Hi All,
As I was tidying up my workshop, I naturally came across an old Music Angel XD-800MkII for EL34's that I'd bought as a DIY project, without tubes, but with case and transformers. GU50 was (is!) the tube I wanted to use.
So I'd merrily started this about a decade ago, but then got distracted onto something else, and never managed to get back to it. As such, it needs tidying up, I.e populating with tubes, and a circuit, etc.
So the original Music Angle (KT88) schematic floats about, and there's the Dynaco ST70 design which is nice, but that's all been done before, and my original concept was to build a local feedback PP, due to the wonderful sound from my local feedback SE.
Anyway, if I just built old stuff, where would I be? Surrounded by nice, working things? The horror..
So anyway after a few false starts and dead ends (LTPs, I'm looking at you!), this is what I came up with. LtSpice only so far.
The 'slightly unconventional' anode resistors for the 6AU6's (which I'm pretty sure I have in stock, from an old raid on a Tek 545b) to B+ is to give it a 'more CCS like' anode load, without having to add Yet More Bits'.
Caveats: I'm not 100% sure of the B+ voltage. Or the output transformers. Or really, the tube models.. but I've attached them.
The tubelibrary I downloaded from the tube Spice model thread, but I re-wired the GU50 model to match the tetrode symbol. I'm not sure why they don't use the pentode symbol, perhaps some SPICE vs LTSPICE thing.
So I don't get much feedback in there, 6dB or so, but I'm low on gain. But at least this concertina + Schade resists the temptation to turn into an oscillator, so that's encouraging. I could add an input stage, of course, but that's another stage, more distortion and more bits. No feedback beyond the primary, as that would be madness
I did try the GU50 in triode mode too, but that was hopeless. In Pentode mode literally everything is better. This design has a wonderful soft clipping too, tubes seem far more 'elastic' than MOSFETs !
Tubes are chosen as to what I already have, a sure fire method I use to ensure I can build stuff...
Please point out any easy fixes, improvements, etc, as in my ideal dream world, I'll build it and it'll be magnificent.
At some point I'll send a small AC signal into the transformers and work out what the ratios might be too. 🔬
Biasing is a bit glossed over, it's awful in the original, but if anyone has a good circuit for PP tube balance (Pot 1) and level (Pot 2) please can you post it? A handy method to casually null the PP magetic field would be cool, with absolute level setting being demoted to a 'baseplate off' think as that drifting doesn't worry me.
All comments welcome, many thanks in advance!
As I was tidying up my workshop, I naturally came across an old Music Angel XD-800MkII for EL34's that I'd bought as a DIY project, without tubes, but with case and transformers. GU50 was (is!) the tube I wanted to use.
So I'd merrily started this about a decade ago, but then got distracted onto something else, and never managed to get back to it. As such, it needs tidying up, I.e populating with tubes, and a circuit, etc.
So the original Music Angle (KT88) schematic floats about, and there's the Dynaco ST70 design which is nice, but that's all been done before, and my original concept was to build a local feedback PP, due to the wonderful sound from my local feedback SE.
Anyway, if I just built old stuff, where would I be? Surrounded by nice, working things? The horror..
So anyway after a few false starts and dead ends (LTPs, I'm looking at you!), this is what I came up with. LtSpice only so far.
The 'slightly unconventional' anode resistors for the 6AU6's (which I'm pretty sure I have in stock, from an old raid on a Tek 545b) to B+ is to give it a 'more CCS like' anode load, without having to add Yet More Bits'.
Caveats: I'm not 100% sure of the B+ voltage. Or the output transformers. Or really, the tube models.. but I've attached them.
The tubelibrary I downloaded from the tube Spice model thread, but I re-wired the GU50 model to match the tetrode symbol. I'm not sure why they don't use the pentode symbol, perhaps some SPICE vs LTSPICE thing.
