Feedback
When I see, how many people desire amp " without any overall feedback ", I am surprised. In " prehistory " of electronic was this solution used for improvement of parameters and it is valid till today. By my opinion problem is somewhere else : overall feedback work good only on linear region of amp. But by some condition is in internal structure of amp overloaded VAS ( for easy example by limitation ) and in this case amp will start to behave such as with disconnect feedback loop. Amplify in this time very rise and start avalanche of saturations in all amplifier stages, what very extend recovery time of all amp and which is very hearable. Solution is very easy and cheap : you must " switch " VAS gain by limitation on unity gain ( or gain of all amp ) and limitation must be " soft ". In this case amp will be much more stabil and you will hear it 🙂 and advantages of high feedback you can to preserve. In all amp do this function only a few cheap diodes. Concrete solution you can to see on thread below, which was started by PMA.
When I see, how many people desire amp " without any overall feedback ", I am surprised. In " prehistory " of electronic was this solution used for improvement of parameters and it is valid till today. By my opinion problem is somewhere else : overall feedback work good only on linear region of amp. But by some condition is in internal structure of amp overloaded VAS ( for easy example by limitation ) and in this case amp will start to behave such as with disconnect feedback loop. Amplify in this time very rise and start avalanche of saturations in all amplifier stages, what very extend recovery time of all amp and which is very hearable. Solution is very easy and cheap : you must " switch " VAS gain by limitation on unity gain ( or gain of all amp ) and limitation must be " soft ". In this case amp will be much more stabil and you will hear it 🙂 and advantages of high feedback you can to preserve. In all amp do this function only a few cheap diodes. Concrete solution you can to see on thread below, which was started by PMA.
Iv'e just finnished reading through a printed version of this thread.
Charles - thanks for your logical & clear discrimination and explainations, if everyone thought & posted with such clarity I think think the threads could be much shorter !
I have a couple of questions. someone earlier quoted this point from M Otala
"But if you build a amp with a very low open loop output impedance...then you can use overall feedback and get even lower distortion..."
In one of your replies you say that a CFP on the o/p would give the 'sound of feedback' similar to global feedback.
my questions are:
1) Could you say something about the charactoristics of this feedback 'sound'
2) do you have any theories about what it is that causing this sound
3) Do you think that making an amp really fast might help to avoid this
4) Do you think that M Otala's point has any credence. Would biasing amps heavily into class A do anything to improve the situation. Have you tried this approach.
5) Are amps of this design from you still available in the marketplace. I would be very interested to hear one.
regards
mike
Charles - thanks for your logical & clear discrimination and explainations, if everyone thought & posted with such clarity I think think the threads could be much shorter !
I have a couple of questions. someone earlier quoted this point from M Otala
"But if you build a amp with a very low open loop output impedance...then you can use overall feedback and get even lower distortion..."
In one of your replies you say that a CFP on the o/p would give the 'sound of feedback' similar to global feedback.
my questions are:
1) Could you say something about the charactoristics of this feedback 'sound'
2) do you have any theories about what it is that causing this sound
3) Do you think that making an amp really fast might help to avoid this
4) Do you think that M Otala's point has any credence. Would biasing amps heavily into class A do anything to improve the situation. Have you tried this approach.
5) Are amps of this design from you still available in the marketplace. I would be very interested to hear one.
regards
mike
Looking at the popularity of single ended tube amplifiers, nowadays the focus is certainly not on low distortion anymore but on acceptable distortion, or even enjoyable distortion. This is not so strange since it is well known that low order harmonics (2nd, 3rd and 4th) are "musical", i.e. fit into the musical pattern, being an octave, a twelfth (actually -1.9% off) and a double octave. Harmonics higher than 5 do not fit into the musical intervals, except for the 8th harmonic, the 16th etc. The lower harmonics enrich the music. Some have said that as long as the harmonic relations mimic the ones from our hearing (decreasing in amplitude for higher harmonics) the distortion will not harm that much. Increasing amplitudes for high order harmonics, as are generated by e.g. crossover distortion, do not fit in that pattern and will be objectionable. Also feedback will shape the distortion spectrum, increasing the relative amount of higher order harmonics, unlike the distortion of our hearing. Maybe it is good to read Cheevers publication: http://www.decware.com/cheever_thesis.pdf and the one from Boyk and Sussman: http://www.its.caltech.edu/~musiclab/feedback-paper-acrobat.pdf The Boyk and Sussman article shows that local feedback is just as harmful as overall feedback with respect to distortion spectrum shaping.
