Reply
 
Thread Tools Display Modes
  #1   Report Post  
Posted to rec.audio.tubes
patrick-turner patrick-turner is offline
external usenet poster
 
Posts: 119
Default Hybrid amp, EL34 drive to compound complementary mosfet output

I've posted up an additional schematic for a Hybrid amp for 39Watts class A which has EL34 driving complementary N&P source followers helped along by common source P & N mosfets which I have not yet ried, but which looks good in theory.
Those wanting to simulate the schematic, or actually try it out by building it may like to post criticism here.
See Fig 16 at the bottom of the page at http://www.turneraudio.com.au/solids...ube-input.html

Patrick Turner.
  #2   Report Post  
Posted to rec.audio.tubes
Alex Pogossov Alex Pogossov is offline
external usenet poster
 
Posts: 95
Default Hybrid amp, EL34 drive to compound complementary mosfet output


"patrick-turner" wrote in message ...
I've posted up an additional schematic for a Hybrid amp for 39Watts class A which has EL34 driving complementary N&P source followers helped along by common source P & N mosfets which I have not yet ried, but which looks good in theory.
Those wanting to simulate the schematic, or actually try it out by building it may like to post criticism here.
See Fig 16 at the bottom of the page at http://www.turneraudio.com.au/solids...ube-input.html

Patrick Turner.


Criticism? Critique? I do not think these are the correct terms. How can a stamp collector criticise a coin collector? Each one would think that his hobby is great, but the hobby of the other person is utter waste of time. We are free willed people and each of us is fully entitled to go insane in a particular and unique way. Therefore we should not judge or criticise which is an infringement on the free agency. Instead of "criticisig" I will make some comments.

1) The whole concept of using a tube to drive MOSFETs is ridiculous becaude it kills the greatest advantage of a solid-state design -- DC coupling and deep feedback. In your circuit you have two (!) differentiating capacitors in the feedback loop: one to the screen, the other to the MOSFET gates. Therefore low-F transients will be horrible. DC offset on the output is not compensated and needs trimming. What happens to the output when the tube is warming up or hot pulled out of socket? An enormous cone ripping pop!
2) This monster uses heavy iron as a 40H choke and other variants even interstage paraphase transformers. Who in the right mind would use solid-state compact devices combined with this monstrosity? Why not make a proper well desidned DC stabilised solid-state amplifier? In class A everything will sound terrific. Just use large heatsinks and more fans.

Now concerning your "quartet" of MOSFETozzi.

A MOSFET has a unique feature -- a quadratic transfer function. Consider a MOSFET with zero threshold and with I = v^2. At 1Amp bias it will have S=2A/V. Consider the MOSFET is biased to 1 Amp by 1 Volt of gate bias.

In the top leg of the push-pull stage output current Ip = (1 + x)^2 = 1 + 2x + x^2, where x is AC component of drive voltage.
For the bottom leg In = (1 - x)^2 = 1 - 2x + x^2.

After combining (subtracting) currents Ip - In the load current will be Iload = 4x.
In other words, in class A, second harmonics are TOTALLY cancelled, and the total effective transconductance is 4A/V. This is a uniwue feature of a MOSFET -- TOTAL cancellation of distortion. Neither tubes with their law of 3/2 not BJTs with their exponential law can do that miracle. This is why several generations of audio designers have been phooled into thinking that MOSFETs are superior to BJTs.

Now consider your approach of taking a current from one MOSFET, converting it into voltage (by 2.2R resistor) and feeding as a drive to the gate of the next MOSFET. Here you violated your own rule of thumb: "Never use a distorted signal from one power device to drive another power device". What you virtually suggest is to use a distorted drain current of the first MOSFET to drive the second one. You add quadratic distortion on top of another quadratic distortion. Now the transfer function will be of the fourth order.

Some maths:
Let us suppose we use the above ideal MOSFETs and using a resistor of 1Ohm, derive drive voltage for the next MOSFET.

Then for the top leg: Ip = ( 1+ x^2)^2 = 1 + 4x + 6x^2 + 4x^3 + x^4. We can see that now we have all the harmonics up to the fourth order. (Note: in this simplified case the transconductance gets doubled. In your circuit you use a AC divider to bring it down for equal current sharing.)

Similarly for the bottom leg: In = 1 - 4x + 6x^2 - 4x^3 + x^4.

Now when you combine (subtract) both legs, the load current Iload = 8x + 8x^3. The additional members of the "qartet" will nicely produce 50% of the third harmonic (at full power)! Note also that this 50% refers to the current. Since the stage works as follower, voltage distortion is reduced by the internal NFB -- the closer the gain to unity, the less is the voltage distortion.

