Two Types of Clip Detection
Driving an amplifier with an excessive input signal so that its output stage is clipping is not in general dangerous to the amplifier, though it can cause excessive power to be delivered to the loudspeaker, and damage it mechanically or thermally. An exception to this is failure of the VAS transistor (see Chapter 7) which may be turned on excessively as the negative feedback loop attempts to pull the output lower than it can go. If the usual VAS current-limiting is applied to protect it when overload protection is operating, then this also protects against excessive VAS currents during clipping.
For a long time there was a belief that the extra high harmonics generated by heavy clipping put tweeters in particular in danger. This hypothesis appears to have been exploded by Montgomery Ross, who shows that the apparent vulnerability of tweeters is in fact simply due to the increased power output associated with clipping, combined with the fact that the tweeter has a lower power rating than bass and middle speaker units because it receives much less power in normal usage.
A clip detect circuit may simply drive an indicator, usually with some sort of pulse-stretching to make brief clipping clearly visible. It may, however, also provide a signal to a microcontroller so that prolonged episodes of clipping cause shut-down. It is important that both positive and negative clipping are detected, because a single unaccompanied voice, or a solo instrument, can have waveforms with considerable asymmetry in their peak values; up to 8 dB is often quoted.
Clip Detection by Rail-approach Sensing
The voltage variations in an unregulated power amplifier supply, due to mains voltage changes and varying current demands on the supply, mean that it is not possible to accurately detect clipping by comparing the amplifier output with a fixed threshold. A far better method is to compare the output voltage with each supply rail. A thoroughly tested way to do this (I have been using it ever since I thought it up in 1975) is shown in Figure 24.33.
Normally the amplifier output is somewhere in the middle between the supply rails and both Q1 and Q2 are continuously conducting. Q2 collector is therefore pulled up to the positive supply rail; the CLIP signal is clamped at +5.1 V by R7 and Zener D2 so it can be applied to a port of a microcontroller such as a PIC.
If the output approaches the negative rail, then Q1 is no longer kept on via R1, R2, and the base drive to Q2 via R3, R4, R5 is removed. Q2 collector therefore drops to the negative rail, and the CLIP signal is clamped at -0.6 V by forward conduction of Zener D2; this small negative voltage is normally safe to apply to a port of a microcontroller without further clamping to reduce it (e.g., by a Schottky diode) but this is a point to check carefully.
If the output approaches the positive rail, then Q1 remains on but now D1 begins to conduct and pulls the junction of R4 and R5 positive, once more removing the base drive from Q2, which turns off, and its collector voltage drops to the negative rail again.
The resistor values R1 to R5 can be adjusted to match the clipping behaviour of the amplifier output stage. Initially D1 can be removed and the detection of negative clipping checked. D1 is then replaced and the relative values of R4 and R5 adjusted so that positive clipping is detected properly.
Note that this circuit detects the approach of the output to the rails rather than clipping as such, so it must trigger slightly before actual clipping. If it was set to trigger slightly after clipping, it would of course never operate. The circuit compensates for supply-rail variations due to both mains voltage changes and supply current drawn, but it cannot allow for the slightly earlier clipping that occurs with low-impedance loads due to the voltage drops in the output stage emitter resistors and the increased Vbe voltages of the output and driver devices.
Clip Detection by Input-output Comparison
A technique that takes everything into account is ‘error sensing clip detection’ which detects actual clipping as it senses when the output signal is no longer simply a scaled-up version of the input signal. This can conveniently be done by monitoring the error voltage, i.e., the difference between the input signal and the feedback signal that is applied to the input pair of the amplifier. However, this detects several conditions as well as clipping; the operation of over-load protection, slew-limiting, a DC-offset fault, or even RF oscillation. A complicating factor is that an error voltage is always present, due to the finite open-loop gain of the amplifier, and this increases with frequency, due to the dominant-pole compensation that is usually employed.
An early instance of this approach was the ‘Input- Output-Comparator’ introduced by Crown in 1977 and used extensively since.
Excerpt from Audio Power Amplifier Design, 6th Edition by Douglas Self © 2013 Douglas Self.