Hi
Hi Marc,
I have seen your suplifier circuit.
Its a good job done with utmost sincerity in terms of bridge tied load comp. balanced class-A amp.
Aksa [HUGH]has said that the dual differential cancels even order harmonics which is true, but the left out ones are odd harmonics which in my opinion are more pleasing to listening in concern with music than even order harmonics.
secondly the additional emitter follower stage is not neccessary in your circuit as stated by AKSA[HUGH], because your amp is a low power version, so theres no need of more current gain.
You have stated somewhere on your website about the disadvantages of mosfets in terms of high gate drive voltages which is true for your kind of circuit, but regarding output impedances the lateral mosfets have higher RDS in comparision with switching mosfets.
If you somehow use switching mosfets in your output stage then you would certainly gain sonic superiorty but you will need additional gate drive requirement also.
hope this illustrates my views regarding your amp.
regards,
kanwar
Marc Vi. said:
Hi Marc,
I have seen your suplifier circuit.
Its a good job done with utmost sincerity in terms of bridge tied load comp. balanced class-A amp.
Aksa [HUGH]has said that the dual differential cancels even order harmonics which is true, but the left out ones are odd harmonics which in my opinion are more pleasing to listening in concern with music than even order harmonics.
secondly the additional emitter follower stage is not neccessary in your circuit as stated by AKSA[HUGH], because your amp is a low power version, so theres no need of more current gain.
You have stated somewhere on your website about the disadvantages of mosfets in terms of high gate drive voltages which is true for your kind of circuit, but regarding output impedances the lateral mosfets have higher RDS in comparision with switching mosfets.
If you somehow use switching mosfets in your output stage then you would certainly gain sonic superiorty but you will need additional gate drive requirement also.
hope this illustrates my views regarding your amp.
regards,
kanwar
Hi Hugh,
Thanks for your comments.
1. My aim is to design an open loop amplifier with a distortion as low as possible. Therefore I use the extra emitter follower. Indeed this gives an additional phase shift but because the amplifier has no global NFB and therefore needs no lag compensation, this is of minor interest. This is a good issue for simulating whether the distortion in an emitter follower + CFP output will be lower than in only a CFP output. I will do this simulation and will report the results.
2. I know that a CFP is more sensitive for oscillation in a class (A)B output and even more with global NFB. Therefore it was my conscious choice to combine class A, no global NFB and a CFP output.
3. It is true that a fully complementary input stage cancels out a little bit the even harmonics and that the higher odd harmonics may influence the colour of the sound in a negative way. (The even harmonics are not completely cancelled out, just a little bit. The even harmonics only disappear fully when there is no emitter degeneration and the input stage is 100% complementary balanced.) As you have seen The Ultimate Suplifier is a BTL amplifier consisting of a left and a right amplifier part. Following a suggestion of Steven, I am now working on a design in which the left (positive) amplifier part has a NPN differential input stage and the right (negative) has a PNP differential input stage. If you go single ended out of those differential input stages, then the even harmonics are not cancelled out while you use the advantage of a differential input pair, namely a low total THD.
Hi Kanwar,
Thanks also for your comments.
Odd harmonics are not more pleasuring for listening, as you can read in literature. Type in google “even odd distortion listening” or go to http://www.pmillett.addr.com/file downloadss/ThesoundofDistortion.pdf
I am no fan of FETs, maybe because I have no experience with FETs
Marc.
Thanks for your comments.
1. My aim is to design an open loop amplifier with a distortion as low as possible. Therefore I use the extra emitter follower. Indeed this gives an additional phase shift but because the amplifier has no global NFB and therefore needs no lag compensation, this is of minor interest. This is a good issue for simulating whether the distortion in an emitter follower + CFP output will be lower than in only a CFP output. I will do this simulation and will report the results.
2. I know that a CFP is more sensitive for oscillation in a class (A)B output and even more with global NFB. Therefore it was my conscious choice to combine class A, no global NFB and a CFP output.
3. It is true that a fully complementary input stage cancels out a little bit the even harmonics and that the higher odd harmonics may influence the colour of the sound in a negative way. (The even harmonics are not completely cancelled out, just a little bit. The even harmonics only disappear fully when there is no emitter degeneration and the input stage is 100% complementary balanced.) As you have seen The Ultimate Suplifier is a BTL amplifier consisting of a left and a right amplifier part. Following a suggestion of Steven, I am now working on a design in which the left (positive) amplifier part has a NPN differential input stage and the right (negative) has a PNP differential input stage. If you go single ended out of those differential input stages, then the even harmonics are not cancelled out while you use the advantage of a differential input pair, namely a low total THD.
Hi Kanwar,
Thanks also for your comments.