So feedback (local and overall) should be kept to a minimum to avoid this spectrum shaping. I'm not sure at the moment what error correction will do though, but I guess it also shapes the spectrum towards higher harmonics.
Replacing overall feedback with local feedback can still be a good idea to avoid problems with frequency compensation, transient intermodulation distortion and interface intermodulation distortion.
Steven
So feedback (local and overall) should be kept to a minimum to avoid this spectrum shaping. I'm not sure at the moment what error correction will do though, but I guess it also shapes the spectrum towards higher harmonics.
Replacing overall feedback with local feedback can still be a good idea to avoid problems with frequency compensation, transient intermodulation distortion and interface intermodulation distortion.
Steven
Steven said:
I thought musical instruments are nothing about harmonic generators. Has anyone tried to see how far out those harmonics go? are they mostly low order harmonics?
Charles Hansen said:
Here's an experience directly from the recording side of the chain
you may find interesting.
I design and build 0 FB microphone preamplifiers for recording
professionals and yes they will qualify for Charle's definition of
0 FB 🙂
Recently I replaced an IP mic pre on Amek console with a discrete
0 FB design.
2 pre's had already been upgraded previously with SSM2017 type
monolythic based designs (someone elses) and the rest were the
consoles standard transformer into 5534 design.
We did some comparisons recording a nice old Gibson J50 ->
Reslo ribbon mic -> Console Pre -> Console bus (opamp) ->
Ampex ATR 2" analog deck. Playback Ampex -> line IP console
(opamp) -> summing bus (opamp) -> OP channel (opamp) ->
amp -> speakers.
So as you can see there were significant amounts of electronics
in this minimalist (by studio standards) chain. AFAIK, the ampex
has discrete FB based electronics.
The results were quite an eye opener. The discrete 0 FB pre had
a much more open, dynamic and real sound compared to the
chip based (2017) and transformer/5534 preamps. Also bear
in mind we had the guitar right there for comparisons.
No electronic chain can replay exactly the sound of that GTR,
however the discrete 0 FB pre did a better job at capturing the
magic of that classic J50 and that's what the engineer wanted,
he didn't give a hoot about THD plots.
It does make life difficult wrestling with these subjective
and objective differences especially when you hear it right
out of the studio speakers.
Cheers,
Terry
Hey Jocko -
I know that you change your little picture from time to time. The latest looks like a Lotus Super 7. Did you get one of these lately? I bet they're a hoot to drive!
Cheers,
Charles Hansen
I know that you change your little picture from time to time. The latest looks like a Lotus Super 7. Did you get one of these lately? I bet they're a hoot to drive!
Cheers,
Charles Hansen
Terry Demol :
Oh, you are comparing different pre-amplifiers with and without transformer, having differen input impedances, input capacitances and gains
A ribbon microphone is a very exotic 1960's retro thing that has particular characteristics that change with things like load impedance and capacitance, they also have very low sensitivity so they usually need 10:1 or more step-up transformers
So I see perfectly normal that you hear differences, to the point that I think that the only thing that doesn't have any influence on the result is the fact of having or not global feedback on the preamp
You have overlooked the rest of the parameters, maybe even the floor reflections and room acoustics that make impossible to have an acoustic guitar recorded through a mic sounding not even close to the original
[I have little experience but I learnt this just the first time I had to mic an acoustic guitar... it even sounds different than expected simply when you listen with the ear at the mic place near the guitar instead of standing up]
In my experience, changing input impedance, or moving the mic, the guitar or the listening position a couple of inches, or even changing the listening volume a couple of dBs, or just knowing that you are using/selling your preamp causes far more changes in perceived sound than all those no-feedback snake oil
Everything sounds brighter when things are done in the way you wanted
Oh, you are comparing different pre-amplifiers with and without transformer, having differen input impedances, input capacitances and gains
A ribbon microphone is a very exotic 1960's retro thing that has particular characteristics that change with things like load impedance and capacitance, they also have very low sensitivity so they usually need 10:1 or more step-up transformers
So I see perfectly normal that you hear differences, to the point that I think that the only thing that doesn't have any influence on the result is the fact of having or not global feedback on the preamp
You have overlooked the rest of the parameters, maybe even the floor reflections and room acoustics that make impossible to have an acoustic guitar recorded through a mic sounding not even close to the original
[I have little experience but I learnt this just the first time I had to mic an acoustic guitar... it even sounds different than expected simply when you listen with the ear at the mic place near the guitar instead of standing up]
In my experience, changing input impedance, or moving the mic, the guitar or the listening position a couple of inches, or even changing the listening volume a couple of dBs, or just knowing that you are using/selling your preamp causes far more changes in perceived sound than all those no-feedback snake oil
Everything sounds brighter when things are done in the way you wanted
mikelm said:
Hello -
Thanks for the kind words. I'll try to answer your questions as best I can. (They are good questions, but perhaps rather difficult to answer.)