I thought you intended to use a quartet to equalise differences between P and N MOSFETs which though called "complementary" are not exactly so. If I understand correctly you attemp was to make top and bottom legs identical by using both genders in each leg.

Well, there exists a proper solution for that -- super-complementary follower:
http://www.valveradio.net/en/home/au...ary-stage.html

However it is not popular because even if top and bottom side are different say by 20% in the transconductance (because of inferiority of P-MOSFETs), the residual asymmetry will be almost completely defeated by the deep global NFB in a solid state amp. In your case where the NFB is perhaps not more than 20dB, symmetry becomes far more important, even in class A.

Perhaps the concept described in the above link will be of more benefit to your case than to any one's else.

Regards,
Alex


  #3   Report Post  
Posted to rec.audio.tubes
Alex Pogossov Alex Pogossov is offline
external usenet poster
 
Posts: 95
Default Hybrid amp, EL34 drive to compound complementary mosfet output

Looks like the "super-complementary" stage had some use in Bryston amplifiers:
http://www.bryston.com/PDF/Schematic...SCHEMATICS.pdf
(Sorry for posting transistorized links on a tube forum.)
"Alex Pogossov" wrote in message ...

"patrick-turner" wrote in message ...
I've posted up an additional schematic for a Hybrid amp for 39Watts class A which has EL34 driving complementary N&P source followers helped along by common source P & N mosfets which I have not yet ried, but which looks good in theory.
Those wanting to simulate the schematic, or actually try it out by building it may like to post criticism here.
See Fig 16 at the bottom of the page at http://www.turneraudio.com.au/solids...ube-input.html

Patrick Turner.


Criticism? Critique? I do not think these are the correct terms. How can a stamp collector criticise a coin collector? Each one would think that his hobby is great, but the hobby of the other person is utter waste of time. We are free willed people and each of us is fully entitled to go insane in a particular and unique way. Therefore we should not judge or criticise which is an infringement on the free agency. Instead of "criticisig" I will make some comments.

1) The whole concept of using a tube to drive MOSFETs is ridiculous becaude it kills the greatest advantage of a solid-state design -- DC coupling and deep feedback. In your circuit you have two (!) differentiating capacitors in the feedback loop: one to the screen, the other to the MOSFET gates. Therefore low-F transients will be horrible. DC offset on the output is not compensated and needs trimming. What happens to the output when the tube is warming up or hot pulled out of socket? An enormous cone ripping pop!
2) This monster uses heavy iron as a 40H choke and other variants even interstage paraphase transformers. Who in the right mind would use solid-state compact devices combined with this monstrosity? Why not make a proper well desidned DC stabilised solid-state amplifier? In class A everything will sound terrific. Just use large heatsinks and more fans.

Now concerning your "quartet" of MOSFETozzi.

A MOSFET has a unique feature -- a quadratic transfer function. Consider a MOSFET with zero threshold and with I = v^2. At 1Amp bias it will have S=2A/V. Consider the MOSFET is biased to 1 Amp by 1 Volt of gate bias.

In the top leg of the push-pull stage output current Ip = (1 + x)^2 = 1 + 2x + x^2, where x is AC component of drive voltage.
For the bottom leg In = (1 - x)^2 = 1 - 2x + x^2.

After combining (subtracting) currents Ip - In the load current will be Iload = 4x.
In other words, in class A, second harmonics are TOTALLY cancelled, and the total effective transconductance is 4A/V. This is a uniwue feature of a MOSFET -- TOTAL cancellation of distortion. Neither tubes with their law of 3/2 not BJTs with their exponential law can do that miracle. This is why several generations of audio designers have been phooled into thinking that MOSFETs are superior to BJTs.

Now consider your approach of taking a current from one MOSFET, converting it into voltage (by 2.2R resistor) and feeding as a drive to the gate of the next MOSFET. Here you violated your own rule of thumb: "Never use a distorted signal from one power device to drive another power device". What you virtually suggest is to use a distorted drain current of the first MOSFET to drive the second one. You add quadratic distortion on top of another quadratic distortion. Now the transfer function will be of the fourth order.

Some maths:
Let us suppose we use the above ideal MOSFETs and using a resistor of 1Ohm, derive drive voltage for the next MOSFET.

Then for the top leg: Ip = ( 1+ x^2)^2 = 1 + 4x + 6x^2 + 4x^3 + x^4. We can see that now we have all the harmonics up to the fourth order. (Note: in this simplified case the transconductance gets doubled. In your circuit you use a AC divider to bring it down for equal current sharing.)

