Loop feedback complicates things, and sometimes in puzzling ways. Many vintage amps have marginal very low frequency stability caused by the stacking of multiple phase shifts. The output transformer's primary inductance appears as a parasitic shunt across the output valves, and its value varies with signal level; usually the coupling capacitors are chosen without proper staggering; the whole thing is often approached without a single dominant pole, and largely ad hoc; etc. etc. So there's no silver bullet answer, but most vintage amps with better quality output transformers benefit from making the cathode bypass caps an order of magnitude larger, and setting a dominant pole with the grid (of output valves) coupling caps. Even better, if possible, is conversion to "fixed" adjustable bias.
All good fortune,
Chris
All good fortune,
Chris
Questions:
Push Pull?
Cathodes tied together, Or Cathodes with individual self bias resistors?
Global Negative Feedback or Not?
Feedback ratio.
Single Ended
Global Negative Feedback or Not?
Feedback ratio.
A bypass cap may need to bypass both the self bias resistor, And the impedance of the cathode (often 1/Gm + "Rp"/u). The effective cathode impedance and the self bias are in parallel (calculate the bypass capacitance needed based on that).
I will for this discussion re-label "Rp" to mean the complete plate load.
And as mentioned earlier, "Rp" at low frequencies is the primary inductance as well as the reflected secondary load.
At very high frequencies, you have the effects of "Rp" including the primary distributed capacitance, and the leakage inductance, and the reflected secondary load.
With real transformers it can get quite complex (pun intended or not, your preference).
Push Pull?
Cathodes tied together, Or Cathodes with individual self bias resistors?
Global Negative Feedback or Not?
Feedback ratio.
Single Ended
Global Negative Feedback or Not?
Feedback ratio.
A bypass cap may need to bypass both the self bias resistor, And the impedance of the cathode (often 1/Gm + "Rp"/u). The effective cathode impedance and the self bias are in parallel (calculate the bypass capacitance needed based on that).
I will for this discussion re-label "Rp" to mean the complete plate load.
And as mentioned earlier, "Rp" at low frequencies is the primary inductance as well as the reflected secondary load.
At very high frequencies, you have the effects of "Rp" including the primary distributed capacitance, and the leakage inductance, and the reflected secondary load.
With real transformers it can get quite complex (pun intended or not, your preference).
Why stop there? A 50 Farad (not micros!) cap is only 5 bucks.
When I threw one in a simulation recently, it worked fine once started. But when I simmed a cold-start it promised to take DAYS to come up to bias point.
And that was with silly-state parts and much more available current than the average tube.
Fender or Sunn used 500uFd once. But that was a shared bias network across a 4-input mixer. In that use, a too-small cap allows bass-bleed from a turned-down input through the other sections.
Oh... that 50 Farad cap is rated 73uA leakage (after you get a charge on it, then another 72 hours), which is just enough to mildly upset a small triode,s bias.
Sure, try it. I have replaced 200uFd with 3*6800uFd, and thought it an improvement, though not the final answer.
sony6060,
1. Depending on the rest of the amp circuit, and if you have room, you might use individual self bias resistors, and associated bypass caps (need more capacitance for each cathode that way).
2. Or you might use individual adjustable fixed bias.
Failing to do 1. or 2. above, then make sure that all the tubes are Very well matched, and by matching, I mean with the Common Bias voltage, and Very well matched at the common plate voltage and common screen voltage of your amp.
1. Depending on the rest of the amp circuit, and if you have room, you might use individual self bias resistors, and associated bypass caps (need more capacitance for each cathode that way).
2. Or you might use individual adjustable fixed bias.
Failing to do 1. or 2. above, then make sure that all the tubes are Very well matched, and by matching, I mean with the Common Bias voltage, and Very well matched at the common plate voltage and common screen voltage of your amp.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.