I’ve been doing a lot of research in hybrid supplies, One because I have a large of tubes that are in this PS, and another because particular op amps are cheap, and usually you can get very good results. I also have an old Navy Electronics manual
And it shows a pentode tube driving a 6080 tube, and the pentode I happen to have a lot of is the 6CB6, so I went with these, and also I’m a big fan of Walt Jung! That guy knows op amps! Anyway here is my result, I’ve already built it and tested it works great so far, now listening tests are in order. BTW notice the 33K resistor coming to G2 from the regulated side, this is what I found in the Navy Manual, a small amount of feedback. A ripple reducer! Interesting!
One note: the trim pot is 100K not 200K
And lower resistor in divider is 6.98k and not 12K
And it shows a pentode tube driving a 6080 tube, and the pentode I happen to have a lot of is the 6CB6, so I went with these, and also I’m a big fan of Walt Jung! That guy knows op amps! Anyway here is my result, I’ve already built it and tested it works great so far, now listening tests are in order. BTW notice the 33K resistor coming to G2 from the regulated side, this is what I found in the Navy Manual, a small amount of feedback. A ripple reducer! Interesting!
One note: the trim pot is 100K not 200K
And lower resistor in divider is 6.98k and not 12K
Attachments
Last edited:
Nice design and execution. But I am a bit puzzled. The very best stabilized and clean output is the one regulated by the opamp, at the output of the 6080 (after the cathode resistors). Then a cap multiplier is added which makes the output impedance higher and the regulation worse.
What is the thinking behind that?
Jan
What is the thinking behind that?
Jan
It’s no different than a buffer stage, very similar to even using an op amp. I’ve seen many PS’s use this to remove any residual ripple. Thanks
According to my measurements the regulator measured 1mV of ripple at full load, which is about 50mA , and after the addition of the circuit the ripple was cut in half for the same load. The actual measurements of total output noise were 130nV/sqrtHz before and 44nV/sqrtHz after.
The noise figure at even 50Hz is about 1nV!!
The noise figure at even 50Hz is about 1nV!!
Last edited:
That is possible, agreed. But I would argue that a regulator with 1mV ripple isn't very good to begin with.
The problem with the last cap multiplier is that it has a kind of ripple as a result of ac load, due to the output impedance.
It can be explained like this: a higher current out of the transistor requires a higher Vbe. Since the Vbe is fixed, any difference in output current MUST cause a difference in output voltage (emitter voltage). It is like an output resistance, although of course there is no physical resistor.
In your case I would delete the last cap multiplier and see how that regulator can be improved. Replacing V1.1 with a transistor (use the one in the output cap multiplier you don't need ;-) will already greatly improve it. But maybe it is already better than 1mV; with those levels, it very much depends on your wiring and where exactly you measure.
Jan
Last edited: