Handbook for Sound Engineers

840 Chapter 25

ultimate state, though, the negative feedback on which
the amplifier depends for predictable performance is

now completely upside down. Now it’s positive feed-
back. Now the amplifier oscillates.

25.7.7 Transient Intermodulation Distortion (TID)

The TID effect, if not fallout from and overwhelmed by

the effects of insufficient slew rate, is due to amplifier
transit times. Not surprisingly, as is nearly always the
case with fad problems (as was TID during the 1970s),
TID has been known and appreciated for as long as
there have been negative feedback amplifier

circuits—the twenties. It is and always has been totally
predictable.

TID is a direct result of the servo nature of an ampli-

fier with a large amount of negative feedback. The feed-

back is intended to provide a correction signal derived

as a difference between the amplifier output and the

applied input signal. It is a simple concept: any differ-

ence between what goes in and what comes out is error

in the amplifier. All we need do is subtract the error.

However, it is not so simple. Since there exists a time

delay in the amplifier, the circuit has to wait for that

amount of time before its correction signal arrives. The

output during this time is uncontrolled and just flies off

wildly in the general direction the input tells it to. Once

the correction arrives, the amplifier has to wait again to

find out how accurate that correction was and so forth,

see-sawing on and on until the amplifier output settles.

Fortunately, this all takes place rapidly (depending on

the amplifier external circuitry), but it still represents a

discrepancy between input and output. It is an effect

peculiar to amplifiers with large amounts of negative

feedback (typical of most contemporary circuitry),

frequently displaying itself quite audibly—especially in

power amplifiers where transit time is quite long with

the usual huge, slow output devices.

Amplifiers that rely on their own basic

linearity—such as tube amplifiers—rather than on a

servo-type nonlinearity correction system, are often

held to be subjectively smoother. A whole subindustry

thriving on the virtues of feedbackless circuitry has

evolved. Nowadays, though, with device speeds

improved as they are, settling times are becoming insig-

nificant in relation to the signal transients with which

they are expected to cope, pushing the frequency area at

which TID could manifest itself far, far beyond

expected audio excitation.

25.7.8 Output Impedance

A lot of devices, particularly the TLO series of BiFETs,

have a quite significant open-loop output impedance.

This is because the IC designers obviously considered

that instead of an active output current-limiting circuit

(standard on most op-amps up until then), a simple

resistor would suffice. Although this built-in output

impedance—by virtue of the enormous amount of nega-

tive feedback used—is normally reduced to virtually

zero at the output terminal, it is still present and

included as part of the feedback path, Fig. 25-20. Any

reactive load at the output is going to materially affect

the feedback phase and phase margin.

Any capacitance from the output to ground will form

a feedback phase-lagging network. This shifts the phase

Figure 25-19. Transit time effects with increasing signal
frequency.

D. Effect in feedback circuit.

A. Relatively low frequency,

little effect.

B. Transit time very

significant, noticeable

phase shift.

C. Worst case! Transit time

half the period of the

signal frequency,

complete phase reversal.

Fixed transit time

Input signal Output signal

Transit

time

Feedback delayed

by transit time