Hi,
I've been playing with a tube/MOSFET hybrid of an unusual design and I thought I would share my experiences. This group seems to be receptive to wacky ideas.
The idea was described on this forum some time ago by forum member smoking amp. It would work with either a MOSFET or a good NPN darlington, like a horizontal output from a TV circuit.
Basically, the idea is to build a push-pull amplifier using a tube on one side and a MOSFET current mirror on the other.
I've implemented it using a cathode current bridge and an op amp driving the MOSFET to self-null the bridge for AC signals. The result is an exactly opposite signal current through the MOSFET, which is applied to the other end of the PP transformer primary.
The DC balance is done simply by LPF'd positive feedback to the op amp + input where it provides a DC reference for the bridge. This may not be the best scheme for transient response as the transformer gets unbalanced by the "distortion current" at high signal levels while the LPF settles. Maybe tune it but that's another story...
Anyway, I've tested it so far on a resistive load and it's very well behaved, with essentially SE harmonic profile. I used a 6BX7 because it's predictably non-linear ;-)
100mW .95% THD
250 mW 1.5% THD
1W 3.1% THD
4W 5.5% THD (Design max power)
with the residuals mostly 2nd harmonic (f2) with a "waterfall" series of f2, f3, and a little f5 until higher power, when the higher order harmonics show up at the onset of output stage grid current.
I think I'm done on the bench and ready to hook it up to a speaker to see what happens with a reactive load & through speaker resonance. If the concept is going to break down it should show up there.
I'll post the plots and other pics but it looks like I can only do done at a time.
Cheers,
Michael J Koster
http://redwoodcoastmusic.com
I've been playing with a tube/MOSFET hybrid of an unusual design and I thought I would share my experiences. This group seems to be receptive to wacky ideas.
The idea was described on this forum some time ago by forum member smoking amp. It would work with either a MOSFET or a good NPN darlington, like a horizontal output from a TV circuit.
Basically, the idea is to build a push-pull amplifier using a tube on one side and a MOSFET current mirror on the other.
I've implemented it using a cathode current bridge and an op amp driving the MOSFET to self-null the bridge for AC signals. The result is an exactly opposite signal current through the MOSFET, which is applied to the other end of the PP transformer primary.
The DC balance is done simply by LPF'd positive feedback to the op amp + input where it provides a DC reference for the bridge. This may not be the best scheme for transient response as the transformer gets unbalanced by the "distortion current" at high signal levels while the LPF settles. Maybe tune it but that's another story...
Anyway, I've tested it so far on a resistive load and it's very well behaved, with essentially SE harmonic profile. I used a 6BX7 because it's predictably non-linear ;-)
100mW .95% THD
250 mW 1.5% THD
1W 3.1% THD
4W 5.5% THD (Design max power)
with the residuals mostly 2nd harmonic (f2) with a "waterfall" series of f2, f3, and a little f5 until higher power, when the higher order harmonics show up at the onset of output stage grid current.
I think I'm done on the bench and ready to hook it up to a speaker to see what happens with a reactive load & through speaker resonance. If the concept is going to break down it should show up there.
I'll post the plots and other pics but it looks like I can only do done at a time.
Cheers,
Michael J Koster
http://redwoodcoastmusic.com
Attachments
I thought it was my idea. 😕
http://www.diyaudio.com/forums/showthread.php?postid=1415022#post1415022
http://www.diyaudio.com/forums/showthread.php?postid=1415022#post1415022
Hi,
Very interesting! I thought it was my idea at first too, but someone pointed me to smoking-amp's earlier posts.
I would very much like to see your ideas on implementing this.
Any ideas on how the DC balance scheme might be done better. Do you think there may be a problem with the slow reaction time ?
Before we get a dozen posts about the MOSFET being backwards, here's a corrected schematic.
Michael
Very interesting! I thought it was my idea at first too, but someone pointed me to smoking-amp's earlier posts.
I would very much like to see your ideas on implementing this.
Any ideas on how the DC balance scheme might be done better. Do you think there may be a problem with the slow reaction time ?
Before we get a dozen posts about the MOSFET being backwards, here's a corrected schematic.
