In my investigation of current source PA possibilities I find it is pretty easy to come up with a scheme that does not include the OPT in the loop (my preference) for SE but not so much for class AB PP (if any one has ideas I am all eyes. 
).
So if including the transformer (not necessarily really high quality stuff) how much GNF do you usually shoot for? 15dB seems to be kind af a natural set point in what I am seeing.
Mike
).So if including the transformer (not necessarily really high quality stuff) how much GNF do you usually shoot for? 15dB seems to be kind af a natural set point in what I am seeing.
Mike
Last edited:
The phase shift in an output transformer limits the amount of feedback possible around it.
Usually a dominant pole is placed in the input circuit to force a -6dB/octave open loop rolloff,
so there is less phase shift at the unity gain point, and improved stability.
If the amp is for PA use 6Db would be plenty, but you still need to check on a scope if it is stable.
For Modern HI FI speakers you need at least 12DB as they are very poorly damped, as designed for transistors with high damping factors and heaps of feedback.
PA speakers may not need a lot of damping.
For Modern HI FI speakers you need at least 12DB as they are very poorly damped, as designed for transistors with high damping factors and heaps of feedback.
PA speakers may not need a lot of damping.
The dominant pole should be in (or following) the first gain stage. It has to be
within the feedback loop to work. Just an input filter won't help stability.
Usually the amplifier sensitivity for full output is designed to be between 1 and 2 Vrms.
Last edited:
Thank you. That is what I have been simulating, right after the input stage. This is the kind of thing that I am considering right now. In all truth will probably use something like 6L6 rather than 6BQ6. This requires a bit over 2Vrms for full 17W output. 6L6 might require more drive.
Could vary the Zout with an adjustable sense resistor I suppose.
Could vary the Zout with an adjustable sense resistor I suppose.
Attachments
I find lower quality OPTs more acceptable with 20~26dB GNFB, but I spend a lot of time tweaking to get good results. It helps to use a low OPT turns ratio and high-current triode output stages, which leads to grid power drive, which is fine with me. I think more folks ought to explore this design space.
Any thoughts on a current FB loop around just the output stage? There are not a lot of stages in this one to start with. Thinking of adding an input stage outside of FB loop for a little more gain and optional bass boost.
In my investigation of current source PA possibilities I find it is pretty easy to come up with a scheme that does not include the OPT in the loop (my preference) for SE but not so much for class AB PP (if any one has ideas I am all eyes.
So if including the transformer (not necessarily really high quality stuff) how much GNF do you usually shoot for? 15dB seems to be kind af a natural set point in what I am seeing.
Mike
It depends, it depends, it depends.
What are you trying to accomplish? I did a design that used PP 807s as the finals. These 6L6-oids like making a lot of h5 and higher order harmonics. Running open loop both the Twin-T test (the residual looked like a sawtooth at 3X the frequency) and listening showed lots of nasty, dissonant harmonic distortion. I cleaned up this mess with lNFB (~7.0dbv as O. Schade recommended) and added enough gNFB to take the edge off. This came to ~ 19dbv of total NFB.
For another project, I used 6BQ6s. Even though the magnitude of the h3 was higher, the Twin-T residual was pretty much sinusoidal at 3X the frequency, so just a slight amount of higher order harmonics. Open loop, there wasn't any of that "pentode nastiness" until you were nearly at clipping, just an excessively "edgy" or "aggressive" sound. I made the gNFB variable between 0 -- 13dbv of gNFB only (no lNFB) For most listening (metal, Techno) I keep the gNFB pot at the halfway point. Softer music likes a bit more. The full 13dbv gives a "solid statey" sound, and not good SS either.
YMMV, depending on the finals, and the sound you're after.
12BQ6 with cathode + global feedback sound gorgeous. A single 6F12P is enough to drive them. I just used 4 and 16 Ohm taps of Schumakher transformers.
https://www.facebook.com/Wavebourn/videos/1354722704586027/
My thoughts are, when you increase depth of feedback around output tubes you increase distortions of the driver. Also, global feedback gets lower depths as the result for the same targeted sensitivity, that increases overall distortions. However, local feedback around output tubes makes clipping more graceful than the global one, but the global one decreases distortions on lower power caused by all stages. There is no other recipe than searching for an optimum decreasing most audible distortions. I would say, it is a science + art together.
It is exactly what I am doing right now:
https://www.facebook.com/Wavebourn/videos/1367788306612800/
I can understand how to do nested current FB in class A (in fact my first simulations were SE with un-bypassed cathode bias resistors (local) and taking a part of the cathode signal back to the cathode of the input stage (global). But I didn't think that you could use un-bypassed cathode resistors in class AB.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.