Hi all,
I started recently to study (and simulate using PSPICE) the "Penultimate Zen". Even if, after some tuning, at the simulator the amplifier sounds great, I would like to ask if nobody is worried by the large amount of global feedback applied.
I'm not religious about that, but I would like to hear some comments.
Please don't say: <I don't care about what Dr. Matti Otala can say about that, it's a sexy configuration and I don't care about TIM>.
Thanks
Teodoro Marinucci
I started recently to study (and simulate using PSPICE) the "Penultimate Zen". Even if, after some tuning, at the simulator the amplifier sounds great, I would like to ask if nobody is worried by the large amount of global feedback applied.
I'm not religious about that, but I would like to hear some comments.
Please don't say: <I don't care about what Dr. Matti Otala can say about that, it's a sexy configuration and I don't care about TIM>.
Thanks
Teodoro Marinucci
The Zen amp does not generate enough open loop gain
to have a lot of feedback - the 2S transconductance of the
power Mosfet compared to the 8 ohm load gives about 16X,
which is 24 dB. Typically this means about 10 dB of feedback,
which is quite low.
TIM is a description of a distortion mechanism in multi-stage
amplifiers, where the job of correcting output distortion tends
to overload the input stage. This input stage, being only an
impedance buffer, is pretty relaxed in its response.
to have a lot of feedback - the 2S transconductance of the
power Mosfet compared to the 8 ohm load gives about 16X,
which is 24 dB. Typically this means about 10 dB of feedback,
which is quite low.
TIM is a description of a distortion mechanism in multi-stage
amplifiers, where the job of correcting output distortion tends
to overload the input stage. This input stage, being only an
impedance buffer, is pretty relaxed in its response.
Speaking purely for myself, I don't regard 10dB of feedback in a solid state piece to be a 'large amount' of feedback. That's way down in tube territory. Most tube circuits use something on the order of 10-12dB NFB. Solid state amps typically use 30 to 40dB or more.
It always amuses me to hear solid state people rant about the "poor" specs exhibited by tube equipment. Every once in a while, I toss out the suggestion to try getting good specs out of a solid state piece with a mere 10dB of feedback. Yes, Nelson, John Curl, and a handful of others manage to get decent specs out of relatively modest amounts of feedback, but most solid state designers can't. Those who are True Believers in high amounts of negative feedback would be better off complaining about the lower gain available from tubes...which in turn limits the amount of negative feedback that can be applied. At least that complaint has some basis in reality.
Of course, there's no law that says you have to use feedback. If you want, feel free to reduce the amount of feedback in the Zen circuit(s).
I haven't had much opportunity to play with electronics recently, but I've gotten solid state designs down as far as ca. 7dB NFB with decent specs. I'd like to get to zero, but that will take better parts and more fiddling time.
Grey
It always amuses me to hear solid state people rant about the "poor" specs exhibited by tube equipment. Every once in a while, I toss out the suggestion to try getting good specs out of a solid state piece with a mere 10dB of feedback. Yes, Nelson, John Curl, and a handful of others manage to get decent specs out of relatively modest amounts of feedback, but most solid state designers can't. Those who are True Believers in high amounts of negative feedback would be better off complaining about the lower gain available from tubes...which in turn limits the amount of negative feedback that can be applied. At least that complaint has some basis in reality.
Of course, there's no law that says you have to use feedback. If you want, feel free to reduce the amount of feedback in the Zen circuit(s).
I haven't had much opportunity to play with electronics recently, but I've gotten solid state designs down as far as ca. 7dB NFB with decent specs. I'd like to get to zero, but that will take better parts and more fiddling time.
Grey
Grey got the point, for many people it's not worth the trouble to linearize every single stage to have descent open loop performance.
Just design a circuit with the highest gain you can with the devices you have, and then the huge amount of global negative feedback will correct all the non linearities... (or maybe not)
Nelson, your answear about the TIM in single stage amplifiers made me wonder something: can we really talk about global feedback in a single stage amplifier?
Isn't global feedback (feeding part of the inverted output back to the input) the same as local feedback (e.g. more source degeneration, or a lower stage gain) here?
Just design a circuit with the highest gain you can with the devices you have, and then the huge amount of global negative feedback will correct all the non linearities... (or maybe not)
Nelson, your answear about the TIM in single stage amplifiers made me wonder something: can we really talk about global feedback in a single stage amplifier?
Isn't global feedback (feeding part of the inverted output back to the input) the same as local feedback (e.g. more source degeneration, or a lower stage gain) here?
First: the assumtpions
Nelson,
I'm starting to describe what I did. At any moment, if you think that my questions, my assumptions, my methodology, are too stupid, say "stop" and I shall stop.
My problem is that, even if I like to build machines delivering good music, I like also to understand (as far as possible to me) how they work. My problem is that my (little) culture is mainly about tubes, so I have some difficulty to understand solid state.
