between THD and S/N ratio meant that designers

could choose to reduce the operating current,

improving S/N ratio for a given THD. Either way,
dbx claim the clean dynamic range is extended
from around 110 dB (202) to 116 dB or more.
In 1980, dbx responded to demands by
customers for a Class A VCA, with the 2001. The

INVESTIGATED
primary benefit of Class A operation is lower and
more consistent distortion in comparison to Class
A -B operation. dbx claim a typical midband THD
of around 0.001% for the 2001, contrast 0.03% for
the 202 and 0.01% for the 202C. The
disadvantages cited include lower dynamic range,

PART TWO because the sum of V,, and In in the core increases
in proportion to the square root of current. So in
the Class A -B condition, an xdB increase in signal
Continuing Ben Duncan's series, Part level causes an x/2 dB increase in noise. Bias in
the Class A core needs to be set as large as the
Two describes the approaches adopted highest anticipated signal current. If this is 200
times higher ( +46 dB), noise is then theoretically
by three designers and manufacturers of 22 dB higher under no- signal conditions.
However, loss in dynamic range is mitigated by
audio VCA ICs and modules the necessarily high headroom of Class A
operation, the 2001 returning a DNR that was
The first widely successful audio VCA transistors' temperatures could be internally only 10 and 16 dB short of the 202 and 202C
using transistors was pioneered and unbalanced horizontally (between Q1 +3 and respectively. dbx also point out their Class A
patented between 1970 and 1973 by Q2 +4, the loggers and antiloggers respectively) if VCA has degraded feedthrough isolation (typically
David Blackmer, co- founder of dbx. In the VCA's DC input blocking were leaky (leaving -66 dB) and greater susceptibility to thermal
the basic log-antilog VCA (Fig 10, Part One, it open to DC offset) while being set to attenuate gradients (typically caused by adjacent hot
Studio Sound, June 1989), operation is limited to high level programme. Highly asymmetric, high components). The extent to which these
one polarity of input signal. In an earlier patent level programme (eg percussion) could arguably specifications are inevitable consequences of Class
for an audio VCA by Embley (1970), single -ended have similar consequences: harmonic distortion. A operation per se should become clearer later.
(unipolar) operation was made workable by According to dbx, thermal inertia residing in the The next stage was monolithic construction.
suitable biasing. Eight years later, competing silicon substrate and the metal packaged Packaged as an IC, temperature gradients across
VCA pioneer Paul Buff, patented a scheme using transistors used in the core of the 202, means the core could be reduced by closing up the
full wave rectification at the input to facilitate distortion would only rise appreciably at the spacing between the core devices by two or more
the required four quadrant operation from a lowest audible frequencies. And while the orders of magnitude and by layout that takes
2- transistor core. But dbx's model 202 was the distortion would be predominantly odd order, dbx account of thermal isobars within the dice. Models
first VCA to contain a core comprising consider that it's overshadowed in practice, by 2151, 2150A and 2155 arrived in 1981 -in the
symmetrical pairs of npn and pnp log -antilog other error mechanisms. familiar SIL package. The series of numbers
transistors for bi -polar or `balanced' operation Late in 1978, dbx introduced model 202C, a disguises a single, selected IC, 2151 being the
(Fig 1). refinement of the 202. Additional circuitry was number stamped on the cream of the batch, with
used to sense when any of the core transistors 2155 being salable to the makers of budget audio
were approaching currents that would cause log systems; and 2150A being the intermediate grade.
dbx errors, and generated a correction signal to
compensate for the error'. This doubled the
After four years of development in conjunction
with an IC manufacturer, dbx had arrived with a
The 202's biasing is Class A -B, meaning it's number of matched npn +pnp devices required, to process capable of putting decent pnp transistors
defined by the instantaneous signal level. As in eight. Further, the correction had to be adjusted on a monolithic substrate in common with
power amplifiers, the minimum standing current to suit the transistors in individual cores, making matched npns. The monolithic log -antilog core
is set by a VBE multiplier. In the days when the the 202C more expensive and difficult to was finally born. The 2100 series ICs even
conformity of npn IC fabricated transistors left a manufacture. The benefits were a reduction in included log conformance correction circuitry (as
lot to be desired (fabricated pnps not even being distortion of 10 dB or more for a given operating in the 2020) to help the core devices along.
mentionable in polite company!), the use of current and S/N ratio. Alternatively, the interplay When operating in Class A-B, the dbx topology
discrete devices in the dbx 202's core was a
foregone conclusion for close matching. To this
end, dbx developed ovens that could hold over 100
transistors under isothermal conditions, while
simultaneously measuring and providing
identification of those with matching VBE
voltages. On the other hand, isothermal
conditions in the 202 were limited to improving
thermal conductivity between the transistor cases
through a thermally conductive ceramic block,
interfaced with a silicone coating. If they
occurred, `vertical' temperature differentials
between the npn and pnp core transistors (Ql, Q2
in Fig 1) would create an offset voltage (due to
VBE's temperature dependence), leading to a
thump when modulated by gain changes. In other
words, the 202's feedthrough isolation might
suffer if it were carelessly sited near parts
radiating significant heat. At the same time,
Class A -B operation means that feedthrough is
inherently low, as the DC (bias) current in the
cell, hence % changes created by VBE mismatch,
are both orders of magnitude smaller than in log-
antilog VCAs operating in Class A. Fig 1: dbx Class A -B log -antilog VCA (excludes log conformance correction
Then again, with Class A -B operation, the core
circuitry)
58 Studio Sound, July 1989

