There is a difference between data sheet and real life. Try to measure your amp. It is not hard to do with $150 worth audio interface and free (or very cheap) software for you computer. If some piece of audio equipment measures bad, it can't sound good.
Last edited:
OK, I am back. As far as the bode plot goes I held off because I am in a time crunch, school and all. I just finished my internship at TE and am back in classes.
As far as fiddling goes, because of my time issue I was kind of reduced to shooting in the dark. The resistors from the KT88 anodes in conjunction with the RC network after the SRPP quieted it.
Some of you think it was foolish to wrap a feedback loop around what I have in the circuit. I loosely based my driver off of a Williamson. Not the most stable, but thought it was worth the engineering challenge, as I love them. If I were to perform this exercise again, do you think it would be wiser to perhaps find a way to direct couple something instead of using all the caps?
As far as listening goes, woah, I better watch it. The bass is really intense. Is there any good recordings out there with complex passages to test these amps well? My musical tastes are open. Just no rap crap or hip hop mediocre culture ********!
As far as fiddling goes, because of my time issue I was kind of reduced to shooting in the dark. The resistors from the KT88 anodes in conjunction with the RC network after the SRPP quieted it.
Some of you think it was foolish to wrap a feedback loop around what I have in the circuit. I loosely based my driver off of a Williamson. Not the most stable, but thought it was worth the engineering challenge, as I love them. If I were to perform this exercise again, do you think it would be wiser to perhaps find a way to direct couple something instead of using all the caps?
As far as listening goes, woah, I better watch it. The bass is really intense. Is there any good recordings out there with complex passages to test these amps well? My musical tastes are open. Just no rap crap or hip hop mediocre culture ********!
Passage 138BC-AD 1611 by Empire Brass Quintet; Some tracks- Spiritual Dance; Hopper Dance;Factus Repente will make SE's sob with fear. There are others more to date:-
Laura Paucini; also Lucia Dalla, Canzone (pop Italian & folk) have a lovely well defined vocals, the slight rasp of Dalla and the regular shake of the tamborines will check ones tweeters and imaging.
Vangelis Reprise typ analogue, but my version has with very little noise. Heavily remixed.
Grover Washington....All My Tomorrows, nice soft sax for coctails, easy company.
The hardest for any power amp to handle with power I have come across is Trompete & Orgel, Maurice Andre/Hedwig Bilgram...(F&DE), there is wind noise from the organ but the lowest passages at 16Hz, with trumpet included will determine how good the power amp LF loop behaves at power.
This test is often better using a dummy load with headphones correctly terminated. All speakers will create some distortion and intermod and the only version that can handle this is a Labyrinth.
richy
Try this Amazon.com: Folk Singer (HDAD Master): Muddy Waters: Music
You can play it in regular DVD player in 24/96 mode. If you have DVD-A player, then 24/192 tracks are outstanding. It is hard to believe that it was recorded 50 years ago. Audio engineers of that time knew what they were doing.
Spot on....Most went through transformer input Neve consoles & modules which commanded a min-95dB down on "A" noise, and excellent top end separation on 64 bus systems. The b/w typ 100Khz.
If I wanted to do an amplifier the right way, and use feedback correctly, I'd measure what needs to be measured. That would be group delay or phase shift vs frequency.
I wonder how many people add negative feedback that actually increases distortion, because the only goal is to have it not oscillate or ring.
I think this is a good point!
The very fact you have phase margin admits the truth of the matter: the correction signal that GNFB punts into the input can be too late by the time it mixes into the output. Perhaps anything over a 5% phase shift should be classified as something made worse by feedback...
If you do not see more than 5% overshoot on the front of square signal, then delay is not a problem.
So if there is no delay, where does the phase shift come from?
And regardless, doesn't a phase shift imply that the FB is now correcting the wrong part of the waveform?
So effectively with phase shift the listener hears the original distorted sound with a quieter inverted phase shifted doubly distorted sound (as it's been through twice now) ?
Saying it isn't a problem maybe a case of defining the problem better, as the subject is hi-fi rather than a tannoy at a school sports day
Phase shift and delay are different things, although they can superficially look similar. A true delay is what things like transmission lines do. You put a signal in one end and then you have to wait before it begins to come out the other end. A true delay in a loop is a real nuisance, and can be difficult to keep stable.
A phase shift, perhaps caused by a low pass filter, gives a change of output voltage instantaneously. It may just ramp up slowly, but it does start moving as soon as the input changes. In a properly designed loop this is all you need.