Odd harmonics are not more pleasuring for listening, as you can read in literature. Type in google “even odd distortion listening” or go to http://www.pmillett.addr.com/file downloadss/ThesoundofDistortion.pdf
I am no fan of FETs, maybe because I have no experience with FETs
Marc.
Marc Vi. said:
Reading the reference, the author makes reference to some
people prefering 2nd harmonic, some preferring 3rd, but
higher order harmonics as being more undesirable.
It appears to me that the proportion of audiophiles who prefer
the 3rd harmonic character (push pull) vs 2nd harmonic (single
ended) are comparable, with a significant segment who go
either way.
With a real simple no-feedback circuit, the differences are very
apparent, and it's an interesting experiment comparing them.
The 2nd harmonic amps are indeed warmer and have a more
romantic quality, but they come apart faster with complex
material. They seem to excel at fairly simple music - vocals, string
quartets, Jazz, and so on.
The 3rd harmonic amps have more contrast and dynamic range,
along with a snappier bottom end. With something more complex
like an orchestral piece, they offer more detail and clarity.
My opinion, of course, and I don't have a general preference
anymore, although I do like to create my 3rd harmonic amps with
balanced single-ended circuits.
This might be a little far afield of the original topic
but what the heck, Nelsons last post reminded me
of it.
I investigated how closely harmonics of a fundamental
are in tune with their closest note on a piano a while ago.
Here is an excel spreadsheet that shows the
relationship between harmonics and their nearest note in the
musical scale. You might have thought they were identical as
I once did, but they are not.
Each row represents a note on a piano. It shows the note
frequency, the harmonic, the nearest note, and how close in
frequency they are.
One section on each page is for the 13 notes starting at C1 and
ending at C2. The other section is for the 13 notes starting at C3
and ending at C4.
C1 refers to the note "C" in the lowest octave on
a piano. C3 refers to the note "C" in the third octave.
Each tab represents an investigation into that particular
harmonic. So a tab for the 2nd harmonic, 3rd, etc.
You can see that higher order odd harmonics go out of
tune. As you go even higher, the even harmonics as well
go out of tune.
This might equate to listening to an orchestra where
the harmonics of the lower notes are not all in tune
with the fundamentals being played by instruments
playing at higher frequencies.
but what the heck
, Nelsons last post reminded meof it.
I investigated how closely harmonics of a fundamental
are in tune with their closest note on a piano a while ago.
Here is an excel spreadsheet that shows the
relationship between harmonics and their nearest note in the
musical scale. You might have thought they were identical as
I once did, but they are not.
Each row represents a note on a piano. It shows the note
frequency, the harmonic, the nearest note, and how close in
frequency they are.
One section on each page is for the 13 notes starting at C1 and
ending at C2. The other section is for the 13 notes starting at C3
and ending at C4.
C1 refers to the note "C" in the lowest octave on
a piano. C3 refers to the note "C" in the third octave.
Each tab represents an investigation into that particular
harmonic. So a tab for the 2nd harmonic, 3rd, etc.
You can see that higher order odd harmonics go out of
tune. As you go even higher, the even harmonics as well
go out of tune.
This might equate to listening to an orchestra where
the harmonics of the lower notes are not all in tune
with the fundamentals being played by instruments
playing at higher frequencies.
I have always had a goal of minimizing THD in my designs according to the error budget of the application. And to do so I use all the tools available - 2 pole compensation, local and nested feedback and global feedback. The 20KHz THD is the test and <0.002% is the goal.
Nested feedback output stages can even make MOSFETs sound great.
Nested feedback output stages can even make MOSFETs sound great.
Harmonics and harmonic series
Weighing in with my first post in this forum, and off topic too....
Addressing mfc's comments on the harmonic series and musical instruments. I have seen this mentioned many times here (I'm slowly working my way through the archive.) The issue is much more complex, and annoying, than you think.
From a straight theoretical point of view it is actually impossible to make a musical instrument that is perfectly consonant. Even a simple fretless instrument has internal issues that prevent this. The piano however is the instrument that displays just about every problem known. Being a essentially a fretted instrument it can only be tuned harmonically in one key. To avoid this most people believe it is tuned to the equal tempered scale. Which essentially makes each interval sound equally bad. Actually pianos are not tuned to equal temperament, but are tuned to a plethora of different tunings - ones which balance different intervals in different keys, to what is hoped will be a generally better result. Well known tunings include Werkmeister III, Kimberger-III, and Young. Of course there are other tunings that are of interest to composers, probably the best known being "just" tuning. However this set of grief is only part of the problem. The other is peculiar to the piano, and is what you are observing.