[Link]
has a number of snags. First, at high signal
levels, each half of the core is biased off with
every polarity reversal. During this transition,
the signal current approaches nil. With the LOGGER ANTI- LOGGER
1 1---
remaining Class A-B bias current being as small D1 D2 10k

as possible, the availability of current is restricted c1

at the very point where transconductance is
diminishing fast. Avoiding outright crossover
INPUT 0- 10k VIC r IVC OUTPUT

distortion at high frequencies while keeping the +

+
bias current low relies on the core's input half
being partnered with a wide -bandwidth, fast
stewing op -amp. At the same time, the logging
t *C1
HF
CAPACITOR
tOR 1OR

transistors' gain has to be degenerated, otherwise l

the poor phase margin and consequent risk of C V, VE d--{ 3k24
CV, -VE
3k24
instability restricts the extent to which op -amp 10OR 1OOR
speed and open -loop gain can be used to 'kiss it I
better'.
SNR in Class A-B antilog VCAs is not quite
what it seems. When signal levels are small, the
standing bias current sets a respectably low noise
floor. But then signal levels directly modulate the
core current, so noise follows the signal. In real Fig 2: Valley Class A log-antilog VCA core. Simplified half- circuit showing pnp log
operating conditions, with programme near zero - and antilog transistors only
level, dbx calculate the noise floor rises by 20 dB,
though 30 dB has been measured' (our own +Vs
measurements of modulation noise will be
included next month). Generally, the noise should PNP
be masked by signal, but noise modulation is LOG
readily measurable (with suitable filtering or DC
stimulation) and is held to be audible under INPUT 0-1

kt
10k
suitable conditions by some sound engineers, as
well as by other VCA makers. One explanation c 68
for the disparity may lie in the sensitivity of the
control ports of dbx's VCAs. If the circuit and
PCB designers are careless, excess noise present 3k24
PNP

D
--( 10 1-
ANTILOG
in the control path is liable to modulate and 33R
amplify the residual noise entering the VCA's 20k

