6CM5 hi-fi amplifier

[gcwills]

Hi all - you may be aware of an amp I built several years ago found at Grant Wills
I continue to receive enquiries about this amp which was built to take advantage of speaker line transformers available here in Australia. I recently developed an alternative power supply based on voltage multipliers as normal hi voltage power transformers are not readily available in Aust. I particularly like the sound of this amp - 6CM5 or EL36 sound great in triode or ultralinear mode. As I recently created the schematics in ExpressSCH, I thought I'd post them here

[SY]

Grant, thanks for posting. One question- why do you RC couple between first and second stages rather than direct couple? This puts another rolloff in the open loop transfer function.

[AKSA]

Good circuit, Grant.... but SY's is a fair question. How much voltage is dropped across the LTP tail resistor?

[gcwills]

Originally Posted by SY

Grant, thanks for posting. One question- why do you RC couple between first and second stages rather than direct couple? This puts another rolloff in the open loop transfer function.

One main reason - the grids of the PI sit at 35V - DC coupling would compromise the biasing of the input stage.

[SY]

Thanks, Grant. I suspect there's some ways around that which would improve the performance.

[mnturner]

Really clever design Grant; I've used line level transformers myself with 6bq5's.
The difficulty in getting tube audio/power transformers in Aus sucks; I'd go as far as to say that unless you can afford to fork out the pretty ridiculous cash to have them custom wound or imported from the states (or to pay for someone else to have them inported), its the most difficult aspect of turning a tube design into an amp here in Australia.
Your use of commonly available parts might make tube audio a little more accessible for others.

[DF96]

Does AC coupling here introduce an extra LF rolloff? Everyone seems to think it does, but I disagree. The PS is essentially a differential amp, which amplifies the voltage difference between its two inputs. There is some common-mode gain too, but that will be much smaller. The differential amp has a resistor between its inputs, and this will form a high-pass filter with a coupling capacitor whichever side the capacitor is on. If there is a capacitor on both sides, as shown in this circuit, then it is simply the series combination of the two capacitors which counts. So still only one LF rolloff!

The PS common mode gain will complicate this but if small enough this should not affect LF stability.

[SY]

Originally Posted by DF96

Does AC coupling here introduce an extra LF rolloff?

Yes it does. There are three LF zeros- the RC from voltage amp to phase splitter, RC from phase splitter to output tubes, and the output transformer. If you're not convinced of that, think of it this way- will DC drift in the first stage tube affect the output stage bias?

[DF96]

If you DC couple from the input amp to phase splitter then you still have 3 LF rolloffs because the capacitor to ground from the 'other' PS grid introduces one. This is the one people don't notice. My point was that this is not necessarily an extra one, but combines with the first one. Some people claim that DC coupling eliminates a rolloff; even Mullard seemed to get this wrong.

DC drift in the first stage only propagates through the common-mode gain of the phase splitter if DC coupled, and not at all if AC coupled. But it will be common-mode at the PS anodes too whereas the signal is differential. The drift will cancel in the OPT so will not affect LF stability.

Going back to AC, what happens at the other grid? There is a low-pass filter, which keeps the grid steady at the same average voltage as the PS input grid. What if the signal is so slow that the grid starts to follow? Then you can see that this LP filter causes the PS to have a high-pass action. Is this an extra rolloff?

[SY]

DC drift in the first stage only propagates through the common-mode gain of the phase splitter if DC coupled

Yes, that seems right. You'd have to put a DC voltage source on that second grid to eliminate the zero. Thanks!