I’d never seen one of these before, and found a lot of useful stuff on a site called vintagehofner.co.uk, which I can heartily recommend for a lot of vintage British gear. Now that was a relatively sensible start, don’t you think? I think this amp could be usefully instructive to us for more than the problems for which it came in.
It has an interesting phase splitter, for a start. Here’s a schematic of a usual phase splitter circuit. There are few amps on the market which don’t use some version of this and the versions are very similar except for the odd difference in component values here and there.
The purpose of a phase splitter is to supply the output valves with two different versions of the same signal. So the signal out of output X in the schematics should be the same as that of output Y; but 180 degrees out of phase. Just to get a basic handle on that, anything that comes out of Y, going positive will be going negative at X by the same amount. A mirror image, if you like to look at it like that. And that, essentially, is opposite in phase.
So we get one input at c3 which goes to the grid. r3 and r4 are called ‘grid leak’ resistors, and serve to reference the grids to ground. r5 and r6 function to bias the valve so that it can operate to cope with the input/output signal levels.
One advantage with this arrangement, is that we get gain through it. In other words, the signals out are some multiple of the signals in. It depends on the resistor values but maybe around a gain of 10.
This might seem like an advantage at first view, but it has a big drawback.
This one looks a lot different. For a start it only uses one half of the valve. A phase splitter nearly always uses an ECC 83, once in a while an ECC82 which is pretty much the same valve except it has a lower vg/ia figure which means a lower gain.
The operation of this circuit is different and uses in part an arrangement called a ‘cathode follower’, but the result is the same. But not quite.
In this circuit r1 and r3 are usually the same value, r2 being in there as a bias element. This circuit gives zero gain. In other words you get the same out as you put in. Actually it’s slightly less. So it must be worse than the other circuit, right?
No, actually, wrong. The key word is ‘accurate’. For a class B (that’s push-pull of some description) the ‘push’ has to be exactly the same as the ’pull’. Or the positive half of the signal has to be the same as the negative half looked at a bit differently. It will otherwise introduce distortion of the sort you really don’t want.
It’s more expensive to build an amp with this ‘gainless ‘ phase-splitter, because you have to find the gain from somewhere else, so another valve is needed, the various components associated, hardware etc, etc.
The problem with the first circuit is that, as it has gain, any differences in the resistor values are multiplied, and that situation will get worse as the resistors etc, age.
This Selmer amp, also had a fault on the trem channel. It didn’t trem. This is well worth another blog, on the subject of ladder/phase shift oscillators, which most amps (certainly British amps) tend to favour. Time for tea.