'
tOR 1OR
audio input.
100R TT

Valley FEEDTHRO T

International I 100 R 50R NPN

LOG
TRIM 47k

15k
Valley International's VCAs are an independent ik 6k8
development of dbx's original, symmetrical log - 0-1 J
3k24
antilog VCA. It began in 1980 with the EGC -101 NPN
(Electronic Gain Control), made by Allison ANTILOG
Research, Valley's predecessor. The TA -101 and
TA -104 arrived in 1982. (Until 1987, Valley ARE THERMALLY COUPLED
International were known as Valley People.) VS
TA-101
Unlike other makers' VCAs, Valley's VCA
modules contain just the core transistors needed
to implement their patented design: TA-104 and
101 are respectively no more than supermatched
quad and octal Transistor Arrays (hence TA). It's
not as penny -pinching as it sounds. Space saving,
cost and manufacturing convenience aside, the
- -o
Diode connected transistor (BE pattern)
core transistors are arguably the only element in
log -antilog configuration that unreservedly
benefits from some kind of integrated Fig 3: Valley Class A log-antilog VCA-the full circuit
construction.
Looking at Fig 3, the shaded area outlines the cancellation leaves the 3rd and higher odd-order transistor in the classic transdiode configuration
contents of Valley's prime VCA array, the harmonics produced by Class A-B operation to (Fig 8, Part One, Studio Sound, June 1989)
TA-101, surrounded by the recommended support dominate, together with associated exhibits current gain, which is a nuisance. For a
circuitry. Initial comprehension should come a intermodulation products. Nasty. Because Valley's start, it makes the transdiode circuit, an
little easier by studying the 'half circuit' in Fig 2, VCAs operate in Class A, distortion is almost otherwise excellent log amplifier topology,
which has been simplified by surgically removing exclusively even order, which can be largely notoriously difficult to stabilise. With 30 to 40 dB
the negative half of the core -meaning it can only nulled by balancing the core transistors. of current gain present inside the feedback loop of
pass a unidirectional signal. Significant odd order distortion products don't the associated op -amp, VHF instability is likely
Like dbx, Valley's patented Class A VCA arise until the signal current tries to exceed the even with a tame UGS chip like LF351. Second,
scheme is built around a symmetrical core, bias current, at clip. For the TA -101's the unwanted gain magnifies voltage and current
composed of pairs of pnp and npn transistors. recommended bias current of 2 mA, current noise (Vn and In). This has a direct effect on audio
Designer Paul Buff enumerates the residual base - clipping implies peak input levels beyond S/N ratio, since current fluctuations in the core
emitter resistance of the core transistors as a 50 VRMS ( +36 dBu), so the conditions for are recovered as signal voltage. Third, the gain in
source of non -linearity. The resulting THD is producing substantial odd -order distortion are the core can magnify the bias current.
proportional to signal current. The symmetry of rather hypothetical. Valley's VCA patent specifies the use of diodes
the pnp -npn core acts to cancel the majority of the Previous attempts at operating log -antilog cells in series with the core devices, to reduce the
2nd harmonic. While yielding low THD figures, in Class A had come unstuck. The logging excess gain. As a result, VHF stability can be
59

[Link]
 guaranteed with a simple lead compensation control signal. With the original ECG-101, use as
capacitor (Cl in Fig 2), which can be kept small a fast acting muting element (one of the toughest
in value, thereby preserving the op -amp's tests of control feedthrough rejection) could result WORLDWIDE AGENTS
bandwidth and slewing capabilities. Precise in a discernible `thump'. In the TA series,
AUSTRALIA: A.T.T. AUDIO CONTROLS PTY LTD
electrical and thermal matching between each improved thermal coupling reduces control -
MELBOURNE (03) 379 -1511
transistor and its associated diode is achieved by feedthrough by at least 20 dB. Again, the bottom
doubling the number of core transistors, then line rests with the designer and how carefully the AUSTRIA: AUDIOSALES
connecting the extra ones as diodes (Fig 3a). With control circuitry is configured. MOEDLING (02236) 26123
the diode degeneration in place, the intrinsic scale David Blackmer's patent `Multiplier circuits' is
BELGIUM: T.E.M.
factor of the Valley VCA's control port doubles, to not limited to a specific bias level (hence
120 mV per decade L, or 120/20 =6 mV per dB. operating class), so it covers the schemes ZELLIK (02) 466 -5010

Bias current is only apportioned equally between promoted by Valley. Although Valley, have CANADA: OMNIMEDIA CORP.
the two halves of the core (pnp +npn) when the argued quite reasonably that the TA series are MONTREAL (514) 636 -9971
control signal is 0 V and gain is unity. With not VCAs in themselves (they are just precision -
DENMARK: PER MEISTRUP PRODUCTIONS
increasing attenuation, the bias current is matched transistor arrays), Valley cite Blackmer's
progressively channelled through the logging patent alongside their own and pay royalties to KARLSLUNDE (02)151300
transistors, and for gain, bias current is directed dbx. At the time of writing, Valley are the only FRANCE: REGISCENE
likewise into the anti -logging transistors. In this pro -audio VCA manufacturers licensed by dbx. PARIS (01) 4396-5050
way, the ratio of bias and signal currents in each
GREECE: ALPHA SOUND
transistor remains constant, as does the total bias
current, irrespective of VCA gain.
Vn is kept at bay by setting the bias current
SSM (Solid State ATHENS (01) 363-8317

HOLLAND: TEM.
high and making the core devices' geometry as
big as is feasible. In is minimised by employing
Microtechnology) BREDA (076)812872