So I don't get much feedback in there, 6dB or so, but I'm low on gain. But at least this concertina + Schade resists the temptation to turn into an oscillator, so that's encouraging. I could add an input stage, of course, but that's another stage, more distortion and more bits. No feedback beyond the primary, as that would be madness
I did try the GU50 in triode mode too, but that was hopeless. In Pentode mode literally everything is better
. This design has a wonderful soft clipping too, tubes seem far more 'elastic' than MOSFETs !Tubes are chosen as to what I already have, a sure fire method I use to ensure I can build stuff...
Please point out any easy fixes, improvements, etc, as in my ideal dream world, I'll build it and it'll be magnificent.
At some point I'll send a small AC signal into the transformers and work out what the ratios might be too. 🔬
Biasing is a bit glossed over, it's awful in the original, but if anyone has a good circuit for PP tube balance (Pot 1) and level (Pot 2) please can you post it? A handy method to casually null the PP magetic field would be cool, with absolute level setting being demoted to a 'baseplate off' think as that drifting doesn't worry me.
All comments welcome, many thanks in advance!
Attachments
Thanks for that, very interesting, especially the 12AU7, I did look at doing that with a 12AT7. The concertina into the pentodes is what I found worked best,
... although my original plan for a phase splitter was a beam deflection tube, I should have a couple somewhere, there's a thread from 2012 that I bought a couple for this project, I'll have to dig them out and see what I got.
https://www.diyaudio.com/community/threads/john-swensons-in-my-eyes-revolutionary-preamp.37499/
The local feedback to the cathodes of the 6AU6 - without a capacitor, having only 320V on the main anodes helps, interesting screen grid arrangement too.. I'll have to look at some of these ideas.
Having now seen RCA make a local feedback PP... also tells me I'm on the right path, thanks!
Attachments
Looks like you were planning output stage shunt feedback. That works well. The RCA does series feedback to the driver cathode, which can work better because you can apply more feedback. The RCA circuit incorporates two stages in the feedback so you have much more overall gm to deal with in the feedback loop.
My Corona SE amp does basically the same thing as the SP10, but I have increased/improved the feedback in a few ways.
First, I added more gain by CCS-loading the driver. CCS load on a pentode gives a lot of gain that can be converted to feedback.
Second, I chose a higher-gain pentode (6BN11, which is a dual 6EW6). Again, for more gain and more feedback.
Third, I buffered the feedback at the driver cathode. I've experimented with using a p-channel fet follower and an op-amp follower to do that job. Distortion performance was identical but the fet follower has a voltage shift that must be dealt with. The important thing that buffering the feedback divider accomplishes is that it stops local degenerative feedback and screen current variations from corrupting that feedback point. The improvement you get in distortion reduction by doing this is astounding.
I'm getting ~.02% distortion at 1W this way with absolutely no feedback around the output transformer.
That said, I expect that the approach in the RCA amp would deliver excellent performance even without those extra tricks.
The shunt feedback you are proposing in the opening post will also work well (I've been down that road too), but the distortion will definitely be higher than the RCA amp approach.
My Corona SE amp does basically the same thing as the SP10, but I have increased/improved the feedback in a few ways.
First, I added more gain by CCS-loading the driver. CCS load on a pentode gives a lot of gain that can be converted to feedback.
Second, I chose a higher-gain pentode (6BN11, which is a dual 6EW6). Again, for more gain and more feedback.
Third, I buffered the feedback at the driver cathode. I've experimented with using a p-channel fet follower and an op-amp follower to do that job. Distortion performance was identical but the fet follower has a voltage shift that must be dealt with. The important thing that buffering the feedback divider accomplishes is that it stops local degenerative feedback and screen current variations from corrupting that feedback point. The improvement you get in distortion reduction by doing this is astounding.
I'm getting ~.02% distortion at 1W this way with absolutely no feedback around the output transformer.
That said, I expect that the approach in the RCA amp would deliver excellent performance even without those extra tricks.
The shunt feedback you are proposing in the opening post will also work well (I've been down that road too), but the distortion will definitely be higher than the RCA amp approach.