by dcgillespie » Wed Apr 28, 2010 9:37 pm
Hifi --
You're right, the specs are lacking, but then they usually are with transformers. In any event, the Antek transformer has significant differences as compared to the original device. Besides being over 25% greater in primary impedance, most importantly, the screen tap is well off the required % for UL operation of 6BQ5 tubes. This is not insignificant, but unfortunately, it is very common with aftermarket transformer manufacturers, who think that a 40% tap is the end all be all for all tubes. Nothing could be further from the truth. For many it is, but the 6V6/EL84 family of tubes perform very poorly at that percent. No doubt this is a major reason your power output was reduced. Undoubtedly, anyone who has measured an ST-35 clone using Hammond iron has been equally disappointed, as they use the same tap % for all of their transformers as well. I can only assume that if the Dynaclone transformer is an exact copy of the original, then it's tapping point is correct.
The effect of using an OPT with a greater primary impedance is basically the same as underloading the amp by placing a 10 ohm speaker on the 8 ohm tap. While that may be good for speaker damping, it works against LF amplifier damping. That was the first strike against LF stability -- as well as reducing power output. The second strike was using coupling caps that were 120% greater than the original value. The third and final strike is the design of the diytube board itself. The diy board uses separate cathode resistors, and massive bypass caps. While the separate resistors do allow for independent bias adjustment which is good for usable output transformer LF peformance, separating the cathodes also eliminates the ability for the stage to self balance itself to near dc signals. This is important, as it is at those kinds of frequencies, where LF instability resides. Add to that, the RC combination for each cathode with the new board is 22 times that of the original design. The combination of these two things has changed the ultra-low frequency characteristics of the output stage significantly.
Using the Dynaclone transformer surely improved matters a lot. Toroids have never been known for their ultra-low frequency performance. But if you are using the equivalent of the original feedback factor (for 8 ohms the FB resistor would need to be 18,708 ohms), and if the Dynaclone transformer has the same ultra-low frequency characteristics as the original transformers, then the combination of the increased coupling cap values along with the new treatment of the cathodes is still a major invitation to LF instability. Using the original value coupling caps could return respectable stability, but it's doubtful that it would be as stable at low frequencies as the original design was. To mimic the original RC factor in the cathode circuit of the output stage, the cathode bypass caps for the new board would need to be around 22 uF each, so you can see how much greater the 470 uF of the new arrangement represents.
One way to check for ultra-low frequency stability is to pulse the input of the amplifier with a low voltage DC source of about .25 vdc, and watch the cone of your speaker. It should react precisely to the pulse with no signs of settling after waves present. Your scope can help watch for this as well. If they are still present, I would reduce the bypass cap values to the point where a stable response to the pulse is be obtained.
Dave
[/i]