HONGKONG/CHINA: TOM LEE MUSIC CO

low impedances at the core transistors' bases and SSM was founded in 1975. More than any of the
other audio VCA manufacturers, SSM are HONK KONG (3) 730 -1098
emitter terminals. Designing for minimum noise
at the outset is important, because with Class A exclusively into OEM monofab, in manufacturing INDONESIA: PT AUVINDO MERIEN PRIMA
operation, the noise voltage at the output is a variety of audio `function' chips for inclusion JAKARTA(01)323029
invariant, meaning that S/N ratio falls in into electronic music and, more recently, pro -
ISRAEL: BARKAI
proportion to increasing attenuation. At the same audio equipment. In recent years, former and
long-serving President Dan Parks realised that RAMAT-GAN (03) 735178
time, the Valley scheme should have none of the
noise modulation effects that characterise log - amalgamation with a major semiconductor maker ITALY: AUDIO LINK
antilog VCAs working in Class A-B. was the only way to go to satisfy the company's PARMA (0521) 598723/4
One other trade -off potentially arising from technological ambitions. In 1988, SSM was
JAPAN: EDGETECH (JAPAN) LTD
Class A operation is excessive control acquired by PMI (Precision Monolithics Inc), a
major manufacturer of innovative, precision ICs. TOKYO (03) 295-4731
feedthrough. It doesn't matter whether the VCA
is a log -antilog or transconductance type. The It's brought the benefit of advanced processes KOREA: YUNG NAK SO RI SA CO
same vertical and horizontal temperature beyond the reach of a small company (in IC SEOUL (01) 267 -9697
gradients that can cause DC offsets and LF manufacture, anything under $100 million is
NORWAY: NORDTEK EQUIPMENT
distortion in Class A -B VCAs result in a shifting small).
DC residual at a Class A VCA's output. If DC Fig 4 displays the contents of SSM's 2014 OSLO (02) 231590

blocking isn't a problem, and assuming the DC (presently their most highly specified VCA chip) PHILIPPINES: STAGE CRAFT INTERNATIONAL INC
offset is small enough not to result in loss of together with the main external parts needed to MANILA (02) 834-0539
headroom, it need only be considered a source of produce a standard VCA. Although substantially
PORTUGAL: CASA CAIIJS ARTIGOS MUSICALS LTD
distortion if it's actually audible. So much integrated, external op -amps (Al, A2) are
depends in turn on the rate of change of the desirable for decent drive capabilities. The OPORTO (02) 0384456
SINGAPORE/MALAYSIA: SWEE LEE
IC CONTENTS
SINGAPORE (03) 336 -2307
[Link]
R BARCELONA (03) 203 -4804

151 - 0100k
c TRIM
SYMMETRY SWEDEN: INTERSONIC
STOCKHOLM (08) 744 5853
10p SWITZERLAND: SINEC
22k
16
SCHÖNENWERD (64) 413747
THAILAND: BANGKOK CINE /GRAPHIC CENTRE
f
BANGKOK (02) 314-3570
l'vs
TRIM
OFFSET
KC.M. BANGKOK (02) 222 -4712
4M7 100k TURKEY: NEFAN
14

*0 +Vs
Vs ISTANBUL (01)1604514

411
r 10p
UK: FIRST AUDIO LTD

BRIGHTON (0273) 693610

22k

RB TR1 TR2 TR3 TR4 j- +

Al VIG
USA.: E.D.C.
NEW YORK (914) 677 -6478
10p DIFFERENTIAL
WEST GERMANY. ADAM HALL
12,t AUDIO
OUTPUTS
USINGEN (6081) 16031
+0 6

O 7
.I.c 22k
AND
CONNECTIONS
A2 VG
DIFFERENTIAL
AUDIO INPUT
k6
CVO 110* 400R

RB VALUE SETS BIAS, HENCE OPERATING CLASS (A orA -B)

