Re: Re: Hi I'm not an actual amp designer, but I play one in academia
Not that I am able to understand all the technical issues but that is not necessary to feel exactly the opposite. It appears Fred is not trying to save face but you are. And on top of that doing it really badly. Why not try an open mind?
BTW: Spelling mistakes are called "spelling mistakes" due to the fact that they are mistakes and not deliberate. In case of diliberate spelling errors this does not qualify as a mistake.
mikek said:
--snip--
You still appear incapable of debating an issue on its merits, merely resorting to sophistic word-play in a vain-glorious, and desperate attempt to save face, even when its clear to all and sundry that the urgument is irrevocably lost.
Not that I am able to understand all the technical issues but that is not necessary to feel exactly the opposite. It appears Fred is not trying to save face but you are. And on top of that doing it really badly. Why not try an open mind?
BTW: Spelling mistakes are called "spelling mistakes" due to the fact that they are mistakes and not deliberate. In case of diliberate spelling errors this does not qualify as a mistake.
Re: Re: Re: Hi I'm not an actual amp designer, but I play one in academia
To what?
Actually, if you look back through my exchanges with fred on this forum, you'll find that he rarely passes up an opportunity to censure me over my spelling errors....merely returning the courtesy old chap..
UrSv said:
To what?
UrSv said:
Actually, if you look back through my exchanges with fred on this forum, you'll find that he rarely passes up an opportunity to censure me over my spelling errors....merely returning the courtesy old chap..
Re: arriving at conclusions without understanding the urgument...
Mike, I actually agree with you on the skew rate argument (but I am not sure what actual good it does to sound). But I have to agree with there is a general "tone" or "attitude" in your post that may shut people off, in spite of all the good arguments you may have in those very posts.
Why don't we let go the personal side of this discussion, and focus on the merits of the design?
We are all adults, aren't we?
mikek said:
Mike, I actually agree with you on the skew rate argument (but I am not sure what actual good it does to sound). But I have to agree with there is a general "tone" or "attitude" in your post that may shut people off, in spite of all the good arguments you may have in those very posts.
Why don't we let go the personal side of this discussion, and focus on the merits of the design?
We are all adults, aren't we?
Re: Re: arriving at conclusions without understanding the urgument...
...you do have a point there ....
....i've got to stop allowing fred's inane'poking a bear' attitude get to me....
millwood said:
Mike, I actually agree with you on the skew rate argument (but I am not sure what actual good it does to sound). But I have to agree with there is a general "tone" or "attitude" in your post that may shut people off, in spite of all the good arguments you may have in those very posts.
Why don't we let go the personal side of this discussion, and focus on the merits of the design?
We are all adults, aren't we?
...you do have a point there ....
....i've got to stop allowing fred's inane'poking a bear' attitude get to me....
Re: arriving at conclusions without understanding the urgument...
Oh, I have no problem doing that. I said I don't understand all of them. I do however understand enough of them to get an opinion and there is absolutely nothing stopping me from looking at two sides of an argument and using skills in, or knowledge of, technical issues, language matters, social behaviour, language matters and commons sense to do so. That goes for the non-technical posts as well.
mikek said:
Oh, I have no problem doing that. I said I don't understand all of them. I do however understand enough of them to get an opinion and there is absolutely nothing stopping me from looking at two sides of an argument and using skills in, or knowledge of, technical issues, language matters, social behaviour, language matters and commons sense to do so. That goes for the non-technical posts as well.
....anomalous behaviour....
I suspect this may come down to freedom from intermittent parasitic oscillation, discussed in:
'New Factors in Power Amplifier Design'
by Tomlinson Holman.
J.A.E.S, Vol. 29, Number 7 pp. 517 (1981)
It is often easier to rid simple designs such Hiraga's, (which eschew S.O.A protection), of such anomalous behaviour.
However, this is easily detectable by real time observation of the distortion residual..
AKSA said:
I suspect this may come down to freedom from intermittent parasitic oscillation, discussed in:
'New Factors in Power Amplifier Design'
by Tomlinson Holman.
J.A.E.S, Vol. 29, Number 7 pp. 517 (1981)
It is often easier to rid simple designs such Hiraga's, (which eschew S.O.A protection), of such anomalous behaviour.
However, this is easily detectable by real time observation of the distortion residual..
Mikek,
Thank you for pointing that out - I am well aware that loss of 'Lifeforce' is short term loss of resolution engendered by oscillation.
However, I have found that in order to keep an open mind one should never state obscure details too baldly; one should skirt around the point a little to foster imagination.
I find ladies more alluring with just a tad of clothing......
