OOPs.
The phasing doesn't work out right for output plate to driver screen feedbacks on individual P-P sides. Would require a criss-cross which would mess up the SE effect. Only way around I guess would be output plate to driver cathode feedbacks. Not sure if that can be made to work with the anti-triode cathode connections in the driver though.
Hmm, well, if the output stage had CFB windings on the OT, then that could be used for correct phase feedbacks to the driver screens. Seems workabe that way, but CFB xfmrs are tricky to get right (sufficient bandwidth). Imagine that, a "McIntosh" clone that has SE sound. We might get arrested!
The phasing doesn't work out right for output plate to driver screen feedbacks on individual P-P sides. Would require a criss-cross which would mess up the SE effect. Only way around I guess would be output plate to driver cathode feedbacks. Not sure if that can be made to work with the anti-triode cathode connections in the driver though.
Hmm, well, if the output stage had CFB windings on the OT, then that could be used for correct phase feedbacks to the driver screens. Seems workabe that way, but CFB xfmrs are tricky to get right (sufficient bandwidth). Imagine that, a "McIntosh" clone that has SE sound. We might get arrested!
Last edited:
It started out a 6v6. M1 gate used to be the screen. But plate impedance
wasn't quite high enough, nor screen current error low enough, at operating
voltage compatible with direct coupling. Thus a hybrid cascode input stage...
Besides, it keeps the crowd that demands a triode for no better reason
than "its a triode" completely anesthetized. No help today for the other
camp, demanding a Pentode for no better reason than "its a pentode."
Last edited:
My next project will be a direct coupled 6080 amp with the pentode driver hanging off the plates of the outputs. Now that I have two top flight Pentode amps I want to see if there really is an edge with a triode amp.
All I need is a cheap input transformer with a 1:1+1 or 2:1+1 ratio. I built a similar amp using a power toroidal as an interstage, but the capacitances were brutal and I had to push 30mA at it to get top end extension. Any suggestions.
Only when I have built this near perfect triode amp will I have an objective opinion on whether triodes are really superior. For me a well implemented class A PP amp is superior to SE, but most people don't realise this because there are so few such designs out there. Some people still seem to think that the Williamson is the top of the PP heap - sad souls.
Shoog
All I need is a cheap input transformer with a 1:1+1 or 2:1+1 ratio. I built a similar amp using a power toroidal as an interstage, but the capacitances were brutal and I had to push 30mA at it to get top end extension. Any suggestions.
Only when I have built this near perfect triode amp will I have an objective opinion on whether triodes are really superior. For me a well implemented class A PP amp is superior to SE, but most people don't realise this because there are so few such designs out there. Some people still seem to think that the Williamson is the top of the PP heap - sad souls.
Shoog
Why not dump the ECL86 and go for a descent tube driver instead:
or why not dump the tubes all together, lets go for Schaded FETs instead:
Sims so great but Ciss is of course the enemy, still -3dB is above 50kHz. Found a 1kV IXYS at Mouser with half the capacitance of DN2540 so that might be even better.
An externally hosted image should be here but it was not working when we last tested it.
or why not dump the tubes all together
, lets go for Schaded FETs instead:An externally hosted image should be here but it was not working when we last tested it.
Sims so great but Ciss is of course the enemy, still -3dB is above 50kHz. Found a 1kV IXYS at Mouser with half the capacitance of DN2540 so that might be even better.
or why not dump the tubes all together
Because there are far simpler and better ways to Schade a FET.
See Nelson Pass' ZEN. For his Anti-Triode equivalent, see Aleph.
But I do appreciate the academics of such an excercize.
If we wish to Schade a Pentode, or stay on topic, then we
probably do still need some kinda tube and maybe an OPT.
Seems you need the *.asc to see why this is so interesting
. NPs Zen has his drawback in extremely low Zin.
Real life plays some good tricks too: With a cold valve, Q2 will be biassed OFF and full B+ will be applied to M2's gate-source (boof!). C6 will be reverse-biassed at this time too (boof, boof!).
C6 is a small value seeing how it carries full signal current. Typical parts show ohms of impedance at this size, and many have ripple current ratings that would be breached in this position. Shoog used 1000uF here, with good reason.
The voltage drops on C6 will, as noted before, be amplified by the ouput valve as a distortion product, together with any other distortions from the 'lytic.