Fig 4: Inside the SSM 2014 advanced transconductance VCA

60 Studio Sound, July 1989

[Link]
 The drive units at the heart of every new Turbosound enclosure are engineering principles, with generous use of die -cast aluminium alloy to ensure
built by Precision Devices Ltd. their indestructibility on the road In fact, our frames weigh twice as much as
A company specifically established by our parent company, Edge most other manufacturers'.
Technology Group, to develop drive units with unsurpassed reliability and audio Guaranteeing that even under massive physical shock, the assembly
performance. won't suffer.
DEVELOPING THE CONE AND COIL. ATTENTION TO DETAIL.
Particular attention has been paid to designing each cone's profile and All this attention to detail means that every step in the manufacturing
materials for its intended application. chain is designed to enhance our enclosures' performance.
The voice coils are wound with pure copper for long term strength and Finally, the components come together in a modular assembly process,
reliability, coated with high temperature enamel, onto an epoxy resin - with testing at every step, before being fitted into enclosures that have been
impregnated woven fibreglass former. built with equal precision.
The whole assembly is cured at high temperature to give it a The result is a finished product that produces accurate, natural sound
completely rigid, integrated structure for greater strength and energy simply through the precise matching of its components. Which means that,
transmission abilities. more than ever, there is no need for external electronic compensation.
AN ADVANCED MAGNET ASSEMBLY. So that's a breakdown of what Precision Devices will mean to you.
The magnet assemblies are of the traditional iron and ceramic And what makes Turbosound enclosures so exceptionally reliable.
'sandwich' type. The difference lies in our attention to detail and the quality of
materials.
Top quality low-carbon steel is machined to very fine tolerances to
maximise contact between plates. And as tight a magnetic gap as possible in
turn gives greater magnetic power and increased cooling for the coil. (In our
OTurbosound4
ONCE HEARD NEVER FORGOTTEN
larger drivers, we also bolt the parts together for extra strength).
TURBOSOD NI] LIMITED, STAR ROAD, PARTRIDGE GREEN, WEST SUSSEX RB13 SEZ. TEL (0403) 711447.
Finally, the loudspeakers' frame designs are based on the soundest of TELEX 878723 TURBO G. FAX (0403) 710155. IMC: TURBO -UK

HOW DO PRECISION DEVICES

MAKE OUR SPEAKERS SO RELIABLE?
HERE'S A BREAKDOWN.
Turbosound TMS -4 er closure

Woven fibreglass former Machined front and back plates

TurboMid'" device'

Die -cast aluminium alloy frame

Suspension Precision turned pole piece

Ceramic magnet
Cone and surround assembly

Pure copper wire voice coil

`TIJRBOSOUND PATENT INFORMATION: Australia: 515, 535 Canada: 1,076,033 Japan: X113424/77 UK: 1,592,246 U.S.: RE 32,183 Nest Germany P274!600/2 Other patents pending.
an Gì.EdgeTech comparry
,oxooxHEW ',OAK 70K,