Michael
Attachments
Michael Koster said:
I would bypass a cathode by cap taking from it DC only that reflects current through triode using this voltage to bias an opamp,
resistor of the same value (un-bypassed!) would use in emitter of a transistor (in source in your case), taking feedback by current from it.
Also, I would feed both triode and opamp from a phase splitter balancing signals for less distortions.
The idea I got from my previous hybrid design where I used a voltage follower loaded on a voltage to current converter. It sounds pristine clean, while if to replace a voltage follower by a triode there will be a "SE triode sound" instead from a small push-pull transformer.
Here was the original:
Here is the distortion spectra of a 1 KHz sine wave.
The output spectrum and the driver spectrum are similar, so I conclude that it is indeed a SEPP amplifier producing a SE characteristic using a PP transformer. f2 is not canceled in the OPT as in PP (never expected it to)
My measurement setup is using a HP339 with the residual monitor output going to a PC running RTA software with a 24/96 sound card. Some of the fundamental shows up in the residual because my 339 needs alignment.
Michael
The output spectrum and the driver spectrum are similar, so I conclude that it is indeed a SEPP amplifier producing a SE characteristic using a PP transformer. f2 is not canceled in the OPT as in PP (never expected it to)
My measurement setup is using a HP339 with the residual monitor output going to a PC running RTA software with a 24/96 sound card. Some of the fundamental shows up in the residual because my 339 needs alignment.
Michael
Hmmm,
Well, my idea was a little different in that it used a CCS'd LTP configuration to drive the complementary current (versus the triode) thru the Mosfet/Pentode side (acting like a folded cascode stage effectively).
No Op Amp was needed that way. I see Michael is using the Op Amp to do the DC biasing and the AC complementing. Similar setup in Wavebourn's.
An Op amp could be used as a bias balancing servo to good effect in the CCS'd LTP approach by controlling the gate DC. But I like the clean current differencing for AC of the CCS'd LTP versus the Op Amp AC servos.
Obviously, everyone here is thinking the same overall idea of making a truly accurate complementary drive on the other side of the P-P xfmr.
https://www.diyaudio.com/community/...omplementary-current-triode.56916/post-638668
or just a Mosfet instead of the Mosfet/pentode cascode:
https://www.diyaudio.com/community/...mp-from-class-a-push-pull.109977/post-1323780
Don
Oh, I would be careful about using a splitter and driving both sides. A diffl LTP splitter probably will introduce some odd harmonic signature of it's own into the sound. A concertina should be safe though, I would think.
The amplifier I built using a triode configured pentode on one side and a Mosfet (FQP2N90) on the other side sounded like a SET to me, so no surprise that the measurements confirm this. Plenty of expensive SET owners/ builders out there that will be surprised (dis-believing maybe?) though.
Well, my idea was a little different in that it used a CCS'd LTP configuration to drive the complementary current (versus the triode) thru the Mosfet/Pentode side (acting like a folded cascode stage effectively).
No Op Amp was needed that way. I see Michael is using the Op Amp to do the DC biasing and the AC complementing. Similar setup in Wavebourn's.
An Op amp could be used as a bias balancing servo to good effect in the CCS'd LTP approach by controlling the gate DC. But I like the clean current differencing for AC of the CCS'd LTP versus the Op Amp AC servos.
Obviously, everyone here is thinking the same overall idea of making a truly accurate complementary drive on the other side of the P-P xfmr.
https://www.diyaudio.com/community/...omplementary-current-triode.56916/post-638668
or just a Mosfet instead of the Mosfet/pentode cascode:
https://www.diyaudio.com/community/...mp-from-class-a-push-pull.109977/post-1323780
Don
Oh, I would be careful about using a splitter and driving both sides. A diffl LTP splitter probably will introduce some odd harmonic signature of it's own into the sound. A concertina should be safe though, I would think.
The amplifier I built using a triode configured pentode on one side and a Mosfet (FQP2N90) on the other side sounded like a SET to me, so no surprise that the measurements confirm this. Plenty of expensive SET owners/ builders out there that will be surprised (dis-believing maybe?) though.
Last edited by a moderator:
smoking-amp said:
No, my idea is to mimic a DC current through the main active element only providing a variable current by a device with high dynamic resistance so overall output curve will be determined by a lower resistance device (the main active element) that works in privileged conditions.
Absolutely agree.
