The topology decided upon is a new direction for me, but not new in the world.
Since hearing Bruce Edgar's Tightans on a CF output EL34 amplifier I have known there can't be much wrong with the little used CF output topology. His speakers killed all the others at VSAC stone dead and every other speaker system I have ever heard by a country mile. His humble CF output amp took all frequencies except the SUB bass which was solid state.
So the CF output stage has to be tried. Theoretically it should have good control of (damp out the peaks in) the loudspeaker and output transformer at all frequencies, and it benefits from 100% local feedback.
The 12SJ7 is a cheap version of the 6SJ7, I bought a bunch for next to nothing having enjoyed the 6sj7 in a guitar head.
The advantage of the 6sj7 is known use as an audio VA and my own experience confirms that. With the pentode input valve we have the two main benefits we require from our first valve, plenty of voltage gain and low input capacitance. Any source should drive this input valve. We haven't got a lot of current from it. I generally agree that you have to slug the output valve. It must depend on the output valve, the ones I use don't mind my low current technique. So I shall hope that in this case also a couple of mA will do and I won't have to use high transconductance pentodes others are working with these days.
But I have no coupling cap with blocking distortion to contend with and I have no grid leak resistor to limit my voltage gain. The direct coupled CF is the optimal load for any driver, so my s6j7 for one offers the optimal load for any source and I have given it the easiest job in the world to do.
All I have to do is float my two valves to the right DC level. Done it hundreds of times, can't remember the year I last built a cap coupled stage of any sort, at least 5 years, wouldn't know where to begin.
From another thread some facts:
"OK have decided to make a mono amplifier with the one to see how it sounds.
Because of a number of factors the main one being the way my mind is wired, I won't use the text book operating point of 200 Anode Volts 70mA Grid Bias of -25v on a 2kohm Load for 2.3Watts Power Output at 5% thd.
I bought from Mike some cheap 5k OPT's a while back, so have drawn that option. Maximum anode voltage allowed is 250v. On my own trace I can see a useful spot to work at of +250v -35.86v 51.6mA and that is what I shall set this up as. But because my curves only go up to 300v I cannot discover distortion. So I have sketched the graphical data for the stock manufacturer's trace of the valve to see what power output and distortion to expect. My own should be on the same lines. I find Power of 2.48 Watts at 4% 2nd Harmonic. So I get more power and less distortion.
I have decided to use direct coupling, low distortion 12SJ7 Volt Amp driver and Cathode follower output.
Having the uTracer makes complexed direct coupled designs much easier. Usually one has to use the available data build it from that find out your valves and VR tubes (so that's three variables if your resistor values are spot on) settling points and then calculate how to reign it in, have another go measure again and find out how to reign that it, etc until you have a satisfactory and stable conclusion. That should be short circuited to a quicker finish with the data for both valves and for any VR tubes you may use to implement the voltages. That just leaves variation in resistances satisfactory calibration of uTracer and human error for it to go wrong. Much less than the usual hair tearing out scenario.
Let's find out how good this valve is."