Feedback really does work, when properly used. The issue of distortion getting reduced but further distorted has nothing to do with delays or phase shifts but is a consequence of instantaneous algebra. To put it simply, if you take 1/(1+x) as describing feedback then you find that you need an infinite series on top to give the same result as a simple polynomial below.
1/(1+x) = 1 -x +x^2 -x^3 +x^4 etc.
A phase shift, perhaps caused by a low pass filter, gives a change of output voltage instantaneously. It may just ramp up slowly, but it does start moving as soon as the input changes. In a properly designed loop this is all you need.
Feedback really does work, when properly used. The issue of distortion getting reduced but further distorted has nothing to do with delays or phase shifts but is a consequence of instantaneous algebra. To put it simply, if you take 1/(1+x) as describing feedback then you find that you need an infinite series on top to give the same result as a simple polynomial below.
1/(1+x) = 1 -x +x^2 -x^3 +x^4 etc.
Thanks for the explanation, I realise the actual electrical delay is will infinitely small (the speed of light over 10" or so) so perhaps I was thinking more of Group Delay?
Negative feedback certainly can work - of course - but IMO a phase shift means that at that instance in time one of the signals in wrong. As the feedback is supposed to sum at the input (to correct the output) I can't see how it can do that properly if the original signal and the corrected signal arrive at the output with phase difference..
I.e. how does the feedback work properly with a phase shift?
What effect does NFB have on inter-modulation distortion?
An old timer Hi-Fi repair man filled me on the three kinds of distortion:
- Total Harmonic Distortion.
- Inter-modulation distortion.
- Distortion due to slewing (the input signal is too fast for the amplifier circuitry).
An old timer Hi-Fi repair man filled me on the three kinds of distortion:
- Total Harmonic Distortion.
- Inter-modulation distortion.
- Distortion due to slewing (the input signal is too fast for the amplifier circuitry).
Last edited:
Negative feedback reduces all types of distortion by the same ratio that it reduces gain, provided that the amplifying stage is reasonably linear (the equations commonly seen assume perfect linearity). This is a reasonable assumption with an ordinary vacuum valve amplifier operated below clipping.
Slewing isn't really distortion because output is no longer a function of input, but it's good to be mindful of the mechanism that causes it, the finite current available to charge capacitances.
All good fortune,
Chris
Slewing isn't really distortion because output is no longer a function of input, but it's good to be mindful of the mechanism that causes it, the finite current available to charge capacitances.
All good fortune,
Chris
It works is you consider the limited bandwidth. No signal that can change infinitely quickly.
Negative feedback works quite well. Just consider that almost all audio DACs in modern players are delta-sigma type, where feedback is what makes them operational.
Yes I've heard this - the premise that the amplifier has to be linear totally baffles me. I wouldn't need to use any (lowering output impedance excepted) if the amp was linear?!
I suspect that means that the correction signal arrives dead on phase and without distortion - but that's wrong isn't it? It actually arrives with phase lag and exactly the same distortion (but inverted) as the original - surely?!
Would this not act like an analog multiplier circuit?
I want NFB to correct PSU issues, crossover distortion and distortion added by coupling capacitors and the tubes themselves, it's of no use to me in a linear amp....
Has anyone done any papers on NFB through a non-linear amplifier? The phrase 'reasonably linear' worries me as it implies a 'not notice' factor whereas I am after the best solution, not the usual boringly mediocre ones
.Feedback works properly with a phase shift. Two ways to see this:
1. do the maths
2. trust those who can do the maths when they tell you it works
Feedback reduces existing IM in exactly the way it reduces anything else in the output which is not in the input. In addition, it creates new IM in exactly the same way it creates higher order harmonics. This is the reason why the amplifier should be as linear as possible open-loop, before you close the loop.
Trying to explain mathemetical ideas in sentences is difficult, that is why mathematical notation was invented. To understand feedback you need to read a good book on feedback (or servo theory - same thing, different application).
1. do the maths
2. trust those who can do the maths when they tell you it works
Feedback reduces existing IM in exactly the way it reduces anything else in the output which is not in the input. In addition, it creates new IM in exactly the same way it creates higher order harmonics. This is the reason why the amplifier should be as linear as possible open-loop, before you close the loop.
Trying to explain mathemetical ideas in sentences is difficult, that is why mathematical notation was invented. To understand feedback you need to read a good book on feedback (or servo theory - same thing, different application).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.