Cheers,
Hugh
Thank you for pointing that out - I am well aware that loss of 'Lifeforce' is short term loss of resolution engendered by oscillation.
However, I have found that in order to keep an open mind one should never state obscure details too baldly; one should skirt around the point a little to foster imagination.
I find ladies more alluring with just a tad of clothing......
Cheers,
Hugh
I've just read through the Mikek/Dieckmann posts after a spell of a couple of weeks.
Hell, what's the point? I would describe it all quite bluntly if it weren't for my admiration for both of them; clever people.
When someone insults you (or your family) in public, you can certainly justify fighting back, even if it is libel in a newspaper. But when it's merely words on a monitor, viewed by choice with a small number of participants, it's hard to take it seriously when its physical presence is nil.
If someone really has a potshot at me, I try to laugh it off, and just walk away. Better not to reply. I have my blood pressure to consider, and won't allow it to rise for anyone. The constant flow of household bills does that for me anyway!
Fred, please, stop reacting, and Mikek, for crissakes build an amplifier, any amplifier, and listen to it carefully as well as masterfully analyzing it.....
Cheers,
Hugh
Hell, what's the point? I would describe it all quite bluntly if it weren't for my admiration for both of them; clever people.
When someone insults you (or your family) in public, you can certainly justify fighting back, even if it is libel in a newspaper. But when it's merely words on a monitor, viewed by choice with a small number of participants, it's hard to take it seriously when its physical presence is nil.
If someone really has a potshot at me, I try to laugh it off, and just walk away. Better not to reply. I have my blood pressure to consider, and won't allow it to rise for anyone. The constant flow of household bills does that for me anyway!
Fred, please, stop reacting, and Mikek, for crissakes build an amplifier, any amplifier, and listen to it carefully as well as masterfully analyzing it.....
Cheers,
Hugh
AKSA,
"that loss of 'Lifeforce' is short term loss of resolution engendered by oscillation."
This certianly seems plausible but isn't it fairly easy to tell if an amp is going to oscillate or ring by looking at the square wave response on a scope?
Or are we talking parastic oscillations at individual transistors that aren't so obvious?
"It's very hard to preserve 'lifeforce' AND stability"
If "lifeforce" is compromised by some oscillation then won't that and stability go hand-in-hand ? I get the impression the two are a trade-off..
Thanks for adding some decent info to this thread.
"that loss of 'Lifeforce' is short term loss of resolution engendered by oscillation."
This certianly seems plausible but isn't it fairly easy to tell if an amp is going to oscillate or ring by looking at the square wave response on a scope?
Or are we talking parastic oscillations at individual transistors that aren't so obvious?
"It's very hard to preserve 'lifeforce' AND stability"
If "lifeforce" is compromised by some oscillation then won't that and stability go hand-in-hand ? I get the impression the two are a trade-off..
Thanks for adding some decent info to this thread.
Hi Dave,
Good point. I knew just after I posted that it would bring on questions!
If you absolutely ensure stability by ramping up the lag compensation cap, you lose the 'lifeforce'. The amp is extremely well behaved, but microdetail is lost. Of course, this is technically because you slow it very considerably.
Reduce the lag comp so it sounds good, and you risk oscillation with many speaker loads. Not all mind, and the Magneplanars, being purely resistive, can dispense with interstage phase correction capacitors, but with many, particularly those which use shunt crossovers like the Sonus Faber.
So, yes, it is indeed a balancing act, and goes a long way to defining exactly why some very good amplifiers are unstable with certain loudspeakers, and not others. This highlights the crucial matching between amps and speakers, something not well understood, and eminently explainable in terms of phase shift, and Bode/Nyquist plots.
However, the question remains; if slew rate is not important beyond a certain figure - usually about 10V/uS at the speaker - then why is lag compensation so vital to good sonics?
Cheers,
Hugh
Good point. I knew just after I posted that it would bring on questions!
If you absolutely ensure stability by ramping up the lag compensation cap, you lose the 'lifeforce'. The amp is extremely well behaved, but microdetail is lost. Of course, this is technically because you slow it very considerably.
Reduce the lag comp so it sounds good, and you risk oscillation with many speaker loads. Not all mind, and the Magneplanars, being purely resistive, can dispense with interstage phase correction capacitors, but with many, particularly those which use shunt crossovers like the Sonus Faber.
So, yes, it is indeed a balancing act, and goes a long way to defining exactly why some very good amplifiers are unstable with certain loudspeakers, and not others. This highlights the crucial matching between amps and speakers, something not well understood, and eminently explainable in terms of phase shift, and Bode/Nyquist plots.
However, the question remains; if slew rate is not important beyond a certain figure - usually about 10V/uS at the speaker - then why is lag compensation so vital to good sonics?
