874 Chapter 25
to match the inadvertent internal characteristics, Fig.
25-56. Jensen transformers are uncannily good in this
respect—no tweaks usually being necessary. There are
external circuit influences that can and will upset the
maximum obtainable common-mode rejection. The
accuracy of the phantom-power resistors is one; any
input pad, regardless of accuracy, is another. Assuming
any reactive (i.e., rising with frequency) common-mode
response has been trimmed out, unequal phantom legs
will enforce a lopsided flat common-mode response
while true floating input pads instantly reduce the
CMRR by nearly the amount of their attenuation. Why?
They do this because they only attenuate the differential
(wanted) signal and not the common-mode one. A
halfway solution is to centrally ground reference the
pad. Given all that, less than perfect common-made
response shouldn’t cause any ill manifestations in a
typical recording environment with fairly short input
leads. A high electromagnetic field of any sort, or an
application with very long leads (or worse yet, a multi-
core), is far more likely to create problems with
untrimmed inputs than with those properly balanced;
vulnerability is greatly increased to all types of
common-mode problems including noise on the
phantom power-supply feed. Indeed, this is a common
compounding of faults on a console that exhibits consis-
tently noisy inputs.
25.10.7.5 Minimum-Gain Considerations
A minor compromise is necessary in the first stage to
prevent its gasping with exhaustion on extremely high
input levels. Ideally, the output of the operational ampli-
fier has to look into an impedance of 600: or greater
(this being the lowest impedance into which it can drive
full-output voltage swing). Maximum gain state isn’t
really a problem. If the first stage is overdriven, then the
second stage will be some 30 dB into clipping; someone
might notice!
At minimum gain though, the first stage in Fig.
25-55 is operating almost as a follower with an output
load of 770:, with the remaining feedback path to
ground. That’s safe and easily within the amplifier’s
driving capability. It would be better though if this small
resistance were still smaller because it is contributing a
little unwanted thermal noise to the otherwise beauti-
fully optimized front end. The calculated degradation is
only minor points of a dB and in practicality is easily
lost in the gray mist that always surrounds the marriage
of calculation with practical noise measurement.
The idea of using a front-end stage that turns into a
follower under operating conditions has proved stable
without any obvious trace of ringing within its band-
width. This is probably because it is only being asked to
look into safe, unreactive loads. Things that will make
any unstable circuit squeal have not affected it. Among
the instruments of torture have been a pulse gener-
ator/storage scope and an RF sweep generator/spectrum
analyzer. The 10 pF compensation capacitor is more an
act of conscience than a practical necessity. No compro-
mise comes from its use here, since at maximum gain
the first amp is working 30 dB below system level (an
implied slew rate of nearly 200 V/μs!). At minimum
gain the incoming signal level is such that it’s most
likely coming from a line source of certainly much
more limited speed than the front end.
Down from the nether world of megahertz, the
microphone amplifier is totally stable for audio, even
with the microphone unplugged and input unterminated;
the input network (of RG and CG) is designed to work
in conjunction with the fairly low-input impedance of
the 5534 (150 k: nominal).25.10.7.6 The LimiterElaboration on the simple two op-amp mic-amp element
consists of arranging an automatic gain control element
in the feedback loop of the second amplifier and
following that with a variable turnover frequency
high-pass filter, Fig. 25-55.
A photoresistor device has its resistive end strapped
across the normal gain-determining feedback resistor.
Its resistance drops in value from very high (megohms)
in inverse relation to the photodiode current to a limit of
around 300: at about 20 mA diode current. This resis-
tance swing in the second amplifier is easily adequate
for use in a peak limiter arrangement. The resistance
change is close to exponential versus diode current,
which could be of use in a gentler compressor, but here
as a limiter the resistance change is quite sudden once
that point is reached.Figure 25-56. Input common-mode “tweak.”
+48 V
Microphone
transformer
AB
100 pF 1000 pFAdjust capacitor value and
connect to A or B to balance
the transformer primary for
input common mode.