First: my assumptions.
I wanted to build 20+20W, class A, hybrid amplifier.
For the tube part I wanted to use the Aikido (www.tubecad.com) line stage. I have already the PCB, the tubes, all the simulations that I understand.
I started thinking about the final stage. I took into consideration the final stage proposed therein: It's a sort of Moscode amplifier, with a 40V positive rail, 1.2A idle current.
It is for that reason that I have "designed" a regulated power supply and I have now a (nice toroidal) power transformer able to deliver 42V, 3A (AC): it seemed already to me that having a 126VA transformer to deliver only 20W was really a waste.
I'm starting to appreciate your topology, and I would like to put it in the place I have already reserved for the final stage.
The circuit I'm using is:
If you tell me "go on" I'll start describing the modifications I have made.
Thanks
Nelson,
I'm starting to describe what I did. At any moment, if you think that my questions, my assumptions, my methodology, are too stupid, say "stop" and I shall stop.
My problem is that, even if I like to build machines delivering good music, I like also to understand (as far as possible to me) how they work. My problem is that my (little) culture is mainly about tubes, so I have some difficulty to understand solid state.
First: my assumptions.
I wanted to build 20+20W, class A, hybrid amplifier.
For the tube part I wanted to use the Aikido (www.tubecad.com) line stage. I have already the PCB, the tubes, all the simulations that I understand.
I started thinking about the final stage. I took into consideration the final stage proposed therein: It's a sort of Moscode amplifier, with a 40V positive rail, 1.2A idle current.
It is for that reason that I have "designed" a regulated power supply and I have now a (nice toroidal) power transformer able to deliver 42V, 3A (AC): it seemed already to me that having a 126VA transformer to deliver only 20W was really a waste.
I'm starting to appreciate your topology, and I would like to put it in the place I have already reserved for the final stage.
The circuit I'm using is:
If you tell me "go on" I'll start describing the modifications I have made.
Thanks
To Bricolo ...
Bricolo,
You sholud always bear in mind that "tube people" are really obssessed by (any kind of) feedback.
There are people disliking cathode followers because of the local, degenerative, feedback.
There are people putting 2200uF electrolytic capacitors to bypass a 50Ohm self-bias resistor.
There are people that, since Otala said that too much global feedback is harmul, once they only hear the word, become anxious.
Me, I'm trying to understand ...
Thanks for any contribution.
Bricolo,
You sholud always bear in mind that "tube people" are really obssessed by (any kind of) feedback.
There are people disliking cathode followers because of the local, degenerative, feedback.
There are people putting 2200uF electrolytic capacitors to bypass a 50Ohm self-bias resistor.
There are people that, since Otala said that too much global feedback is harmul, once they only hear the word, become anxious.
Me, I'm trying to understand ...
Thanks for any contribution.
If you don't want any feedback, neither local nor global, you would end up with an unusable system with infinite gain. Because of vacum/solid state technical limitations, you always have a resistance that causes a degenerating local feedback, and capacitances creating miller paths
If you don't want any feedback, you should stay far away from any linear electronical device.
And without using global and degenerating local feedback, you are just using your components with the highest DC gain they can give you. That means close to an integrator: high DC gain and low -3dB frequency. Isn't this told to have poor audio performance?
If you don't want any feedback, you should stay far away from any linear electronical device.
And without using global and degenerating local feedback, you are just using your components with the highest DC gain they can give you. That means close to an integrator: high DC gain and low -3dB frequency. Isn't this told to have poor audio performance?
Greatings
Mystery of Amps design, NFB and musicality ... the never ending discussion !
Speaking about my own recent experience with my modest multi amplication system starting from classic Class AB amplifiers with high global NFB , to the ZEN V1 amps (with some upgrade taken on newer versions in fact, and I think with even lower NFB),
there is no doubt that the little big ZEN is far far far more "musical" (ability to retranscript sound naturally, as there are in real life) there's probably better, but the change is hudge for my ears. I simply re-discovered all my CDs and I can now listen at decent levels for hours.
The better load matching of the ZEN with my midrange and tweeter may be also probably in favor of this judged better "musicality" .
I like this paper from D.Cheever that actually made some sense for me after my above experience.
http://w3.mit.edu/cheever/www/cheever_thesis.pdf
one of the keys would be to admit that our ear is not a perfect sensor ... and no simulator did find that.
I wonder if someone had the time to re-experiment some of the exps that Cheever did, trying different amps design families measuring what he defined as "T.A.D" and making blind listening testing. I'd love to learn more about these things.
Mystery of Amps design, NFB and musicality ... the never ending discussion !
Speaking about my own recent experience with my modest multi amplication system starting from classic Class AB amplifiers with high global NFB , to the ZEN V1 amps (with some upgrade taken on newer versions in fact, and I think with even lower NFB),
there is no doubt that the little big ZEN is far far far more "musical" (ability to retranscript sound naturally, as there are in real life) there's probably better, but the change is hudge for my ears. I simply re-discovered all my CDs and I can now listen at decent levels for hours.