Similarly for the bottom leg: In = 1 - 4x + 6x^2 - 4x^3 + x^4.

Now when you combine (subtract) both legs, the load current Iload = 8x + 8x^3. The additional members of the "qartet" will nicely produce 50% of the third harmonic (at full power)! Note also that this 50% refers to the current. Since the stage works as follower, voltage distortion is reduced by the internal NFB -- the closer the gain to unity, the less is the voltage distortion.

I thought you intended to use a quartet to equalise differences between P and N MOSFETs which though called "complementary" are not exactly so. If I understand correctly you attemp was to make top and bottom legs identical by using both genders in each leg.

Well, there exists a proper solution for that -- super-complementary follower:
http://www.valveradio.net/en/home/au...ary-stage.html

However it is not popular because even if top and bottom side are different say by 20% in the transconductance (because of inferiority of P-MOSFETs), the residual asymmetry will be almost completely defeated by the deep global NFB in a solid state amp. In your case where the NFB is perhaps not more than 20dB, symmetry becomes far more important, even in class A.

Perhaps the concept described in the above link will be of more benefit to your case than to any one's else.

Regards,
Alex


  #4   Report Post  
Posted to rec.audio.tubes
patrick-turner patrick-turner is offline
external usenet poster
 
Posts: 119
Default Hybrid amp, EL34 drive to compound complementary mosfet output

Alex mentioned....

"patrick-turner" wrote in message ...
- show quoted text -
I've posted up an additional schematic for a Hybrid amp for 39Watts class A which has EL34 driving complementary N&P source followers helped along by common source P & N mosfets which I have not yet ried, but which looks good in theory.
Those wanting to simulate the schematic, or actually try it out by building it may like to post criticism here.
See Fig 16 at the bottom of the page at http://www.turneraudio.com.au/solids...ube-input.html

Patrick Turner.

Criticism? Critique? I do not think these are the correct terms. How can a stamp collector criticise a coin collector? Each one would think that his hobby is great, but the hobby of the other person is utter waste of time.

No, your hobby isn't a waste of time.

We are free willed people and each of us is fully entitled to go insane in a particular and unique way. Therefore we should not judge or criticise which is an infringement on the free agency. Instead of "criticisig" I will make some comments.

***

1) The whole concept of using a tube to drive MOSFETs is ridiculous becaude it kills the greatest advantage of a solid-state design -- DC coupling and deep feedback.

But in fact, tube amps can have several C&R and one R&L couplings all preventing DC coupling and preventing deep FB yet they have fabulous sounding bass and they do music so well that SS is out-classed and seen as inferior by many ppl.

In your circuit you have two (!) differentiating capacitors in the feedback loop: one to the screen, the other to the MOSFET gates. Therefore low-F transients will be horrible.

LF transients are fine, no problem...
Try building and testing something shown at my website, and then you'll know.

DC offset on the output is not compensated and needs trimming. What happens to the output when the tube is warming up or hot pulled out of socket? An enormous cone ripping pop!

Tube warm up does not cause any cone ripping pop. Where did you get such silly ideas?

The warm up gives no pop at all and nobody pulls hot tubes from sockets, and operating bothers are no worse that any fully tubed amps.

2) This monster uses heavy iron as a 40H choke and other variants even interstage paraphase transformers. Who in the right mind would use solid-state compact devices combined with this monstrosity?

You are always free to use a CCS feed to load the EL34 or EL84 anodes which drive mosfets. But a choke is just fine. The use of an IST is also just fine and one can be made easily with wide bandwidth because the Vg-s voltages are so low.

Why not make a proper well desidned DC stabilised solid-state amplifier? In class A everything will sound terrific. Just use large heatsinks and more fans.

I've already done all that, but fans surely don't belong in lounge rooms because they make noise. Tubes and mosfets are a natutally good combination, and especially with no DC coupling and class A. DC coupling and fully bjt circuits suit sharehlders mostly, and possibly sound well, but things like Sugden class A amps do not have any appeal to me - especially after repairing so many.

Now concerning your "quartet" of MOSFETozzi.

A MOSFET has a unique feature -- a quadratic transfer function. Consider a MOSFET with zero threshold and with I = v^2. At 1Amp bias it will have S=2A/V. Consider the MOSFET is biased to 1 Amp by 1 Volt of gate bias.