When one 6P3S tube in my PP amp run away thermally I had similar SET signature.
I still have no time to breadboard my version of a triode SE with PP output transformer... I even found a name for it, "Alligator" 😀
smoking-amp said:
Actually, I think your "figure 2" is functionally more similar where you get the MOSFET drive from the cathode, only I use the op amp to produce an exact AC current mirror around the bias point.
Now that I think about it, the LTP approach does seem to be a refinement on the basic idea. But where does the distortion current go? By this I mean the increase in average current (conductance, Rp...) at full signal vs. quiescent.
hehe something like this would probably not appeal to any diehard SET fancier
Wavebourn said:
Yes, this is exactly the way I think about it. I guess the "privileged conditions" are provided by the high dynamic resistance of the MOSFET shoulder which should give the triode a SET world view 🙂
as opposed to the dual tube PP where the dynamic resistance change of the opposite is the mechanism which cancels f2 distortion (and also the mechanism which converts f2 to f3 in diffamps)
It will be great to compare notes.
Michael
Attachments
Re: Wavebourn
Ahh, yes. The voltage amplifier and complementarily modulated current source P-P scheme. A somewhat different approach to getting SET sound. Some additional flexibility there since the voltage controlled current can be set up to act like a complementary current or can be set higher to keep the triode current constant or any shade in-between. Looks like Michael's AC servo control could be set up to do similar things also.
When I conjured up the CCS'd LTP triode/Mosfet version, I was encountering heavy resistance to the idea that a P-P design could accurately mimic a SET design soundwise, so I was determined to get exactly the same harmonic spectrum by accurate complementary current.
Some resistance even arose to the idea that a cascode Mosfet stage could have the same current out from the drain as input to it thru the source. In any case, it is still important to use a low capacitance Mosfet (with at least 30V across it to minimize capacitance variation), which Fairchild solved nicely just in time for me with the FQP... series.
Nice to get twice as much power out too (compared to a single tube SET)
So, Anatolyi, does this "Alligator" amplifier have plans to devour SET amps for lunch?
Don
Ahh, yes. The voltage amplifier and complementarily modulated current source P-P scheme. A somewhat different approach to getting SET sound. Some additional flexibility there since the voltage controlled current can be set up to act like a complementary current or can be set higher to keep the triode current constant or any shade in-between. Looks like Michael's AC servo control could be set up to do similar things also.
When I conjured up the CCS'd LTP triode/Mosfet version, I was encountering heavy resistance to the idea that a P-P design could accurately mimic a SET design soundwise, so I was determined to get exactly the same harmonic spectrum by accurate complementary current.
Some resistance even arose to the idea that a cascode Mosfet stage could have the same current out from the drain as input to it thru the source. In any case, it is still important to use a low capacitance Mosfet (with at least 30V across it to minimize capacitance variation), which Fairchild solved nicely just in time for me with the FQP... series.
Nice to get twice as much power out too (compared to a single tube SET)
So, Anatolyi, does this "Alligator" amplifier have plans to devour SET amps for lunch?
Don
The scalable amplifier. I call it the "Variamp" because you can dial in whatever power multiplier you want. You do need to reduce the turns ratio for higher currents so maybe a variac-like OPT would be in order.
Actually you don't need to reduce the turns ratio. This technique can also be used to improve the damping factor of tube amps without resorting to NFB.
BTW, there is a true push pull version of this where you have 2 tubes in PP and 2 opposing current mirrors. I have been calling it Perfect Push Pull because you can ideally balance both AC and DC (no issues with current shift due to signal level).
The Variamp would have switch settings for SE and PP flavors at various power levels and damping factors.
Michael
PS I think any of these designs would be a great idea for type 45s because they're so rare and expensive (only need one and no matching problems) and you could drive a wider range of speakers with it.
Actually you don't need to reduce the turns ratio. This technique can also be used to improve the damping factor of tube amps without resorting to NFB.
BTW, there is a true push pull version of this where you have 2 tubes in PP and 2 opposing current mirrors. I have been calling it Perfect Push Pull because you can ideally balance both AC and DC (no issues with current shift due to signal level).
The Variamp would have switch settings for SE and PP flavors at various power levels and damping factors.
Michael
PS I think any of these designs would be a great idea for type 45s because they're so rare and expensive (only need one and no matching problems) and you could drive a wider range of speakers with it.