1) In my experience, the application of feedback (pretty much any sort of feedback, as I defined it earlier in this thread) results in a sonic character that is somewhat "mechanical" or "electronic" in nature. In contrast, a zero feedback design tends to be more "organic" or "natural". I'm sorry I can't give a better description than this, but once you hear it I think you'll know what I mean. There are a few caveats:
a) Feedback seems to have less of a detrimental effect on tube electronics than it does on solid state electronics.
b) Many people prefer the sound of feedback circuits. It tends to have more *apparent* resolution, while usually having less *actual* resolution. It tends towards the "hi-fi spectacular" presentation rather than the "musically emotional" presentation.
c) It may take a while to get "dialed into" the sound of zero feedback circuits. Basically you have been listening to feedback circuits for your entire life. When you first hear zero feedback, there may be some time required to "acclimate" to it.
d) It is easier to hear what zero feedback does when the entire playback chain is zero feedback.
2) I think there are some plausible mechanisms as to what creates the "feedback" sound, but I am hesitant to discuss them in this forum. Based on my past experience the main result is that people want to argue with me because I haven't proven to their satisfaction that these mechanisms are "real".
3) It might. It even probably would. But to me the better question is "why use feedback at all"? Instead of trying to minimize the sonic damage it inflicts, why not just use zero feedback circuits and avoid the problems altogether?
4) Yes of course he is correct. You can always reduce the static distortion measured on a test bench into a resistive load of any circuit by applying feedback.
5) Everything that Ayre makes and has ever made is zero feedback. Even the power supply regulators are zero feedback. This applies to our video circuits as well as our audio circuits. In that case, you can see the difference that zero feedback makes.
(Confession -- our very first product was a power amplifier that had DC feedback around the output stage only. Since then we have avoided even using any DC feedback, including any form of servo. Most of our products are true DC amplifiers, although our phono stages are cap-coupled.)
Best regards,
Charles Hansen
Ayre Acoustics, Inc.
All posts that were not directly related to the main topic (which is building a non feedback amplifier), were moved here http://www.diyaudio.com/forums/showthread.php?s=&threadid=28764
The discussion is really interesting so far, so please keep it on track.
Peter Daniel
diyAudio Moderator
The discussion is really interesting so far, so please keep it on track.
Peter Daniel
diyAudio Moderator
It was mentioned earlier that a non-feedback amp with a low output impedance would be desireable. I just finished building such an amp today. Yet it's probably not what people in this thread are looking for.
It's a Class-D/Class-I Balanced Current Amplifier (BCA, see also the crownaudio patent and website for details). It doesn't have feedback in the audio chain (yes, the opamps and comparators used for modulation do have feedback, but this is only used for modulation, not for error correction of the output). Some initial tests today showed that it's fairly load-independent.
The output looks really nice on the scope, and even when driven into clipping, the sinewave only gets a bit squashed / compressed. It also handles DC without effort (into resistors, don't want to loose a speaker...), and since the output stage is a switching bridge with freewheeling diodes, it also shouldn't have a problem with inductive loads (haven't tested that yet).
Unfortunately I can't measure THD. Can anyone tell me how this is done? This is just a hobby for me, and in fact this is the first power amp I ever built, so I'm not very experienced. Is there a way to measure THD using an Audio Generator and an Oscilloscope? Or maybe with a PC, a high-quality soundcard, and some software? I'd really like to get some idea about the performance of this amp.
