Consoles 837audio engineers alike. Slew rate is the speed (measured
usually in volts per microsecond, V/μs) at which an
amplifier output shifts when a step source of extremely
high speed is applied to the input. All the early-genera-
tion op-amps had slew rates on the order of 0.5 V/μs,
but no one really understood it or its implications or
effects then, and it was not the issue it is now.
Why is slew rate a problem? If the audio signal that
the device is attempting to pass has a rise-time that
exceeds the amplifier’s slew rate, then obviously distor-
tion is created—the amplifier simply cannot react quickly
enough to follow the audio. Slew is an issue at high
frequencies (rapid transitions) and high levels (a quiet
signal is moving at fewer V/μs than the same signal
louder). This by happy accident means that a lot of
program material that does not contain high amplitudes at
high frequencies can be passed by slow, low-slew-rate
devices with impunity. Unfortunately, a lot of sources
found in recording studios or on stages don’t fit that bill;
they’re loud and have serious high-frequency content.
The speed limitation was nearly always in the differ-
ential and dc level-shifting stages of the devices. It is
quite difficult to fabricate on an IC wafer ideal classes
of transistors in configurations necessary to improve
matters without compromising other device characteris-
tics (such as input bias current, which affects both input
impedance and offset performance).
Feed forward, in which a proportion of the unslewed
input signal is fed around the relatively slow-responding
lateral pnp stages, improving slew rate and bandwidth
appreciably, is used to great effect in the LM318; a slew
rate of some 70 V/μs is achievable by this technique. It
was in this area of slew rate, combined with a signifi-
cantly improved noise performance (again another
parameter suffering from difficulty in fabricating appro-
priate devices in a relatively dirty wafer), that the next
major breakthrough occurred in devices commonly used
for audio applications—the Harris 911. Although
dramatically improved, the slew rate was still not fast
and was also asymmetrical (+5 and 2 V/μs).25.7.2 Bipolar Field Effect Transistors (BiFETs)
A breed of op-amps called BiFETs, or bipolar field
effect transistors, emerged. These devices have a
closely matched and trimmed field effect transistor
input differential pair (hence, the typically unimagin-
ably high 10 M: input impedance) and a reasonably
fast 13 V/μs structure. These devices are typified by the
TLO series from Texas Instruments, Inc. and devices
such as the LF356 family from National Semiconductor
Corp. Selected versions can, when source impedance is
optimized, give noise figures better than 4 dB at audio
frequencies, which is thoroughly remarkable for units
costing very little more than a 741.
The speed of the devices has been achieved by the
replacement of the conventional bipolar transistor differ-
ential input and level-shifting circuitry with FET config-
urations. Incidentally, the intrinsic noise characteristic of
these FET front ends is significantly different from that
of bipolars and seems perceptually less objectionable.
There are currently a few devices designed specifi-
cally and optimized totally for inclusion in high-quality
audio equipment. With a quoted noise figure of better
than 1 dB at audio, a slew rate of 13 V/μs, and the
ability to drive a 600: termination at up to +20 dBm,
the Signetics Corp. NE5534 (or TDA1034) was in the
vanguard of these; many nice devices have since
followed.
These are all somewhat more expensive than the
BiFET types, but none is prohibitively so, unless the
target design is extremely cost sensitive. Today’s
designer is spoiled by the ability to choose appropriate
devices for each application almost regardless of cost,
and as will be shown, sometimes the less grand and
glorious parts are sometimes the better choice.
Noise in any competently designed and operated
console can be attributed mostly to two sources:- Mixing amplifiers with an appreciable number of
sources and, hence, a lot of makeup gain - The input stage, especially a microphone amplifier
with a fair amount of gain in it
Once a background noise level is established from
the front-end stage (at a level obviously dependent on
the amount of gain employed there), the difference in
noise contribution further down the line between an
amplifier with a typical unity gain noise of 120 dBu
and one of 115 dBu is for the vast majority of consid-
erations totally insignificant. Concentration on these
two hot spots will define the noise performance of an
entire console.
In circumstances where extremely low system noise
floors are actually necessary (rather than just deemed a
good idea) and where such a noise level isn’t being
totally swamped by the source (which it usually is), then
devices like the 5534 make sense elsewhere. Not so
much that they are that much quieter within themselves
but that their substantial output driving capability
allows circuit impedances to be reduced, resulting in a
worthwhile difference to noise floor. It’s nice to know
that there is also maybe a chance of chucking enough
current at capacitors in filters for them to work properly
at high frequencies and high levels. Using the 5534 as a