I don't know what amp is that schematic, but I ever built 2 amps very similar in the year around 1999 also with that 2 diode and even simpler, most people liked them.
Esperado,
It seems young (and uncivilized ) people (always the same) just do not realize that phase or delays matters in a feedback path.
Phase or delays ?
Phase delays or time delays ?
Once again, the usual confusion.
Bob Cordell wrote a short paragraph on delay in feedback.
It's a reccurent topic :
http://www.diyaudio.com/forums/tubes-valves/177139-feedback-delay-time.html#post2362902
http://www.diyaudio.com/forums/blogs/jan-didden/454-feedback-how-late-time-same-time-all-time.html
Kindhornman,
I was only saying that I don't think that there was anyone building a class-D amplifier from scratch as a diy project but speaking about commercial designs that are created by Bruno. I am not sure that there is at this time a demand to build diy class-D amplifiers. Not yet anyway.
This is why I feel a bit uneasy with Bob Cordell chapter on class D, it's not directed towards DIY. I think it's a topic needing a dedicated book.
It seems young (and uncivilized ) people (always the same) just do not realize that phase or delays matters in a feedback path.
Phase or delays ?
Phase delays or time delays ?
Once again, the usual confusion.
Bob Cordell wrote a short paragraph on delay in feedback.
It's a reccurent topic :
http://www.diyaudio.com/forums/tubes-valves/177139-feedback-delay-time.html#post2362902
http://www.diyaudio.com/forums/blogs/jan-didden/454-feedback-how-late-time-same-time-all-time.html
Kindhornman,
I was only saying that I don't think that there was anyone building a class-D amplifier from scratch as a diy project but speaking about commercial designs that are created by Bruno. I am not sure that there is at this time a demand to build diy class-D amplifiers. Not yet anyway.
This is why I feel a bit uneasy with Bob Cordell chapter on class D, it's not directed towards DIY. I think it's a topic needing a dedicated book.
Last edited:
Don't it was clear i told about input capacitances of the (base and emitter) input transistor ?
I suppose you know the charge (and discharge) curve of a capacitor ? (i'm too able to be supposing and arrogant)
What is the difference, at a given frequency, between a phase difference and a delay ? How can-you tell looking at the two waves in an oscilloscope ?
Unless you have a flat bandwidth in open loop, you will have phase turns at the upper frequencies between original signal and feedback signal.
They will act like a delay (increasing with frequency) between error and correction. As the amplifier is not perfect, this will increase high order distortions.
To jan.didden, (because your comment is interesting), and before to leave definitively this thread, I agree with what you said about the benefit to get as much feedback at 20KHz as you can, that i prefer to describe (looking at my green and red graph ;-) as moving the decreasing slope (apologize for my poor English) at higher frequency.
All what i tried to propose, with a lot of pain here, was CFA is a nice way to move the slope to the right of several octaves, or to get higher slew rates, or...all is correlated and so simple..
About "how much feedback as you can", read in an absolute way, i'm not so sure. I usually work in a very pragmatic way, trying to set each stage gain (IE local feedbacks) for minimal distortion results in closed loop and fixed global gain. (Yes, michaelkiwanuka, i own a distortiometer)
This will be my last contribution in this thread.
OT/
Sounds, they're all dependant on exhaust tuning. My VR6 has sweet howl at 6K+ rpm and a deep growl at 2K
/OT
MikeK, how much time have you spent on controlled listening tests?
Not being argumentative or taking sides .. just after clarification to put your comments & pontificating in their proper perspective.
Nonsense jan.
In a VFB you can have 70dB or more loop gain at LF and say 50 dB at 20k. In a CFA it's quite easy to have 50dB at 20 k, where of course it matters, and 50 dB at LF.
I don't know about you, but if the distortion at 1k is already at circa 10 ppm with 50dB of loop gain - CFA example - I see little point in piling on even more feedback which is what you are suggesting.
As for the comment on the sound, well I really don't know where you got that one. PIROOMA perhaps?