The performance in dc balance is lower than the top-servo design we worked out above. The screen current dc error appears to be entirely ignored, and the current sensing resolution is so low that inevitable mismatches in Vbe between transistors will show up as unbalanced current in the trafo - I don't see how a toroidal will tolerate this.
Looks to me like the dc and ac performance will both be degraded compared to the design we already posted on this thread, and by the time the self-destruct features are fixed, it will need more parts to build, too.
I had similar issues when i built my direct coupled Tabor clone. On switch on it destroyed all CCS in the tail. It was very frustrating.
In the end I used a 900V IXY part to carry just 18V of bias. this has been stable - but they don't stay well matched as they heat up and so there is more hum than I like and there is probably a bit of core saturation as well.
Getting these issues ironed out is no small task (and can be both frustrating and expensive).
Shoog
In the end I used a 900V IXY part to carry just 18V of bias. this has been stable - but they don't stay well matched as they heat up and so there is more hum than I like and there is probably a bit of core saturation as well.
Getting these issues ironed out is no small task (and can be both frustrating and expensive).
Shoog
Re: my earlier OOPs in post 81.
Thinking about the class AB output problem again this morning, I have concluded that the crossed plate to screen feedbacks I mentioned in post 80 are correct after all for maintaining the triode and anti-triode distortia, besides the correct signal phasing. So my correction in post 81 is wrong. This class AB anti-triode stuff really deserves another thread to do it justice, so I'll leave the details for that.
Lars,
I think my simplification of the two caps, down to one cap, across the Q1,Q2 bases (or M4,M5 gates in latest versions) for filtering is incorrect. AC can still vary both gates together unless the two caps are going to B+ (like Rod has in his diagrams).
Ken,
I thought the Aleph? (or Zen?) just used a counter modulated CCS. While close to anti-triode, it doesn't particularly maintain the sum of currents constant. That arrangement does have some interesting properties for a triode pull-up/load though. If the modulated CCS acts at a level to counter any real load Z and perhaps just a bit more for some additional negative resistance to counter the usual Mu drop-off of triodes at low current.
Have to look up the Aleph & Zen again, I can't remember them in detail. But for an unusual Fet triode, I would put a modulated cascode above the FET with a resistive divider from the output down to the Fet source, to control the cascode's gate. The FET would be operated in its low voltage "triode" zone, with attenuated output on its "plate".
I just looked up Zen 4, which appears to be the one with the counter modulated CCS output. The Aleph J looks like a bootstrapped Mu follower output with some modulated CCS. I didn't notice any Schade like stuff in either one, just global NFB.
Don
Thinking about the class AB output problem again this morning, I have concluded that the crossed plate to screen feedbacks I mentioned in post 80 are correct after all for maintaining the triode and anti-triode distortia, besides the correct signal phasing. So my correction in post 81 is wrong. This class AB anti-triode stuff really deserves another thread to do it justice, so I'll leave the details for that.
Lars,
I think my simplification of the two caps, down to one cap, across the Q1,Q2 bases (or M4,M5 gates in latest versions) for filtering is incorrect. AC can still vary both gates together unless the two caps are going to B+ (like Rod has in his diagrams).
Ken,
I thought the Aleph? (or Zen?) just used a counter modulated CCS. While close to anti-triode, it doesn't particularly maintain the sum of currents constant. That arrangement does have some interesting properties for a triode pull-up/load though. If the modulated CCS acts at a level to counter any real load Z and perhaps just a bit more for some additional negative resistance to counter the usual Mu drop-off of triodes at low current.
Have to look up the Aleph & Zen again, I can't remember them in detail. But for an unusual Fet triode, I would put a modulated cascode above the FET with a resistive divider from the output down to the Fet source, to control the cascode's gate. The FET would be operated in its low voltage "triode" zone, with attenuated output on its "plate".
I just looked up Zen 4, which appears to be the one with the counter modulated CCS output. The Aleph J looks like a bootstrapped Mu follower output with some modulated CCS. I didn't notice any Schade like stuff in either one, just global NFB.
Don
Last edited:
Re: the two caps versus one cap for Q1,Q2 (M4,M5) servo filtering.
I guess it doesn't matter so much with a CCS in the tail, that just makes the tail/CCS bob up and down with the AC when only one cap is used. Probably still better to use the two caps though, no need to exercise the CMRR needlessly. Unless keeping the caps out of the audio path is super important.
I guess it doesn't matter so much with a CCS in the tail, that just makes the tail/CCS bob up and down with the AC when only one cap is used. Probably still better to use the two caps though, no need to exercise the CMRR needlessly. Unless keeping the caps out of the audio path is super important.
