[Diabolical Artificer]

Here's what I've built so far - see schematic. There are two mistakes in the drawing, anode of EF37A connected to g3/cathode and g2 connected to 250k R. Oh and a third, the resistor on the cathode of the EF37A shown dotted is going to be a 2k pot for NFB, it should be to the right hand side connected directly to the cathode and the 0.1u g2 bias cap will be connected there also when NFB is applied.

The amp works, kind of, but there's a problem with the second stage. Up to .5v AC sig is a clean sine, but over that, it's clipping, not sure why. Here's some figures. There is 236v HT to the first three stages - EF37A ... a 107v, K1v, g3 40v ( bit low ? ) input 1khz @ .535v AC P-P, output up to 50v AC P-P. 6SN7 ( 2nd stage ) a 108v, k 3.5v so bias a bit too positive, load line suggests - 6-8v bias better, still as it's only taking 500mV P-P before clipping, this suggests something wrong. Have tried more HT - no improvement. This second stage can put out up to 109v P-P into the PS.

The phase splitter is a LTP which started with two 83k anode R's, I increased the out of phase to 120k to balance the two signals. Measurements = a1 173v , g1 44v , both k are 68v, a2 is 164v, g2 49.5v so sections off by about 5v so not too bad.

The six 807's should have 300v HT, but my valve tester PSU ( which I'm using to power them )which should be capable of 250mA will only go up to 269v, I think the cap on the PSU is too small at 220u. 807's biased at -15v, 1v on cathode. As you see in the schematic, the pots are to adjust each valve to get them running as near as possible to each other. However when I adjusted the first four valves, adjusting the fifth - bottom left say threw out the rest. I have to look again at this set up. The valves are all of different makes but all have a similar gm. When tested I ended up with 3 pairs of matched valves - to within 10%, 5% on one pair.

I'm going to see if I have a hefty tfmr for the HT. Rough guesstimate says I'll need 500VA or above. With a 12v g - g on the 807's I got a 15w clean sine out, but stage 2 is messing things up. As soon as I turn the volume up, the signal clips and the valves or something start ringing. I tried cutting the second stage out but the sine out of the OPT was all wrong; picture a line of "n's " joined up.

One other problem I had was ripple, I'm going to sort out the power supply for the first three valves with some filters and bigger caps. Another problem I've yet to sort out concerns applying NFB. From what I've read this is found by applying a 10khz square wave to the input and adjusting the pot on the cathode on the EF37A. This should improve stability, improve the speaker damping etc. I've been unable to find a good text on phase compensation and how to calculate the value of the capacitor that is usually in parallel with the NFB resistor. It may be among the books and texts I've recently downloaded, but have yet to read.

First pic schematic, 2nd, oscillogram of output/input 1khz sine ( top trace, probe x10 = 11v P-P - bottom trace probe x1 535mV 1k. ) 3rd pic temporary chassis and PSU's, 4th 6 x 807's.

TFL, any suggestions most welcome, Andy.