In the top leg of the push-pull stage output current Ip = (1 + x)^2 = 1 + 2x + x^2, where x is AC component of drive voltage.
For the bottom leg In = (1 - x)^2 = 1 - 2x + x^2.

After combining (subtracting) currents Ip - In the load current will be Iload = 4x.
In other words, in class A, second harmonics are TOTALLY cancelled, and the total effective transconductance is 4A/V. This is a uniwue feature of a MOSFET -- TOTAL cancellation of distortion. Neither tubes with their law of 3/2 not BJTs with their exponential law can do that miracle. This is why several generations of audio designers have been phooled into thinking that MOSFETs are superior to BJTs.

I'm not phooled, but I do prefer mosfets. Susan Parker has a website which uses class A mosfets of same N channel with OPT and gets extremely low THD with no global NFB. Very easy to drive, and bood sound. I've done the similar thing with 4 mosfets in common source mode, and overlapping FB loops to reduce THD to low enough, and Rout to low enough, with OPT, C&R couling, no problems.

Now consider your approach of taking a current from one MOSFET, converting it into voltage (by 2.2R resistor) and feeding as a drive to the gate of the next MOSFET. Here you violated your own rule of thumb: "Never use a distorted signal from one power device to drive another power device". What you virtually suggest is to use a distorted drain current of the first MOSFET to drive the second one. You add quadratic distortion on top of another quadratic distortion. Now the transfer function will be of the fourth order.

Yes, but the source follower output is subject to follower FB. I'll have to try the circuit out before I believe a single word you say.

Some maths:
Let us suppose we use the above ideal MOSFETs and using a resistor of 1Ohm, derive drive voltage for the next MOSFET.

Then for the top leg: Ip = ( 1+ x^2)^2 = 1 + 4x + 6x^2 + 4x^3 + x^4. We can see that now we have all the harmonics up to the fourth order. (Note: in this simplified case the transconductance gets doubled. In your circuit you use a AC divider to bring it down for equal current sharing.)

Similarly for the bottom leg: In = 1 - 4x + 6x^2 - 4x^3 + x^4.

Now when you combine (subtract) both legs, the load current Iload = 8x + 8x^3. The additional members of the "qartet" will nicely produce 50% of the third harmonic (at full power)! Note also that this 50% refers to the current. Since the stage works as follower, voltage distortion is reduced by the internal NFB -- the closer the gain to unity, the less is the voltage distortion.

Your previous paragraphs have totally bamboozled all who may have read them, including me, and I propose to build the circuit to find out about all properties of my schematics. That's how I learn, rather than forecasting bull**** that may or my not happen in the real world.


I thought you intended to use a quartet to equalise differences between P and N MOSFETs which though called "complementary" are not exactly so. If I understand correctly you attemp was to make top and bottom legs identical by using both genders in each leg.

My guess is that if N&P mosfets are used in top and bottom halves of circuit, they become more like two equal halves. Just using P on top and N on bottom cannot give best possible low THD because of differnces in P and N device characteristics. If you had say an EL34 and 6L6 used in a typical tube amp OP stage then you get considerable 2H because the tubes have different gm and Ra.
But say you have a PP tube amp with 4 tubes, then you could use EL34 + 6L6 on each side, and as long as biasing was adjusted right, it would give low 2H like 4 x EL34, or 4 x 6L6.

Well, there exists a proper solution for that -- super-complementary follower:
http://www.valveradio.net/en/home/au...ary-stage.html

I'll consider that later, but you have it all common source, and with tube drive, I'd want it source follower, so I'll have to play around with the idea.


However it is not popular because even if top and bottom side are different say by 20% in the transconductance (because of inferiority of P-MOSFETs), the residual asymmetry will be almost completely defeated by the deep global NFB in a solid state amp. In your case where the NFB is perhaps not more than 20dB, symmetry becomes far more important, even in class A.

No. The simplest version of my hybrid using P and N source followers only and 1 x EL34 gives excellently low THD. Using higher mosfet supply rails of say +/-22Vdc, and having 3 x N and 3 x P source followers will give even better performance. There is no need for any global NFB of any other kind besides the screen FB I have. But for those wanting more FB, just use an additional input tube, add the FB loop as usual. Mosfets can be driven with EL34 in triode without the source follower mosfets included in the anode to screen connection. The GNFB done conventionally will work just fine.

Perhaps the concept described in the above link will be of more benefit to your case than to any one's else.

Definately perhaps only, because unless anyone builds examples and tests them, the real truth always remains unclear.

Patrick Turner.

