892 Chapter 25
- Hangup 1 is the interaction between frequency groups.
Hanging on two control chains that operate at the same
frequency either adjustably or through overlap can at
best be deceiving or at worst self-defeating. In the
Baxandall (as with most other arrangements), if
maximum gain (say, 15 dB) is attained at a given
frequency by one control, a second similarly tuned
chain, cranked for maximum, will not give the
expected additional 15 dB gain. The overall loop is
already operating close to the maximum gain defined
by the stopper resistors. A notable measured result is
for the maximum boost-and-cut capability of a
sweep-mid bell curve to be restricted at the extent of
its range where it overlaps across the shelving
high-frequency and low-frequency curves.
A rough rule born from hard experience of
squeezing the most EQ from the least electronics is to
not allow overlap incursion beyond the point where
either curve has ±6 dB EQ effect individually. Over-
lapping is best achieved from the comfort of another
EQ stage, although that too invokes other
compromises. - Hangup 2 is noise. The basic Baxandall, using purely
passive frequency-determining components, is a
fairly quiet arrangement. With controls at flat, it is
theoretically only 6 dB noisier than the unity-gain
noise of the amplifier plus additional thermal noise
due to network resistances—all in the 100 dBu
region. The noise character varies with the controls,
as would be expected of an amplifier whose gain is
directly manipulated at the frequencies in ques-
tion—high-frequency boost, more high-frequency
noise, and so on.
As soon as active filtering is involved, more noise is
unavoidably introduced, often highly colored and
consequently much more noticeable. What is worse is
that it’s present all the time irrespective of control posi-
tions. Even with its appropriate control at neutral center,
it is quite usual to hear a midsweep swoosh in the noise
changing with filter frequency. This is, along with the
strange spectral character of the noise emerging from
some filters, notably, the integrator-loop variety, a result
of unoptimized impedances and dubious stability almost
inherent to their design.
25.11.26 Swinging Output Control
The source impedance versus feedback impedance
ratiometric approach of the Baxandall is not the only
way of achieving symmetrical boost-and-cut, as stun-
ning as its simplicity and elegance may be. A method of
enclosing the controls within the feedback leg of a
noninverting amplifier is developed in Fig. 25-69. This
has the advantage of leaving the noninverting input of
the op-amp free, obviating the need for a preceding
low-impedance source or buffer amplifier. Roundabout
to this swing is the necessity of a buffer amplifier or
quite high destination load impedance since the output
is variable in impedance and included within the feed-
back loop of the op-amp. Serious control law modifica-
tion, potential phase margin erosion with consequent
instability, and certain head room loss are among the
penalties for careless termination.
Unity gain in Fig. 25-69A is achieved when the
attenuation in the feedback chain equals the output
attenuation; the feedback attenuator causes the op-amp
to have as much voltage gain as the output attenuator
losses. Replacing the two bottom legs of the attenuators
with a swinging potentiometer, Fig. 25-69B, provides a
boost-and-cut facility; when the pot is swung toward
min, the feedback leg is effectively lengthened to
ground, and the amplifier gain is reduced somewhat.
Meanwhile, the output attenuator is shortened consider-
ably, reducing the output accordingly. At max the
reverse occurs. The feedback leg is shortened,
increasing the loop gain of the op-amp while the output
attenuator is lengthened, losing less of the available
output. A small stopper resistor defines the overall gain
swing about unity, which would otherwise range from
zero to earsplitting, respectively.
Introducing reactances and complex impedances into
the potentiometer ground leg (or legs as in Fig. 25-69C)
results again in boost-and-cut control over the
frequency bands in which the reactances are lowest, i.e.,
high frequency for capacitors, low frequency for induc-
tors (real or fake), and so on. This arrangement, which
is in a few professional systems and in some Japanese
hi-fi, has only one major drawback other than the previ-
ously mentioned output-loading considerations. In order
to achieve reasonable control dB-per-rotation linearity,
the two attenuators (feedback and output) need to be of
about 3 dB loss each with the control at center. This
implies that the obtainable output voltage is 3 dB below
the output swing capability of the op-amp, landing a
head room deficit of that amount in the equalizer
stage—probably where it is most needed.25.11.27 Swinging Input ControlAvoiding the head room headache but utilizing a rather
similar technique, the swinging-inputs gain block of
Fig. 25-70 is very promising. Here, the feedback attenu-
ator remains unchanged, but the output attenuator is