Cheers,
Hugh
Hugh,
Thanks for the prompt reply.
From a technical P.O.V. decreasing the Lag cap will increase gain bandwidth product and increase feedback above the dominant pole frequency. Could this possibly bring subjective benefits?
If you were to decrease the value of the Lag cap and increase the value of the degeneration resistors in the differential pair you could keep the gain-bandwidth product constant. If that brought about the same subjective benefits as just decreasing the lag cap you could rule out the gain-bandwidth product idea.
I think the dual feedback path approach is interesting (like in the leach amp) where feedback at high frequency is taken from the output of the VAS or driver stage. This by-passes any poles formed in the output stage and loading. This at least should improve overall stability especially with difficult loads and therefore enable you to reduce the value of the lag cap.
Sure distortion would rise at high frequencies as feedback decreases around the output stage but this should be in the ultra-sonic range anyway.
Thanks for the prompt reply.
From a technical P.O.V. decreasing the Lag cap will increase gain bandwidth product and increase feedback above the dominant pole frequency. Could this possibly bring subjective benefits?
If you were to decrease the value of the Lag cap and increase the value of the degeneration resistors in the differential pair you could keep the gain-bandwidth product constant. If that brought about the same subjective benefits as just decreasing the lag cap you could rule out the gain-bandwidth product idea.
I think the dual feedback path approach is interesting (like in the leach amp) where feedback at high frequency is taken from the output of the VAS or driver stage. This by-passes any poles formed in the output stage and loading. This at least should improve overall stability especially with difficult loads and therefore enable you to reduce the value of the lag cap.
Sure distortion would rise at high frequencies as feedback decreases around the output stage but this should be in the ultra-sonic range anyway.
Dave,
The dual feedback path indeed has the effect you describe. It also compensates very well for highly capacitive loads, a well worthwhile benefit, without much disturbing 'lifeforce'.
I have tried degeneration whilst reducing lag comp, and this does compensate to a degree, and it does improve sonics marginally.
All this is built into my AKSA amplifiers, and they sound pretty good. It's not rocket science, but it is pushing the limits, since global nfb amps are oscillators waiting to happen and the diversity of speakers in the market makes this eventuality almost inevitable. Further sonic gains are very difficult to come by, and stability is intimately related to rail voltage, so for very powerful amplifiers, the handicap is very real.
The only other options I can think of are SET, SE mosfet and sliding bias SEPP regimes, which need to be power followers for best stability, with modular voltage amplification. The first two are power limited, and the last VERY difficult to design since the stability of the servo control is very tetchy. According to NP the best SEPP amps use loose control over the current sink and set very high quiescents. You can't fool Mother Nature, Nelson opines. I've some experience in this with my successful Glass Harmony, but I need a little more time to verify it.
Cheers,
Hugh
The dual feedback path indeed has the effect you describe. It also compensates very well for highly capacitive loads, a well worthwhile benefit, without much disturbing 'lifeforce'.
I have tried degeneration whilst reducing lag comp, and this does compensate to a degree, and it does improve sonics marginally.
All this is built into my AKSA amplifiers, and they sound pretty good. It's not rocket science, but it is pushing the limits, since global nfb amps are oscillators waiting to happen and the diversity of speakers in the market makes this eventuality almost inevitable. Further sonic gains are very difficult to come by, and stability is intimately related to rail voltage, so for very powerful amplifiers, the handicap is very real.
The only other options I can think of are SET, SE mosfet and sliding bias SEPP regimes, which need to be power followers for best stability, with modular voltage amplification. The first two are power limited, and the last VERY difficult to design since the stability of the servo control is very tetchy. According to NP the best SEPP amps use loose control over the current sink and set very high quiescents. You can't fool Mother Nature, Nelson opines. I've some experience in this with my successful Glass Harmony, but I need a little more time to verify it.
Cheers,
Hugh
AKSA said:
You know i might just do that.....
.....No......seriously....you do have a point about the life-force phenomenon....
Dave said:
Yes....important point here.....sadly an amp. with a stable global feedback loop will sometimes display signs of parasitic oscillation.......this is local parasitic, and is a function of layout, trace and component parasitic reactances...etc.
Most likely to occur at leading and/or trailing edges of transients.
In broad terms, (very broad!), the simpler the design, the less likely this will occur...
Real time observation of THD residual is the key to painless detection...
traderbam said:
ultra-sonic....depending on magnitude, interaction, and location of pcb, and component parasitics....
......audible 'cause results in intermodulation of oscillatory component with audio program......modulation products expressed in audio band....thus effects audible...measurable....
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.