That's good. The schematic at least looks better using one cap that is out of the audio path.
Oh, Lars, on your recent Mosfet'd schematics in post #84, the equal gm's of the DN2540's (outputs) will null out the SE effect in the output stage.
The input stage will of course still be putting a SE effect onto the signal.
Which actually brings up an interesting question. The input "triode" in these designs is putting the opposite SE signature onto the signal as the "SE" emulating type output stages. So some driver to output dist. cancellation occurs.
Of course normal SETs have this issue too. But if one wanted maximize 2nd H, maybe the input stage should be a neutral LTP type or a clean OP Amp. or even the anti-triode side of a dual driver should be driving the triode/anti-triode output (just driving the output "triode" side).
Ie, for non dist. cancellation between stages, an anti-triode driver should drive a triode output. (and visa-versa for full P-P, triode drives anti-triode) This is what I found was the key for solving the class AB triode/anti-triode output this morning using class A triode/anti-triode drivers.
For lower capacitance FETs, I was using the Fairchild FQP1N50 or FQP1N60 types (and FQP1P50), but darned if they haven't discontinued them now. At least I bought a couple sleeves of them, will last for a while.
Don
Oh, Lars, on your recent Mosfet'd schematics in post #84, the equal gm's of the DN2540's (outputs) will null out the SE effect in the output stage.
The input stage will of course still be putting a SE effect onto the signal.
Which actually brings up an interesting question. The input "triode" in these designs is putting the opposite SE signature onto the signal as the "SE" emulating type output stages. So some driver to output dist. cancellation occurs.
Of course normal SETs have this issue too. But if one wanted maximize 2nd H, maybe the input stage should be a neutral LTP type or a clean OP Amp. or even the anti-triode side of a dual driver should be driving the triode/anti-triode output (just driving the output "triode" side).
Ie, for non dist. cancellation between stages, an anti-triode driver should drive a triode output. (and visa-versa for full P-P, triode drives anti-triode) This is what I found was the key for solving the class AB triode/anti-triode output this morning using class A triode/anti-triode drivers.
For lower capacitance FETs, I was using the Fairchild FQP1N50 or FQP1N60 types (and FQP1P50), but darned if they haven't discontinued them now. At least I bought a couple sleeves of them, will last for a while.
Don
Last edited:
Member
Joined 2009
Paid Member
The Zen4 is where I'm currently looking for ideas, I might actually build one to hear it for myself.
I'm still trying to understand the tube circuit that's the subject of this thread. I see a 3-winding transformer as the core of it. One winding for the speaker output, one driven by the SE triode and one driven by a CCS. Do I understand correctly that the CCS is NOT modulated as it is with the Zen4 CCS but rather is fixed and used simply to balance the dc current through the output transformer or is it proposed that is will be modulated like the CCS like Zen4 to make the 'anti-triode' ??? - or am I getting things royally confused...
Hey guys,
I wasn´t completely satisfied with the filtering so I tried this one instead. With a simple integrator function at the - input and the +input LP-filtered to B+/ground we null the hum almost completely(anyway if my sims are right):
Don, the sims indicate SE as the MOSFETs doesn´t contribute much to distortion. Seems like all we see is driver distortion.
Bigun, see the output stage like a transformerloaded LTP. The signal is fed from the pentodes cathode into the MOSFETs source.
I wasn´t completely satisfied with the filtering so I tried this one instead. With a simple integrator function at the - input and the +input LP-filtered to B+/ground we null the hum almost completely(anyway if my sims are right
):An externally hosted image should be here but it was not working when we last tested it.
Don, the sims indicate SE as the MOSFETs doesn´t contribute much to distortion. Seems like all we see is driver distortion.
Bigun, see the output stage like a transformerloaded LTP. The signal is fed from the pentodes cathode into the MOSFETs source.
Lars, the FETs all have stoppers in their gates, but since their required bandwidth is so low, you could use higher values for all of them. I have found stability improves for 1K stoppers in some circuits, and if you used something like 100K - you might get some free extra filtering!
the only precaution - for higher values of stoppers it is more important to be sure that they are mounted very near the gate pin.
the only precaution - for higher values of stoppers it is more important to be sure that they are mounted very near the gate pin.
Lars, C3 sounds like a good idea! Be careful though - C3 will make the FET gate follow the drain at power ON, & you may overstress the gate oxide. A gate-source zener might be advisable.
