864 Chapter 25
input-impedance amp terminated only by the collection
of vile resonances and phase-shifting elements that are
an open-circuit transformer. An open-circuit imped-
ance-defining resistor (Ro in Fig. 25-42) with a value 10
or 20 times that of the amp OSI, helps tame this. It also
marginally tames the secondary resonance.
There are a variety of techniques for dealing with
this resonance. They vary from pretending it doesn’t
exist to actually using it as part of a front-end, low-pass
filter to keep ultrasonic garbage out of the electronics.
Minimization of the high-frequency bump is attempted
as much as possible passively, prior to the amp; the
taming network in Fig. 25-42 represents a typical
approach. Here, a series resistor-capacitor combination
in conjunction with the open-circuit impedance-defining
resistor is used. The values are calculated to produce a
step-type response, Fig. 25-43, which when combined
with the hump at the high-frequency end of the trans-
former response, produces a more acceptable roll-off
characteristic. Naturally, the interreaction between this
network and the complex impedance of the transformer
is not quite that simple. The network capacitance reacts
heavily with the transformer inductance, shifting the
resonance frequency in the process. It is this fact that
has led to the misconception that the capacitance
somehow magically tunes out the resonance.
Open-circuit stability is dramatically improved, Fig.
25-43. The network takes an even larger slice out of the
overall high-frequency response, keeping impedances at
the top end comfortably low.
25.10.6.4 Bandwidth
Providing the compensating high-frequency roll-off
around a subsequent amplifier, in the form of exagger-
ated feedback phase-leading around the mic-amp itself
in this case (CF), has the advantage that the noise
performance of the combination at higher frequencies
remains unimpaired by an impedance mismatch
resulting from a passive network.
Problems result in several areas. Compensation
around the mic-amp becomes limited when the elec-
tronic gain approaches unity, while compensation
around a late fixed-gain stage means that all stages prior
to it, including the mic-amp, have head room stolen at
the frequency of the resonance and to a degree of the
magnitude of the resonance. This may or may not be a
problem depending on how far the lower side of the
resonant curve invades the audio band.
The passive method reduces the magnitude of the
resonance. The ultimate low-pass roll-off slope is that of
the high-frequency side of the resonance, which more
accurately is a lightly damped inductance-capacitance,
low-pass, 12 dB/octave filter. The active method uses an
additional 6 dB/octave curve in the compensation
making a total of 18 dB/octave, but it relies on the reso-
nance being of a manageable degree to begin with. A
measure of both techniques is usually required; their
balance and relationship are an experimental process to
optimize for each different type of transformer.
This enforced filtering is of considerable advantage,
helping to keep all sorts of unwanted ultrasonic noise
from finding its way into the mixer. It also represents a
major advantage of transformer inputs over solid-state
varieties.
A further advantageous filtering is the falling source
impedance seen by the amplifier at extreme low
frequencies. This is due to the fact that the winding
inductive reactance reduces with frequency. This is a
definite help in combating the generation of excess
low-frequency noise in the first amplifier.25.10.6.5 Common-Mode TransferThere are two different amplitude response curves to be
considered. The first, the normal differential input, has
been fairly thoroughly determined. The second, by
virtue of its mechanism, relies on imperfections within
the main filter element itself (the transformer) rides over
and oblivious to our carefully calculated filterFigure 25-42. Basic microphone preamplifier with compensation components.MicrophoneRoRGCG
Gain553410 k 7
CF33 MF 100 718 k 7