Regards,
Alex
  #5   Report Post  
Posted to rec.audio.tubes
patrick-turner patrick-turner is offline
external usenet poster
 
Posts: 119
Default Hybrid amp, EL34 drive to compound complementary mosfet output

Alex mentioned...

Looks like the "super-complementary" stage had some use in Bryston amplifiers:
http://www.bryston.com/PDF/Schematic...SCHEMATICS.pdf
(Sorry for posting transistorized links on a tube forum.)

Patrick replies now with ......
The link does does not take me to a particular schematic, but yes, I recall Ben Duncan's 1996 book I read where the use of N and P devices on each side of PP circuit are used, in order to give automatic device characteristic matching and hence less THD etc.

In the absense of much stuff about tubes, discussion of SS items should be tolerated when considering mosfets driven by vacuum tubes. Alex may surely be peeved and dismayed, and aghast at such absurdly ridiculous techniques I propose which I think do have a few saving graces IMHO, even when a dreadfully horrible IST is inserted offendingly fair smack in the middle of the signal path.
But just because I recommend the use of an IST, its actually OK because there is already room for an IST in a hybrid amp where an OPT isn't used. But the IST should be a slightly wider better BW device compared to something one may purchase from Hammond. As I mentioned, the IST has only to handle much lower voltages than when driving output tubes because the Vg-s signal of mosfets is rarely more than 2Vrms.

I had a look at the link for super-complementary follower:
http://www.valveradio.net/en/home/au...ary-stage.html

Its not too bad, and had two bjts to give +&- drive voltages and all that's OK.

But one may use a simpler method where the mosfets are arranged so two are in source follower mode and two are in common source mode.
Say you have +15Vrms powering 5 ohms at output.
The pair of Vg signals into N&P source fols are the same phase, same amplitude although DC bias is different. So Vg would be about +16.3Vrms, with Rs = 0.47r.
Vg signal to N&P common source mosfets is also the same phase and amplitude and will be about -1.3Vrms.

The -1.3Vrms comes from an anode RLa, say 2k2 to B+, and the +16.3Vrms comes from the cathode and an equal sized RLk 2k2 to a bootstrapped point connected to Vo.
Input to the concertina phase inverter is about 17Vrms, and this triode should be an EL84 in triode with low Ra, so that any distortion voltage between Vo and source follower gates is nearly all transfered to gates of common source mosfets. The only disadvantage of the scheme is that one has to use the extra triode between an input amp which needs to make slightly more than Vo. However, it should work OK and and HF gain can be adjusted with phase/gain reduction RC network applied to stop HF oscillations if thy can at all be provoked with a capacitance load of any value at Vo.

Next time I get a clear day to spend in the workshop without being driven mad by irate customers wanting their gear fixed last month, I'll try out a few of these ideas along with many more that keep coming to mind.

My own view is that the concertina phase inverter floating on Vo isn't really needed so I'll try my ideas first which are elegantly so simple.

BTW, just to illustrate why Alex should not get cranky about LF transients and his intolerance of RC and LR coupling when using SS devices, ppl should go to http://www.turneraudio.com.au/solids...ono-mosfet.htm

Here we have RC coupling, and an OPT used with mosfets. The sound is indistinguishable from a good class A amp using 4 x EL34 per channel. There are NO BAD HABITS od bangs, pops, wheezes, farts or other BS when the amp is turned on or off. Bass is extraordinarily GOOD.

I have plans to install new PTs with E&I lams because the toroidals from Jaycar are too noisy. Then I can also include a winding for B+ and for tube input stages instead of the simple existing 3 bjts. The OP stage will be source follower mode and existing OPT BW will increase somewhat from the existing 200kHz+, and I expect excellent sound, and I'll have what will be a more interesting pair of amps with lower THD/IMD than at present.
Patrick Turner.

Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is On
HTML code is Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
MOSFET output stage RichD Tech 427 October 15th 08 06:10 PM
EL34 output tube balancing? combover Vacuum Tubes 8 October 11th 06 12:27 AM
A hybrid MOSFET OTL Sander deWaal Vacuum Tubes 30 February 20th 04 09:50 AM
A hybrid MOSFET OTL Sander deWaal Audio Opinions 4 February 14th 04 10:25 PM
FS: Smart Devices 2x150VT Tube/MOSFET hybrid amplifier Chris Groppi Marketplace 0 February 5th 04 10:37 PM


All times are GMT +1. The time now is 02:10 PM.

Powered by: vBulletin
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 AudioBanter.com.
The comments are property of their posters.
 

About Us

"It's about Audio and hi-fi"