Attachments
Re: Throwing in my hat too
Interesting...
What's for the transistor in parallel with the tube?
kenpeter said:
Interesting...
What's for the transistor in parallel with the tube?
Re: Kenpeter remember...
Oh Boy! Now we have triode emulation and anti-triode emulation.
Better be careful, those SET fanciers will be fit to be tied now. I'll throw in one more outrageous design, then I'm going to hide for a few days until the flames subside. (Even Broskie got purple- plexed about this one.) Probably have to change our names to get back on the forum if this keeps up.
http://www.diyaudio.com/forums/showthread.php?postid=240542#post240542
Don
Oh Boy! Now we have triode emulation and anti-triode emulation.
Better be careful, those SET fanciers will be fit to be tied now. I'll throw in one more outrageous design, then I'm going to hide for a few days until the flames subside. (Even Broskie got purple- plexed about this one.) Probably have to change our names to get back on the forum if this keeps up.
http://www.diyaudio.com/forums/showthread.php?postid=240542#post240542
Don
smoking-amp said:
Yes, this is the idea! My "Tower" room heaters (schemo of one above) ate them already, but consumed 12A from 120V outlet constantly. Audiofools did not like them because they sounded too clean, "No signature". 😀 Despite they had no "Transistor Signature" as well! Also, they were insane (no massive output toobs), regarded as a shameless heresy!
An "Alligator" will sound less pristine, but it will have a nice triode SET signature with cheaper but better output transformers (less of iron and copper is needed).
I think of it as a commercial project, but schemo will be available in public domain...
Re: Michael
"But where does the distortion current go? By this I mean the increase in average current (conductance, Rp...) at full signal vs. quiescent."
Good point, my setup was using a large 100 Watt Edcor output xfmr which was able to handle the current imbalance at large signal level. But it really should have something added to fix this.
Maybe a servo control that adjusts the tail current to keep the triode and Mosfet avg. "DC" currents balanced. (as long as the tube has monotonically increasing DC avg. current with signal level, but could be a problem when hitting triode saturation though since the servo loop polarity might reverse.)
Or a servo to add avg. "DC" current to the Mosfet side (another Mosfet current source in parallel) to balance up the xfmr.
Without the servo, one can also just adjust the idle bias current off center some too (if the xfmr can handle it). Make the offset at idle equal to the opposite offset at max signal, then you only need to deal with 1/2 the original max offset current imbalance in the xfmr. Not too good for a toroidal xfmr though.
Don
"But where does the distortion current go? By this I mean the increase in average current (conductance, Rp...) at full signal vs. quiescent."
Good point, my setup was using a large 100 Watt Edcor output xfmr which was able to handle the current imbalance at large signal level. But it really should have something added to fix this.
Maybe a servo control that adjusts the tail current to keep the triode and Mosfet avg. "DC" currents balanced. (as long as the tube has monotonically increasing DC avg. current with signal level, but could be a problem when hitting triode saturation though since the servo loop polarity might reverse.)
Or a servo to add avg. "DC" current to the Mosfet side (another Mosfet current source in parallel) to balance up the xfmr.
Without the servo, one can also just adjust the idle bias current off center some too (if the xfmr can handle it). Make the offset at idle equal to the opposite offset at max signal, then you only need to deal with 1/2 the original max offset current imbalance in the xfmr. Not too good for a toroidal xfmr though.
Don
Re; Wavebourn "Don, what about this one?"
Hmmm, are those diodes supposed to be thermionic ones? I think this will make for lots of distortion, since triodes are supposed to have constant Mu, not be diode functions. This looks like emulation of the problems, not the benefits, of triodes. Back to the drawing board.
Since we are on this diode kick, I have seen a patent for a SS amp where they put thermionic diodes in series with the output transistors bases (usual emitter follower output, complementary totem pole), so they would have transconductances similar to tubes. Only problem is, it is similar to pentodes, not triodes. But maybe it helps with smoother crossover.
Don
Hmmm, are those diodes supposed to be thermionic ones? I think this will make for lots of distortion, since triodes are supposed to have constant Mu, not be diode functions. This looks like emulation of the problems, not the benefits, of triodes. Back to the drawing board.