The effect you are seeing with piano tuning is actually intentional, and forms part of the piano tuner's art. Pianos are tuned with stretched octaves. Why? Well the problem is that piano strings are not infinitely thin. This results in the witness point (the effective point where the string end appears to be) being slightly different for the different harmonics. Thus a string can actually be out of tune with its own harmonics. Part of the art of a piano tuner is to create a consonant whole, where this effect has the least impact aurally. Over time it was found that this occurred when the octaves were slightly stretched. You may find that a piano is of the order of 30 cents (.3 of a semitone) sharp at the highest notes, and a similar amount flat at the lowest. However this effect is most pronounced on pianos with shorter strings, and a concert grand may exhibit only one third the stretching of a piano suited for home use. This is simply a result of the strings being longer relative to their width, and thus the change of witness point of less relative import.
However, pianos are pretty well the only instruments that are stretch tuned. There have been some efforts to import some ideas to the guitar, but they are limited, and in general the intonation of a guitar is done to provide a harmonic consonance that suits its player, since it is only possible to do this in one zone of the neck. The remainder of the neck becoming more and more dissonant. Most other instruments have no tempering, and no stretching. All fretless instruments for instance. A fretless instrument can, in principle, always play exactly the correct pitch, perhaps why string ensembles work so well. Woodwinds and brass are very much more constrained, and are either designed for a particular key, or are simply fixed by the physics of creating overtones, and cannot be tuned. This leads to all sorts of fun, because they will always be out of tune to some extent when played in an ensemble.
Fundamentally it is insoluble. This comes down to the unique prime factorisation theorem. Is is simply not possible to create the notes of a musical scale only using the ratio 3:2. Which is what equal temperament tries to do. Only octaves and the fifth can ever be correct. The rest, it is simply a matter of how incorrect. And no other tuning will allow key modulation or transposition. So we are stuck.
We can go on from here to reach conclusions about the nature of harmonic distortion, but really we should go to a new thread.
Weighing in with my first post in this forum, and off topic too....
Addressing mfc's comments on the harmonic series and musical instruments. I have seen this mentioned many times here (I'm slowly working my way through the archive.) The issue is much more complex, and annoying, than you think.
From a straight theoretical point of view it is actually impossible to make a musical instrument that is perfectly consonant. Even a simple fretless instrument has internal issues that prevent this. The piano however is the instrument that displays just about every problem known. Being a essentially a fretted instrument it can only be tuned harmonically in one key. To avoid this most people believe it is tuned to the equal tempered scale. Which essentially makes each interval sound equally bad. Actually pianos are not tuned to equal temperament, but are tuned to a plethora of different tunings - ones which balance different intervals in different keys, to what is hoped will be a generally better result. Well known tunings include Werkmeister III, Kimberger-III, and Young. Of course there are other tunings that are of interest to composers, probably the best known being "just" tuning. However this set of grief is only part of the problem. The other is peculiar to the piano, and is what you are observing.
The effect you are seeing with piano tuning is actually intentional, and forms part of the piano tuner's art. Pianos are tuned with stretched octaves. Why? Well the problem is that piano strings are not infinitely thin. This results in the witness point (the effective point where the string end appears to be) being slightly different for the different harmonics. Thus a string can actually be out of tune with its own harmonics. Part of the art of a piano tuner is to create a consonant whole, where this effect has the least impact aurally. Over time it was found that this occurred when the octaves were slightly stretched. You may find that a piano is of the order of 30 cents (.3 of a semitone) sharp at the highest notes, and a similar amount flat at the lowest. However this effect is most pronounced on pianos with shorter strings, and a concert grand may exhibit only one third the stretching of a piano suited for home use. This is simply a result of the strings being longer relative to their width, and thus the change of witness point of less relative import.
However, pianos are pretty well the only instruments that are stretch tuned. There have been some efforts to import some ideas to the guitar, but they are limited, and in general the intonation of a guitar is done to provide a harmonic consonance that suits its player, since it is only possible to do this in one zone of the neck. The remainder of the neck becoming more and more dissonant. Most other instruments have no tempering, and no stretching. All fretless instruments for instance. A fretless instrument can, in principle, always play exactly the correct pitch, perhaps why string ensembles work so well. Woodwinds and brass are very much more constrained, and are either designed for a particular key, or are simply fixed by the physics of creating overtones, and cannot be tuned. This leads to all sorts of fun, because they will always be out of tune to some extent when played in an ensemble.
Fundamentally it is insoluble. This comes down to the unique prime factorisation theorem. Is is simply not possible to create the notes of a musical scale only using the ratio 3:2. Which is what equal temperament tries to do. Only octaves and the fifth can ever be correct. The rest, it is simply a matter of how incorrect. And no other tuning will allow key modulation or transposition. So we are stuck.
We can go on from here to reach conclusions about the nature of harmonic distortion, but really we should go to a new thread.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.