. MESALAB, Dept. AS, 3942
system in the construction section of entire system. Mesa Avenue, Sarasota, FL 33583
this series. In short, you have lots of options,
but unless you build the "full house"
HOW MUCH? system you are going to compromise
The system's cost varies and is based some aspect(s) of performance; you TOP QUALITY Speaker Cables with banana
more on the cost of electronics than on will still have electrostatic detail and plugs. 25 ft. only $24.95. SNAKE CABLES,
the cost of the ESL. ESL materials for quality, but you will sacrifice either Dept. SB, P O Box 242, Littlerock, CA 93543.
T4/80
four cells should run under $100 if you SPL's, frequency response, or
use perforated metal. The cost of bass/midrange detail. In any case, you
frames to mount them is too variable will have the satisfaction of having DECCA RIBBON SPEAKER DK1000 $400.00
now $200.00, frt. prepaid, check or money
to predict; mine were about $20. Grille built your own speakers, and they will order, N.Y.S. residents add 7%. We stock
cloth will cost about $12 per yard. sound well. But by spending the time, drivers—KEF, Polydax, SEAS, Philips,
Transformers are $35 each (from me, money, and effort required to build a Peerless, and passive radiators. Metalized
"full house" system, you will find the mylar caps 1-20uf, air core chokes #14-16-18
1578 Austin St., Atwater, CA 95301), wire. SPEAKER SHOP INC., 3419 Bailey
but you may have some lying around sound you can get will exceed anything Ave., Buffalo, NY 14215. (716) 837-1557.
for free, or perhaps you can find some commercially available; and you will T2/80
cheaper from a transformer company. have the satisfaction of having done it
LONG HAIR WOOL carded-cleaned for stuf¬
You can build T.L. woofer enclosures yourself. □ fing speakers. $7.50/lb. including shipping.
for around $50 each plus the cost of J. EBBERT, 770 Holly Rd., Wayne, PA 19087.
drivers which may be as high as $100 (215)687-3609 T2/80

per channel. Electronic crossovers can
be had for about $23 a pair from Old
Colony (P.O. Box 243, Peterborough, STATE OF
NH 03458) (plus power supply and
chassis), or from DeCoursey for about
THE ART
ORDER THE BEST BRITISH
$80 a pair. Eq/gain stage can be built drivers, x—overs, non resonant
for $30 the pair, or perhaps somewhat enclosures with transmission line
less. Part 11 of this series includes complete con¬
For design drawings, tech, specs
Your largest outlay will be for struction details for the ESL's and part III
and prices send $ 2.00 bills to:
IMPEX, 111 N. Fairview, Moun'
amplifiers. Presumably you already will detail the construction of bass systems.
Prospect, III. 60056
have at least one to use on either the -Ed.

-loKLf
Our Speakerlab Thirties are the culmination of our acoustic technology and speaker
building experience — in collaboration with one of the world s top acoustic designers, Mila
Nestorovic. A speaker with better transient response in a smaller enclosure than any system
we have ever of fered. They’re our Stradavarius.
Because our ears receive most information about a musical sound in its very first instant,
a speaker’s ability to react instantly to an impulse (called transient response) is extremely
important. Our Thirties image so well that the speakers themselves actually seem to
disappear. The pluck of a guitar string or the snap of a snare drum appear instantly in the
Thirties field of sound. . .
Because they use the patented Nestorovic Woofer System®, Thirties achieve almost a
full octave of extra low bass. Through a unique phase-control network and incredibly high
flux density 8" and 10" woofers, you get amazingly tight bass transient response. 1 he result is
tremendous bass from a relatively small enclosure. Handling up to 350 watts/channel,
Thirties are simply unlike any speaker system you have previously heard.
Experience the Thirties: handcrafted, technically precise instruments.
And Thirties arejust one of Dept. SB2
ten different speaker designs from
Speakerlab. We also have a wide <■
range of speaker parts available.
Send for our free catalog — it’ll tell ;
you more about us, our kits and
our famous patented Nestorovic
/^eakeflo^
Woofer System®, Wave Aperture
Drivers®, Polylam® Cones and
Subwoofer Drive Systems.
I Dept._,-
Dept. 101, 735 N. Northlake Way
Seattle, Washington 98103

Speaker-Builder-1980-02

Uploaded by

Copyright:

Available Formats

Speaker-Builder-1980-02

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Speaker-Builder-1980-02

Uploaded by

Copyright:

Available Formats

A Three-cabinet suspended design | All about electrostatics

Cable Type il/ft. pF/ft. gH/ft.

Speaker Builder | 2/80 6

cables were virtually indentical at

AUDIO SOURCE "HIGH DEFINITION"

AUDIO SOURCE "HIGH DEFINITION"

7 2/80 I Speaker Builder

Speaker Builder | 2/80 8

9 2/80 I Speaker Builder

100kHz to 5p.Sec. pulse 5p.S pulse 5[iS pulse 5[kS pulse

Speaker Builder | 2/80 10

11 2/80 I Speaker Builder

T he system described evolved over

Speaker Builder | 2/80 12

Fig. 2. Dimensions (d) of driver, box or room must be compared to wavelength of

Assuming a pp is 6mm and f 70Hz a

13 2/80 I Speaker Builder

Speaker Builder | 2/80 14

Speaker Builder | 2/80 16

Fig. 6. On-axis frequency response of a point source mounted in different

avoiding all rigid coupling between the

17 2/80 I Speaker Builder

Speaker Builder | 2/80 18

S ince the publication of my speaker

Speaker Builder | 2/80 20

21 2/80 I Speaker Builder

Speaker Builder | 2/80 22

HORNS • WOOFERS • MIDS • TWEETERS Q

MEDIATE DELIVERY. GOOD SELECTION OF MUSICAL IN¬ address

Speaker Builder | 2/80 24

BUILD THE WORLD’S BEST

25 2/80 I Speaker Builder

Speaker Builder | 2/80 26

speakers. The "tube vs. transistor"

amplifiers rather than the complex

27 2/80 I Speaker Builder

HI PASS ESL AMP STATOR

VARIOUS HIGH VOLTAGE TO OTHER

Speaker Builder | 2/80 28

large cells rather than small ones

PHASE CONCEPT LTD. PCL— 101A.

distance for phase alignment.

33 2/80 I Speaker Builder

BOXED BOOKS BASS

Speaker Builder | 2/80 34

35 2/80 I Speaker Builder

drivers which may be as high as $100 (215)687-3609 T2/80

Speaker Builder | 2/80 36

You might also like