The better load matching of the ZEN with my midrange and tweeter may be also probably in favor of this judged better "musicality" .
I like this paper from D.Cheever that actually made some sense for me after my above experience.
http://w3.mit.edu/cheever/www/cheever_thesis.pdf
one of the keys would be to admit that our ear is not a perfect sensor ... and no simulator did find that.
I wonder if someone had the time to re-experiment some of the exps that Cheever did, trying different amps design families measuring what he defined as "T.A.D" and making blind listening testing. I'd love to learn more about these things.
N-Well said:
C'est vous, mon pot Clouseau ?
If you do a search you may find a number of gents overhere who seem to find Dan Cheever's piece de resistance a wee bit off.
Me, i stopped taking thesis seriously after reading one on submarine chemical carriers in the port of Rotterdam.
Attachments
When I started this thread I was curious about the general characteristcs and behaviour of the "penultimate Zen".
GNF and IMD are only one of the many points of interest.
As an xample, it seems to me that having 10uF for C2 seems too much: at the simulation I get a pronounced frequency dip just below 10Hz.
More: I would like to understand what are the parameters that control the behaviour of the circuit. It is quite clear to me that R13 controls the idle current, and that R4 controls the symmetry of the output signal.
But ... how these parameters are interrelated ? Are there other parameters playing similar role ? What if I want to use a different active device ? What if my lodspeakers are 4Ohms ? What if I want less idle current ?
Thanks
P.S. Thanks for the thesis.
GNF and IMD are only one of the many points of interest.
As an xample, it seems to me that having 10uF for C2 seems too much: at the simulation I get a pronounced frequency dip just below 10Hz.
More: I would like to understand what are the parameters that control the behaviour of the circuit. It is quite clear to me that R13 controls the idle current, and that R4 controls the symmetry of the output signal.
But ... how these parameters are interrelated ? Are there other parameters playing similar role ? What if I want to use a different active device ? What if my lodspeakers are 4Ohms ? What if I want less idle current ?
Thanks
P.S. Thanks for the thesis.
Yesseuh ! Clouseau izz backe !
The regreted Peter Sellers did a great job and made French accent quite ..euh well euh ... popular let's say ... wof wof wof ...
My understanding of his thesis is that NFB combined with non linear gain stages will inevitably produce a train of high order hamonics (actually I did not catch all the maths here ...) , and if the resulting harmonic distibution is not similar to our own ear Harmonic distorsion spectrum (our ear generates its own distorsion spectrum when receiving sounds, yes..) , we can percieve it as "bad" sound (to use a short cut, i could not say anything else anyway ...)
Indeed looking at some of the charts in the paper, increasing the NFB is decreasing the higher harmonics, but the trouble would be that even such low levels may remain percievable and would be enough to denaturate sound of musical instruments for example.
Some so called "numeric" amplifers (that I never eard myself) are claimed to be very musical,and I was wondering if this was because they could be extremely linear ( a proper pulse modulation would allow it) and may require very small amount of feedback or none may be ? (never studied schemas/principle myself I must say)
I am not really competent to say if Cheever thesis was build with enough rigor. To clear this topic (apparently not closed), years of research would be needed (and who's going to found this these days ? )
As he wrote himself a few time in his paper, he has made some hypothesis at some point and he ended up his thesis with some path to explore furthermore.
so we have to continue build, experiment and listen, and in the mean time, please our ears by the way..
The regreted Peter Sellers did a great job and made French accent quite ..euh well euh ... popular let's say ... wof wof wof ...
My understanding of his thesis is that NFB combined with non linear gain stages will inevitably produce a train of high order hamonics (actually I did not catch all the maths here ...) , and if the resulting harmonic distibution is not similar to our own ear Harmonic distorsion spectrum (our ear generates its own distorsion spectrum when receiving sounds, yes..) , we can percieve it as "bad" sound (to use a short cut, i could not say anything else anyway ...)
Indeed looking at some of the charts in the paper, increasing the NFB is decreasing the higher harmonics, but the trouble would be that even such low levels may remain percievable and would be enough to denaturate sound of musical instruments for example.
Some so called "numeric" amplifers (that I never eard myself) are claimed to be very musical,and I was wondering if this was because they could be extremely linear ( a proper pulse modulation would allow it) and may require very small amount of feedback or none may be ? (never studied schemas/principle myself I must say)
I am not really competent to say if Cheever thesis was build with enough rigor. To clear this topic (apparently not closed), years of research would be needed (and who's going to found this these days ? )
As he wrote himself a few time in his paper, he has made some hypothesis at some point and he ended up his thesis with some path to explore furthermore.
so we have to continue build, experiment and listen, and in the mean time, please our ears by the way..
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.