Has anyone ever listened to a Crown Audio Class-I BCA amp? Are there any other BCA amps on the market? I think they must be great, if even I can build one on veroboard, that is not too bad. I'd really like to listen to one some day.
Cheers
Felix
It's a Class-D/Class-I Balanced Current Amplifier (BCA, see also the crownaudio patent and website for details). It doesn't have feedback in the audio chain (yes, the opamps and comparators used for modulation do have feedback, but this is only used for modulation, not for error correction of the output). Some initial tests today showed that it's fairly load-independent.
The output looks really nice on the scope, and even when driven into clipping, the sinewave only gets a bit squashed / compressed. It also handles DC without effort (into resistors, don't want to loose a speaker...), and since the output stage is a switching bridge with freewheeling diodes, it also shouldn't have a problem with inductive loads (haven't tested that yet).
Unfortunately I can't measure THD. Can anyone tell me how this is done? This is just a hobby for me, and in fact this is the first power amp I ever built, so I'm not very experienced. Is there a way to measure THD using an Audio Generator and an Oscilloscope? Or maybe with a PC, a high-quality soundcard, and some software? I'd really like to get some idea about the performance of this amp.
Has anyone ever listened to a Crown Audio Class-I BCA amp? Are there any other BCA amps on the market? I think they must be great, if even I can build one on veroboard, that is not too bad. I'd really like to listen to one some day.
Cheers
Felix
sfx said:
Hi sfx,
A bit off-topic, but here you go:
http://www.sumuller.de/audiotester/maine.htm
http://www.dazyweblabs.com/shannonsoft/page3.html
http://audio.rightmark.org/
Maybe it is a good idea to show us your BCA amplifier. I think it deserves a separate thread. At least I'm curious to how you did it, and in general to the pros and cons of balanced current amplification.
Steven
Steven said:
But there is nothing like this. Or better say, there are two possible ways:
1. Mask the imperfections of the audio chain by specific amps sound. You may get some results, you may get some listeners and admiration and you may get better results than many middle-end machines can give.
2. Or you try to reach the lowest possible distortion, maximal linearity at all the stages, matched cable signal transfer at higher power (50 Ohm load), volume control with switched low impedance attenuators, HF filtering, HF and RF intermodulations suppression. This way is far more difficult than that one mentioned in point 1. But brings results, not achievable by masking distorting colorating instruments.
PMA said:
What do you think of de Lima's arguments with regard to greater amplifier distortion ultimately being capable of resulting in less distortion being output from the speakers?
http://www.audiopax.com/whitepapers.html
se
Steve,
I would need more time to study the paper in detail. I suppose that it is very difficult to compensate one kind of distortion by another kind. And, the character of speaker distortion is quite different from distortion of any active element.
During development of the audio chain with very low distortion you get through several steps. Firstly the results are disappointing, disclosing the imperfections that were masked by distorting instrument. Then after solving the details like signal transfer, filtration, PSU, mains filter etc. etc. you start to get absolutely pleasant pure sound with details that have been masked before. I am very surprised of the capabilities of the electro-acoustic transducer like electrodynamic speaker, we still have enough to expect from it.
I would need more time to study the paper in detail. I suppose that it is very difficult to compensate one kind of distortion by another kind. And, the character of speaker distortion is quite different from distortion of any active element.
During development of the audio chain with very low distortion you get through several steps. Firstly the results are disappointing, disclosing the imperfections that were masked by distorting instrument. Then after solving the details like signal transfer, filtration, PSU, mains filter etc. etc. you start to get absolutely pleasant pure sound with details that have been masked before. I am very surprised of the capabilities of the electro-acoustic transducer like electrodynamic speaker, we still have enough to expect from it.
Steve Eddy said:
Steve, I read the first part. I am extremely sceptical about the reversing of the speaker leads to cancel THD. Reversing the leads does nothing to change the RELATIVE phase of the speakers' distortion components to the fundamental. If reversing would cancel the 2nd harmonic, why doen't it cancel the fundamental as well, resulting in...silence?
Jan Didden
- Status
- Not open for further replies.