David McGown wrote:
Yes. Global feedback, around 20 dB or so. Basically using the feedback network from the Dynaco Mark 3 and it seems to work fine. Made sure the divider at the cathode of the input tube was the same, and used a bypass cap around the rest of the cathode resistor needed to bias the stage. I probably should fine tune it, but the amps are so sounding fine as is, been hesitant to tweek them any more.
David
Roscoe Primrose wrote:
David McGown wrote:
with an additional feedback loop and a weird resistor bypassing the coupling cap to each output tube
With that resistor bypassing the coupling cap, are you sure you actually had a negative voltage on the grid?
Roscoe
I tried it with and without, and had the same problem. That had me running in circles for a while.
David
David, as an update on this, I just built the same thing, only using the Acro TO-330 outputs and EL34s. I will say that that bias circuit is tricky. I had to fiddle with it quite a bit to get the output tubes stabilized at the right operating current, about 65mA each. And if it's any comfort, this "stable" Williamson is anything but! Of course, the Acros are notorious for HF instability, whereas the Dynacos are probably a lot calmer. Even so, at first measurement I couldn't attach a probe to the input without the thing going haywire. I modified it with my all-purpose, rather aggressive HF rolloff filter at the input (1200pF + 820) and it was stable with .22uF across the 8 ohm load. Here are some pics, first the stock circuit at 10kHz, then the stabilized version, and a frequency plot of the stable version. You can see it's still "singing the mass" out to 100kHz, but it's flat to 20kHz. The red line represents the phase shifts.