[Link]
 IN VG, OUT L TASCAM
IN VIG OUT, R
European Distributors
Hi -Fi Stereo Center Kam
Muenchner Bundesstrasse 42,
5013 Salzburg,
AUSTRIA.
Tel: (06621 37701
VCA CONNECTIONS VCP CONNECTIONS
Beltronics
Rue de la Celidee Straat 29,
1080 Brussels,
22k 22k BELGIUM.
, OUT L Tel: 1321 2 424 0233
So and Hoyem A /S,
IN IN Bulowsgarden,
Bulowsvej 3,
OUT 1870 Frederiksberg C,
DENMARK.
Tel: (011 22 44 34
VIG, OUT R
Studiovox Ky,
(a) inverting and non -inverting VCAs (b) Voltage controlled panners Atomitie 5C,
SF -00370 Helsinki,
FINLAND.
Fig 5: SSM 2014 circuit permutations Tel: (80) 562 3411
Harman France,
external parts count can be reduced by omitting THD residual that's close to Class A. For class Peripole 243,
33 Av. de Lattre de Tassigny,
Al +2 but then the on -board output amps (shown selection, only one resistor value need be changed. 94127 Fontenay s /Bois Cedex,
disabled in Fig 4 by tying their outputs 14 and In fact, for demanding applications, SSM FRANCE.
16, to -Vs) have limited output current. The recommend that level detection (an application Tel: (01) 4876 1 144
problem is an intrinsic one, concerned with the circuit is supplied with their evaluation PCB) is Elina SA,
management of heat, rather than a process arranged to switch from Class A-B over to A, at 59/59A Tritis Septemvrious St.,
Athens 103,
limitation: the +20 dBu into 600 f2 line driving high signal levels. Class switching promises to GREECE.
capability that is demanded by pro -audio overcome all the associated tradeoffs -with the Tel: 1011 8220 037
designers has to be turned down for their own caveat that attack and release time constants Greenlands Radio Centre,
good! That's because any such output stage in the need careful attention °, depending on the nature PO Box 119,
close confines of a monolithic VCA will almost of the programme. According to Ron Dow, SSM's 3900 Godthab,
GREENLAND.
certainly produce thermal gradients across the founder and chief designer, provided the class Tel: 299 21347
cell, unbalancing it. If such a VCA existed, there change has a time constant above 10 ms, the
would be difficulties in deciding at which point to GBC Italiana spa,
main sonic effect of the changeover is a 16 dB TEAC Division,
trim THD and feedthrough, as the device warmed change in the noise floor, which shouldn't be Viale Matteotti, 66,
up. At best, such a VCA might gain a reputation audible at the high programme levels where the Cinisello Balsamo, Milan,
ITALY.
for only sounding its best after reaching thermal class change occurs. SSM's model 2018 (due later Tel: (02) 618 1801
equilibrium, several hours after the console or this year) is scheduled to yield further process
processor has been switched on. Worse still, a refinements leading to reductions in THD, noise
Hljodriti - Hot Ice,
PO Box 138,
`heat- saving' Class A -B output stage would and feedthrough. Hafnarfirdi,
thermally modulate the cell to produce a THD ICELAND.
(Part Three will appear next month). Tel: (011 53776
rising steeply below 200 Hz.
AEG Nederland NV,
Returning to SSM's 2014, the overall circuit is
more complicated than any of the previous VCAs, Technical Aletta Jacobslaan 7,
1066 BP Amsterdam,
because overall NFB is employed. This gives rise
to several permutations: a choice of inverting and definitions and NETHERLANDS.
Tel: (0201 5105 473
non -inverting VCAs (Fig 5a) and even VCPs
(Fig 5b). abbreviations Audiotron A /S,
Seilduksgt, 25,
PO Box 2068 Grunerlokka,
All SSM's VCAs are based on a marriage Cell: Active heart of a variable 0505 Oslo 6,
between the current -ratioing transconductance cell transconductance VCA NORWAY.
Core: Active heart of a log -antilog VCA Tel: 1021 352 096
and log-antilog core. In his patents of 1981 and
1984, co- designer Douglas Frey has developed a DNR: Dynamic range Goncalves,
topology that is suited to low cost monolithic HFE: Current gain in a bipolar transistor Avenida 5 de Outubro, 53, 1,
1c: Collector current (in a bipolar transistor) Lisboa 1,
production, meaning it has to beget good yields. PORTUGAL.
To begin with, the cell can be implemented with In: Current noise Tel: (011 544029
just npn or pnp devices-or both. Restricting the Isothermal: Areas of equal temperature
Audio Profesional SA,
cell to npn devices alone (as in 2014) saves the Monofab: Monolithic fabrication, a true (not Paseo Maragall 120,
hybrid) IC Entlo 3a,
cost and design headaches involved in developing
NFB: Negative feedback 08027 Barcelona,
a usable pnp process that doesn't infringe existing SPAIN.
S/N ratio: Signal to noise ratio
patents. Second, when operating in Class A -B Tel: 1931 349 7008
UGS: Unity gain stable
mode, matching between the cell's four devices is VBE: Base-emitter voltage of a bipolar Erato Audio Video AB,
less critical than usual, for good results. In transistor Aeogatan 115,
particular, Frey cites improvements in control 116 24 Stockholm,
VCP: Voltage controlled panner SWEDEN.
feedthrough rejection over a previous Vn: Voltage noise Tel: (08) 743 0750
transconductance type of VCA, where rejection ZOL: Zero operating level, eg +4 dBu Telion AG,
depended on `extreme matching' -something that IC, H1, Ilk: The lower case letters indicate Albisriederstrasse 232,
is difficult for even the best IC makers to incremental (or `small signal') changes in 8047 Zurich,
SWITZERLAND.
accomplish with high yields and hence low cost. these quantities Tel: (011 493 1515
Another perennial problem area, the need for References
isothermal conditions, is readily accomplished in 1 G Bergstrom, Signal Correction Circuit for Teac Deutschland GmbH
Electrical Control Systems, US patent no Bahnstrasse 12,
the 2014, being a full -scale monofab. 6200 Wiesbaden -Erbenheim,
4,234,804 (1980)
The unique features of SSM's patented VCA 2 P Buff, `Specsmanship and the new generation WEST GERMANY.
technique extend to programmable operation: of VCA', RE/P, pp.138 -49, Oct 1982 Tel: (061211 71580
3 B Duncan, `Which Amplifier Technology ?',
biasing can be Class A, A -B, or even Sliding -Bias
Class A". And in Class A -B mode, distortion
Studio Sound, Dec 1988 TASCAM
4 R Bransbury, `Automation and the VCA',
cancellation circuitry comes into play, yielding a Studio Sound, Aug 1985

62 Studio Sound, July 1989

[Link]