You could also change the FETs back to MPSA92 (or Zetex ZTX550 series if you like higher voltage). With C3 = 470n/400V polypropylene, 47K base resistor will probably be enough - Mr Miller is multiplying the 470n by the gain - and the differential gain of each stage is very large.
BJTs will give better accuracy and temperature stability, and don't need a gate stopper or zener in this position.
Use a "C3" on both sides of the differential amp.
You could also change the FETs back to MPSA92 (or Zetex ZTX550 series if you like higher voltage). With C3 = 470n/400V polypropylene, 47K base resistor will probably be enough - Mr Miller is multiplying the 470n by the gain - and the differential gain of each stage is very large.
BJTs will give better accuracy and temperature stability, and don't need a gate stopper or zener in this position.
Use a "C3" on both sides of the differential amp.
Lars and Rod,
Nice upgrades. Looking good.
Bigun,
The "CCS" winding side (the anti-triode side here) is modulated, similar to the Zen, but more constrained here. The Bottom CCS on the tied cathode/sources (of triode and anti-triode sides) is forcing the two sides to sum to a constant current. So they are exactly complementary currents on the two sides outputs to the windings (triode versus anti-triode sides).
(the servo circuit in this thread, attached to the Mosfet gate, may be causing some confusion, its only for DC balancing. The Mosfet gate is essentially just at a fixed DC voltage as far as the audio operation is concerned)
The modulated CCS in the Zen is free to be adjusted to any gain, so is not constrained in this way. Also, there are less constraints on the distortion spectra of the modulated CCS there, since it is simply driven from a gate or base input. The anti-triode here is driven from the source (or cathode for a tube case) and so the same current in, also comes out the drain or plate. (ie, no current lost thru gate or grid, other than small capacitive effects) Precise complementarity here, not approximate complemetarity like Zen or ordinary P-P.
In addition, the triode side, here, is operated in a way so as to preserve its environment like a SE triode. This is accomplished by keeping it's cathode voltage as near constant as possible. This is where a large gm for the anti-triode device is very helpful. Very little voltage variation on the cathode/source connection is required to operate the anti-triode. If the two sides have equal gm's you just get ordinary triode class A P-P.
So there are gross similarities between triode/anti-triode and Zen or ordinary P-P. The difference is in the details and distortion spectra.
-----------
small correction to my post 91:
For class AB output: crossed plate to screen fdbks in post 80 AND the non-crossed CFB cathodes to screen fdbks in post 81 are BOTH workable with the triode/anti-triode driver stage.
Don
Nice upgrades. Looking good.
Bigun,
The "CCS" winding side (the anti-triode side here) is modulated, similar to the Zen, but more constrained here. The Bottom CCS on the tied cathode/sources (of triode and anti-triode sides) is forcing the two sides to sum to a constant current. So they are exactly complementary currents on the two sides outputs to the windings (triode versus anti-triode sides).
(the servo circuit in this thread, attached to the Mosfet gate, may be causing some confusion, its only for DC balancing. The Mosfet gate is essentially just at a fixed DC voltage as far as the audio operation is concerned)
The modulated CCS in the Zen is free to be adjusted to any gain, so is not constrained in this way. Also, there are less constraints on the distortion spectra of the modulated CCS there, since it is simply driven from a gate or base input. The anti-triode here is driven from the source (or cathode for a tube case) and so the same current in, also comes out the drain or plate. (ie, no current lost thru gate or grid, other than small capacitive effects) Precise complementarity here, not approximate complemetarity like Zen or ordinary P-P.
In addition, the triode side, here, is operated in a way so as to preserve its environment like a SE triode. This is accomplished by keeping it's cathode voltage as near constant as possible. This is where a large gm for the anti-triode device is very helpful. Very little voltage variation on the cathode/source connection is required to operate the anti-triode. If the two sides have equal gm's you just get ordinary triode class A P-P.
So there are gross similarities between triode/anti-triode and Zen or ordinary P-P. The difference is in the details and distortion spectra.
-----------
small correction to my post 91:
For class AB output: crossed plate to screen fdbks in post 80 AND the non-crossed CFB cathodes to screen fdbks in post 81 are BOTH workable with the triode/anti-triode driver stage.
Don
Last edited:
I was thinking about pentode-antipentode driver + inverter("splitter") loaded directly with separate plate feedback (local feedback on each side)
http://www.diyaudio.com/forums/showthread.php?t=151545
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.