Amplifier stability and speaker cable - a cautionary tale
Production dept. asked me to provide them with a new small power amplifier to be used on one of the test rigs. Only a few watts were needed, just enough to sweep-test every drive unit listening for buzzes by ear. The previous B&K oscillator (now faulty awaiting repair) had a speaker drive output stage, but the newer B&K doesn't have an inbuilt power amp.
As It happens, I'd half assembled into a case a commercial "DIY" power amp module comprising of heatsink, power supply and circuit supplied in one neat resin-potted housing. It's a design that's been available for years and sold to the disco/pa market to drive long speaker lines.
Assembly complete (wiring mains, phono input socket and psu transfomer) and hooking up to an oscillator and speaker and all seemed well. Until I turned the volume up more at which point there was a tell-tall audible distorted character to the pure tone: the character of circuit instability. I clipped the 'scope across the speaker feed to see, as expected, that one half of every sine wave had a burst high-frequency supersonic noise, which whilst inaudible itself, manifest as distortion in the audio band. Reduce the audio input and stability returned.
Disconnecting the speaker so the amp was not actually generating power into a load, and again, stability. The solution was a 'zobel network' comprising a small resistor and capacitor in series soldered directly across the output terminals. Cost about GBP 0.10 (ten pence) and commonly fitted to amplifiers to partially-cancel the inductive/capacitive reactive load of a speaker cable and/or loudspeaker itself. But not this one.
So the message from this is .... we want to assume that audio power-amplifiers are well designed and stable into the sort of loads that typical cables + speakers present, such as this combination. But that may be over optimistic on our part. Unless we have test equipment we cannot look for instability and if we identify it by ear, the damage may have already been done. There may indeed be latent instability which is only triggered on signal or transient peaks, depending upon drive level as in this case. The more exotic the speaker cable and/or the loudspeaker load the more we tempt instability. That's why I design Harbeth speakers to present as benign a load as possible to the amp and advise you to use simple, standard cable that looks like simple standard cable.
If anyone is interested I'll take a photo off the 'scope to show the distorted waveform before the zobel was fitted.
Alan A. Shaw
Harbeth Audio UK