Last edited:
if IMD is the bigger audible problem with distortion than harmonics - then high loop gain at our most sensitive hearing range, in the range of most musical instrument's fundamental or peak energy seems more valuable than "flat loop gain"
since higer low frequency loop gain feedback reduces the error of IMD difference products folding down into lower, more sensitive frequencies by the higher loop gain at those frequencies - and IMD difference products are anharmonic so masking isn't as applicable
IMD also includes TIM, FM, PID as subclasses - and they are reduced by the amount of feedback gain at the product frequency
additionally you have to add resistive shunt or internal feedback in typical VFA to flatten the loop gain response over audio - perhaps "throwing away" the gain at worst, at best just not applying it as effectively by using it in the local loop(s)
this is a well known result - that higher global open loop gain feedback gives greater distortion reduction than the same gain divided up, "linearizing" local stages in series - and why feedback theorists from Bode, Mason, Horowitz to Lurie all recommend maximizing global feedback within the limits of the global loop stability constraints
since higer low frequency loop gain feedback reduces the error of IMD difference products folding down into lower, more sensitive frequencies by the higher loop gain at those frequencies - and IMD difference products are anharmonic so masking isn't as applicable
IMD also includes TIM, FM, PID as subclasses - and they are reduced by the amount of feedback gain at the product frequency
additionally you have to add resistive shunt or internal feedback in typical VFA to flatten the loop gain response over audio - perhaps "throwing away" the gain at worst, at best just not applying it as effectively by using it in the local loop(s)
this is a well known result - that higher global open loop gain feedback gives greater distortion reduction than the same gain divided up, "linearizing" local stages in series - and why feedback theorists from Bode, Mason, Horowitz to Lurie all recommend maximizing global feedback within the limits of the global loop stability constraints
Last edited:
Well, I am not advocating flattening the gain by loading the TIS although I have done that for experimental purposes.
There is NO scientific study indicating a link between higher leavels of -ve feedback and better sound, and especially at the very low levels we are talking about here. Lots of opinions and passionate arguments, but no facts.
Bode, Mason, Horowitz et al may well teach what to do to get better measured performance, but they cannot answer the question: does more feedback at lower frequencies make an amplifier sound better?
There is NO scientific study indicating a link between higher leavels of -ve feedback and better sound, and especially at the very low levels we are talking about here. Lots of opinions and passionate arguments, but no facts.
Bode, Mason, Horowitz et al may well teach what to do to get better measured performance, but they cannot answer the question: does more feedback at lower frequencies make an amplifier sound better?
Michael,
I just read your statement. So if these CFA things are demonstrably inferior, why could one not tell the difference in controlled listening tests?
Excellent question.
I have never claimed I am a purist but regardless how one achieves it, higher gain at lower frequencies does make a substantial difference in improving the sound. After adding Linkwitz Railey equalization into the feedback and input of my amp, I have never gone without it in 10 years and everyone ever listening to music at my place remains glued to the chairs. You could not do this befor digital music since record warp would destroy your system.
You have to have the power supply to sustain operation at this elevated gain at LF. There is just no way that anyone can argue away this fact and the only way you can experience it is adding it into your system.
You have to have the power supply to sustain operation at this elevated gain at LF. There is just no way that anyone can argue away this fact and the only way you can experience it is adding it into your system.
Well this is actually the point I was making. If the two cases have the same LG at 20kHz, as in your example, they will have the same sound quality (other things being equal, for the discussion).
The CFA curve looks flat (it IS flat), and the VFA looks like it starts to drop already at very low frequency. But any way you look at it, the VFA will have better specs below 20kHz. Whether that matters is another issue, but the LF distortion of the VFA will be lower, the DC offset will be lower, the damping factor will be better; the important 20kHz spec and subjective sound quality will be the same or better, although the CFA will look 'flatter' and having a wider bandwidth!
jan
Last edited:
Bonsai, basically your question reduces to: does lower non-linearity make an amplifier sound better?
But compared to what? To be meaningfull, the question should be framed in 'all other things being equal', and then the answer is yes.
OTOH if you increase negative feedback but in that process change some other parameter in the amp, the result may be higher feedback but lower sound quality.
jan
Diminishing returns. At a given point it just becomes a numbers pushing sports.Excellent question.
Excellent question.
Now you are saying that you could one not tell the difference in controlled listening tests??
Those who claim that CFA is demonstrably superior are more aggressive here then opposite.
In my opinion both could sound equally good when properly designed.
OT/OT/
Sounds, they're all dependant on exhaust tuning. My VR6 has sweet howl at 6K+ rpm and a deep growl at 2K
/OT
Sure, the sound depends on the construction of the exhaust (obviously). However, opposed to a flat-plane crankshaft, the firing order of a cross-plane crankshaft is irregular (LRLLRLRR or so), resulting in the typical V8 burble sound of US V8 engines (brab brab brab brab), like one of the cylinders is not firing. A nice vrooooom requires a much more complicated exhaust system, which rearranges the order of the exhaust pulses, see pic.
/OT
Attachments
When ears are no more adequate enough all is left
are measurements and in this matter VFAs "sound better"
the same way that an average SS amp will indeed sound
better than distorted SETs unless , of course , that distorted
signals sounds more pleasing , wich can be the case.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.