Oh, I meant just deleting C3 and C4 to an open circuit condition to remove the source Fdbk paths altogether. To get to the 0.005% model you mentioned. (same as your 2nd schematic, just above, shows)
Yes, for using the SG1 and SG2 taps in place of the plates for both, as just an experiment. Not a final solution necessarily unless it works better. (and keeping DC isolation caps in place as already configured/needed) (also, some adjustment of the Fdbk resistors needed to compensate for the lower AC voltage at the screen taps versus the plate taps)
I trust the screen taps on real OTs to be better coupled to the secondary than the primary plate ends. It may not show much difference in an ideal Spice sim however. The main issue is whether that provokes any oscillation due to the leakage L between plate tap and UL tap (which is now in the N Fdbk loop causing some phase shift at HF).
Yes, for using the SG1 and SG2 taps in place of the plates for both, as just an experiment. Not a final solution necessarily unless it works better. (and keeping DC isolation caps in place as already configured/needed) (also, some adjustment of the Fdbk resistors needed to compensate for the lower AC voltage at the screen taps versus the plate taps)
I trust the screen taps on real OTs to be better coupled to the secondary than the primary plate ends. It may not show much difference in an ideal Spice sim however. The main issue is whether that provokes any oscillation due to the leakage L between plate tap and UL tap (which is now in the N Fdbk loop causing some phase shift at HF).
Last edited:
I connected both feedback taps to the screens. I changed R19 and R20 to 100k, to increase negative feedback signal to the MOSFET sources. Compensates for lower AC voltage at the screen taps vs. the plate taps.
The result was slightly higher distortion than the combination Schadeode/E-Linear from before. 1kHz THD @ 1W into 4 ohms = 0.044%, but the increase is all 2nd harmonic (-68dB instead of -75dB). 3rd harmonic stayed down at about -75dB. That might actually sound a little nicer at low to medium levels. THD at 23W out rose to a hair over 1%.
Interesting. I guess there's a possibility that feedback from the UL screen tap might interact unfavorably with the feedback loop from the output tube's plate to grid?
Is this a non-issue when using plate-grid local NFB around a pentode (no UL screen taps in the OPT)? Or would this kind of NFB using a pentode hookup be more likely to cause oscillations in an OPT? Would this be a reason why an E-Linear hookup might sound better than it simulates (or measures)?
I was hoping to use the simplest 'Schadeode' style circuit possible, using as few parts as possible,so I could cram it all into this Dyna ST70 chassis. But it looks like I should find a nice, big chassis in which to build this. I'd like to have room to dry different things without getting all twisted up in there. The dang ST70 chassis is so small...
--
Last edited:
N Fdbk from the plates should be the safest for stability due to only the output tube being in the loop (or output and driver for the CFB driver type loop). (ie, no OT in loop) Taking the N Fdbks from the UL taps is more dangerous for stability, since the plate output has (somewhat) to pass -thru- the OT to get to the UL tap (although the auto-xfmr effect should help to minimise any phase shift there, ie, both the plate tap and the UL tap have the plate current passing thru the winding, ideally the UL tap would just be an attenuated version of the plate signal )
As long as the UL taps proves to be stable for the various N Fdbks pick-off, then I would expect it to provide the lowest distortion at the secondary, being typically more closely coupled to the secondary than the plate taps. (I have measured some single bobbin OTs, and the outer-most plate winding tap has rather high leakage L typically, while the inner-most plate winding tap has lower leakage L, asymmetric, the UL taps are better matched and more closely coupled to the secondary) No guarantee that the UL taps are more closely coupled on all OTs, but there is less AC variation on the UL winding section, so it is good winding practice to place that section close to the secondary windings. And UL operation itself demands that the UL winding be closely coupled to the secondary for the UL N Fdbks to work well.
As long as the UL taps proves to be stable for the various N Fdbks pick-off, then I would expect it to provide the lowest distortion at the secondary, being typically more closely coupled to the secondary than the plate taps. (I have measured some single bobbin OTs, and the outer-most plate winding tap has rather high leakage L typically, while the inner-most plate winding tap has lower leakage L, asymmetric, the UL taps are better matched and more closely coupled to the secondary) No guarantee that the UL taps are more closely coupled on all OTs, but there is less AC variation on the UL winding section, so it is good winding practice to place that section close to the secondary windings. And UL operation itself demands that the UL winding be closely coupled to the secondary for the UL N Fdbks to work well.
Last edited:
smoking-amp said:
Thanks for the reply. I'll be using Dynaco A470 OPTs (from a Stereo 70). They're well thought of, which I hope means that they were wound well. I suppose this is a build-it-and-see kind of thing. The 'Schadeode' circuit uses very few parts, which would make it good for experimenting. I'll try to lay it out so that I can move the feedback connections between 6L6 plates and screens, to compare. Or disconnect the screen taps and go pentode. All look good in simulation.
Wavebourn said:
I've seen mention of this many times. SY wrote about this circuit a number of times too. I really need to build one of these, but I would want to change tube types. 6L6GC instead of 7027, probably 6AC7 instead of 6CB6A (?), and would I really need that 7199 up front? Why not a 6SN7 cathodyne for the phase splitter? Or if the pentode voltage amp is really necessary, I suppose a 6U8A or 6GH8A would work. Or 6F12P?
Last edited:
I saw mention of this post from smoking-amp, describing the square-law characteristics of 12AT7 in relation to the infamous RH series 'Schade' feedback amps, with mention of how MOSFETs have a similar square-law characteristic that offers possibilities for distortion cancellation in 'Schade' feedback setups.
http://www.diyaudio.com/forums/tubes-valves/159191-couple-idiot-questions.html#post2057579
If I'm understanding that correctly, it got me to wondering if the behavior of a MOSFET driver stage could be why M. Koster's "Schadeode" circuit functions so well. If so, then the push-pull version should be quite good for a higher power amp with low distortion. It might not be quite as good as the RCA 50-Watt Amplifier, but it's a heck of a lot simpler. I mean a LOT simpler.
Could be...
--
http://www.diyaudio.com/forums/tubes-valves/159191-couple-idiot-questions.html#post2057579
If I'm understanding that correctly, it got me to wondering if the behavior of a MOSFET driver stage could be why M. Koster's "Schadeode" circuit functions so well. If so, then the push-pull version should be quite good for a higher power amp with low distortion. It might not be quite as good as the RCA 50-Watt Amplifier, but it's a heck of a lot simpler. I mean a LOT simpler.
Could be...
--
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.