Consoles 871resistors—manifests itself also as sometimes quite
substantial dc offsets that will likely have to be trimmed
to not eat up too much head room.
Curiously, instability tends to show itself as common
mode. This fault manifested itself to the author for the
first time (too) early one installation morning; a peak
program meter (PPM) across such an output read
nothing, listening elicited a little bit of hum, but a scope
on either leg to ground showed 10 Vp-p square waves,
driving the tape machine to which the output was
connected into shock.
Integrated versions of the cross-coupled circuit such
as the SSM 2142 have the great advantage of extremely
closely matched/trimmed resistor values, and hence far
more predictable performance than discrete versions.
25.10.7 A Practical Microphone-Amplifier Design
Optimizing front-end sound is nothing more than
shrewd judgment in juggling the nearly endless elec-
tronic operating conditions so that adequate perfor-
mance is obtained over the wide range of expected and
common input signals. Any wrinkles should be
arranged to exert influence only under quite extraordi-
nary operational conditions.
The microphone amplifier example described here,
Fig. 25-53, is a somewhat developed version of a basic
front-end design and is in grave danger of becoming an
industry standard.
Initially most striking is the manner in which a
single-track potentiometer is used to vary simultane-
ously the gains of two amplifying elements—the
front-end (noninverting) stage and the succeeding
inverting amplifier. Since the first stage is (as far as its
inputs are concerned) a conventional noninverting
amplifier, transformer input coupling is no more prob-
lematic than with simpler microphone amplifiers (e.g.,
Fig. 25-41, a standard generic microphone amplifier).
With maximum gain distributed between two stages,
large gain is possible without any danger of running out
of adequate steam at high frequencies for feedback
purposes in either of the two amplifiers. This, inciden-
tally, also makes for reasonably simple stabilization of
the amplifiers, something not easily accomplished with
simpler single-amplifier circuits achieving the same
gain swing. Other than the obvious simplicity and
economy of one-pot gain control, two nice features
inherent in the design are interesting from the points of
view of system-level architecture and operation.25.10.7.1 System-Level ArchitectureSystem-level architecture is largely concerned with
operating all the elements of a system at the optimum
levels and/or gain for noise and head room (i.e., at a
comfortable place somewhere between the noise floor
and clipping ceiling). Where gain is involved, it’s impor-
tant that the resultant noise be due primarily to the gain
stage that has been optimized for noise (or rather lack of
it) such that it can then mask all subsequent and hope-
fully minor contributions. At no point in the gain
swing—particularly at minimum gain—should it be
necessary to attenuate unwanted residual gain. This
amount of attenuation gets directly subtracted from
overall system head room. What good is 24 dB of head
room everywhere else, if you have only 16 dB in the
front end?
In this respect circuits similar to Fig. 25-53 score
well, and the graphs of Fig. 25-54 show why. Fig.
25-54A represents the gain in dB of a simple nonin-
verting amp varying with the percentage rotation of an
appropriately valued linear pot in its feedback leg. This
is like the gain/rotation characteristic of the first amp of
Fig. 25-53. Similarly, Fig. 25-54B is the gain/rotation
plot for a linear pot as the series element in an inverting
amp, such as the second gain stage of Fig. 25-53. For
the first half of the rotation, the first stage provides all
the gain swing and most of the gain; only about 6 dB is
attributable to the inverting stage at midpoint. Toward
the end of the rotation, this position reverses with the
front end remaining comparatively static in gain; the
extra swing and gain come from the inverting stage.
Noise criteria are met, since the first (optimized) stage
always has more than enough gain to allow its noise to
swamp the second stage, with the exception of
minimum gain setting. There it hardly matters anyway
because the front-end noise contribution is going to be
at a similar level to the overall system noise floor (i.e.,Figure 25-53. Shared-gain two operational-amplifier input
stage.
Approx
31 dB
max
0 dB minApprox
31 dB
max
Vin 0 dB minVo
R 1R 3R 4R 2Min MaxR 1 = R 2 22 k 7
R 3 = R 4 20 k 7