Since we are on this diode kick, I have seen a patent for a SS amp where they put thermionic diodes in series with the output transistors bases (usual emitter follower output, complementary totem pole), so they would have transconductances similar to tubes. Only problem is, it is similar to pentodes, not triodes. But maybe it helps with smoother crossover.
Don
Oh, here is another one for possible SET "emulation":
http://www.diyaudio.com/forums/showthread.php?postid=871030#post871030
Put a real SET amp in for the top corrector amp, and a SS amp for the main power amp (bottom). It does require some feedback for the SET amp though, which is not likely to give the usual open loop SET sound unfortunately.
Maybe could move the feedback subtraction point forward to compare the final (total) amplifier output signal with the SET plate signal and feed back correction to the SET input to minimize the difference. Maybe some positive fdbk that way though. Must be some way.
Wait, I got it. Make the SET amp with partial cathode feedback (CFB). That way the SS amp is effectively supplying the plate voltage for the open loop SET amp (backward thru its OT), and the SET amp is effectively monitoring the final speaker output thru the CFB fdbk. Well, low feedback anyway.
Don
http://www.diyaudio.com/forums/showthread.php?postid=871030#post871030
Put a real SET amp in for the top corrector amp, and a SS amp for the main power amp (bottom). It does require some feedback for the SET amp though, which is not likely to give the usual open loop SET sound unfortunately.
Maybe could move the feedback subtraction point forward to compare the final (total) amplifier output signal with the SET plate signal and feed back correction to the SET input to minimize the difference. Maybe some positive fdbk that way though. Must be some way.
Wait, I got it. Make the SET amp with partial cathode feedback (CFB). That way the SS amp is effectively supplying the plate voltage for the open loop SET amp (backward thru its OT), and the SET amp is effectively monitoring the final speaker output thru the CFB fdbk. Well, low feedback anyway.
Don
smoking-amp said:
Sounds like you're suggesting splitting the difference so zero signal has "negative" offset and FS has positive offset, in the same amount.
Or maybe it would be better to put all of the negative offset in at idle, where you have more headroom in flux density. As signal increases to FS, the DC decreases to zero. Many PP transformers are built to handle 5mA or more offset so it shouldn't be a problem. Hmm, I can probably prototype this in my hybrid amp. The idle current is 34 mA and FS is 37 mA. I was originally going to do it this fixed way but both kenpeter and VoltSecond suggested this LPF servo would be more trouble free to start with ad I'm glad for that but I still wonder about the transient response.
I'm also working on a bias regulator for PP outputs that has no drift, offset, or distortion current issues, but I can't see how to combine that idea with this thing yet...
Thanks!
Michael
Re: Michael
"Sounds like you're suggesting splitting the difference ..."
Yes.
Although, on thinking about this again, I think I may see an elegant solution (at least for the CCS'd LTP version). All one needs is an appropriate resistance added across the tail CCS to track the increased current needs with increasing signal (due to dist.).
This strikes me as being related to the WE harmonic equalizer (HE) scheme. Just put in a tail resistance that nulls the 3rd harmonic dist. So simple. Why didn't I think of that before.
The high Gm of the Mosfet will keep tail voltage variation small, so it will require a lower value tail resistance than for P-P tubes HE. Unfortunately, this will attenuate the AC complementary current too. So maybe use something more like a gyrator circuit (inductor emulation). Probably can do this with just modifying the CCS tail circuit. Or just use a real inductor for the tail.
Don
"Sounds like you're suggesting splitting the difference ..."
Yes.
Although, on thinking about this again, I think I may see an elegant solution (at least for the CCS'd LTP version). All one needs is an appropriate resistance added across the tail CCS to track the increased current needs with increasing signal (due to dist.).
This strikes me as being related to the WE harmonic equalizer (HE) scheme. Just put in a tail resistance that nulls the 3rd harmonic dist. So simple. Why didn't I think of that before.
The high Gm of the Mosfet will keep tail voltage variation small, so it will require a lower value tail resistance than for P-P tubes HE. Unfortunately, this will attenuate the AC complementary current too. So maybe use something more like a gyrator circuit (inductor emulation). Probably can do this with just modifying the CCS tail circuit. Or just use a real inductor for the tail.
Don
- Status
- Not open for further replies.