There is a forum on CFA and what they are. This led to the forum to design and compare CFA vs VFA which has designed both and many are in the building phase. Some have listened and given their views on the sound. And then spin-off forums on individual's designs occurred as well. You can make very sophisticated circuits of wide band vs narrow and high feedback vs lower fb. Decide for yourselves.
I have noted that most of the top level models from commercial products are CFA. This has been true for many years now. My own involvement back in the late 70 and early 80's were CFA designs and I have used them ever since because of their wide open loop BW and slew rate and other qualities that I think make for better sound. As far as distortion numbers go, IMO when the circuit is better than -100dB at any practical level and freq then you can move on to improving other parameters like PSRR.
These forums have closed the gap considerably IMO with even no gnfb producing fantastically low distortion. And PSRR has been all but solved with new topologies in CFA. BUT, this level has not filtered down into products you can buy at your local audio-video store yet. So DIY'ers have a chance to hear the best if they do the work.
THx-RNMarsh
I have noted that most of the top level models from commercial products are CFA. This has been true for many years now. My own involvement back in the late 70 and early 80's were CFA designs and I have used them ever since because of their wide open loop BW and slew rate and other qualities that I think make for better sound. As far as distortion numbers go, IMO when the circuit is better than -100dB at any practical level and freq then you can move on to improving other parameters like PSRR.
These forums have closed the gap considerably IMO with even no gnfb producing fantastically low distortion. And PSRR has been all but solved with new topologies in CFA. BUT, this level has not filtered down into products you can buy at your local audio-video store yet. So DIY'ers have a chance to hear the best if they do the work.
THx-RNMarsh
Last edited:
While I agree with most of your above post, Richard, in one respect the DIY has not budged an inch for the last quarter century - it is still mostly dedicated to providing the basic circuitry only.
In other words, they will do a great job of designing the basic audio circuit, but will usually skip altogehther eveyrthing else, like the basic protection circuits against excessive DC at the output, or overheating.
I believe the user should be given the opportunity to build a complete item, not just its key part, all under the assumption that the DIYer will find somebody else's solution for that. And even if he does, how can he be certain that it will successfully blend in?
I build in overvoltage/overcurrent circuits in, and if the user decides that he doesn't want them, then something goes wrong and he fries his speakers, at least I won't be to blame, and nobody can say I could have but just didn't bother to. I build in excessive DC and overheat protection, so again, it's the user's duty to decide, but by the same token, it's his responsability as well.
In other words, they will do a great job of designing the basic audio circuit, but will usually skip altogehther eveyrthing else, like the basic protection circuits against excessive DC at the output, or overheating.
I believe the user should be given the opportunity to build a complete item, not just its key part, all under the assumption that the DIYer will find somebody else's solution for that. And even if he does, how can he be certain that it will successfully blend in?
I build in overvoltage/overcurrent circuits in, and if the user decides that he doesn't want them, then something goes wrong and he fries his speakers, at least I won't be to blame, and nobody can say I could have but just didn't bother to. I build in excessive DC and overheat protection, so again, it's the user's duty to decide, but by the same token, it's his responsability as well.
I see, I'll have a look through the thread later. I don't even know what TIM is.
I believe in the low distortion school then, after all, why else would LME49990 sound better than LME49710?
By the way I like Maserati's too, they look damn good.
I just had an idea, why not make two "The Wire" amplifiers, one with LME49710 and one with LME49990, then blind test them? If this is a viable test, I'll consider this.
I think there is optimum distortion for each amp type. In general any amplifier below 0.1% THD can be said to be hi fi. At that point humans become unreliable judges(always are in truth). I would say when a SE valve amp that might be even 1% THD and is still hi fi with the right distortion spectrum. As Jean Hiraga pointed out the 5 th ( and 7th, 9 th) harmonic is the problem . It should be < - 80dB if possible. Jean also said a high THD distortion curve reducing exponentially might sound less distorted than an amplifier that had only 5 tH harmonic albeit - 80 dB. As speakers seldom go below 1% THD I am reluctant to think distortion THD is what we are listening to. Radford HD 250 claimed zero distortion. I recall it as a bad sounding amplifier. Amplifiers with very low THD often have excellent house keeping circuitry. That possibly is more important, it keeps the coloration of the PSU from seriously affecting the amplifier. Hum ripple the obvious one. As far as I know no one out 7 000 000 people now and the almost equal number that might have lived before could hear THD below 0.1%.
If a blind test was done between a first class SE valve amp and lets say a Chord amplifier I suspect you might be very hard pushed to say which is least distorted. It is almost inevitable you will think the valve amp sounds more like music you heard in real life and thus is less distorted. You may never have that chance or you might try something that can not work together. Alas if that's right. A first class valve amp has a walk in quality that is like 3 D. Most are too romantic which is distortion, those I don't like. I have valve radio if I want that. A well constructed Williamson amp of 1947 is almost without any sound distortion (colouration). It is like a transistor amp with slightly more 3D than most of them, maybe a touch cold in presentation. If the feedback is adjusted it can be made to sound more natural. The Marantz Model 9 is an excellent example of a Williamson inspired design. On the whole the amplifier I like most. Mostly I dislike valve amps. The ones I do like are the one to dream to own. The Mackintosh I mentioned earlier seems to be one. The little Heath EL84 amp also.
there is one school of thought - and one cult
way wrong - read Cordell's papers - on his site - high negative feedback is not correlated with TIM - not "a cause"
compensation, particularly Miller dominant pole and low input diff pair bias are possible problems - but we now degenerate the input pair, run more current than the input noise optimum - trading a small extra current noise with BJT for lots more input stage linearity and current to drive the VAS/ Miller C
its not just Bob, who only debugged Otala's incorrect conflation of GBW and "open loop bandwidth" - and then built hardware to measure TIM in Otala's own terms, and an amplifier with high feedback that measured ultra low TIM, DIM-30 in the noise floor
but early "converts" Walt Jung, Marshall Leach initially embraced Otala's prescription - wrote about it, showed circuits - and then both later wrote that they realized the "high open loop bandwidth, flat loop gain" prescription was not necessary to suppress TIM - that high feedback was possible to design with low TIM
Cherry weighed in at the time - against special interpretation of TIM, just another distortion mechanism that can be designed for
there is no theory supporting Otala's "school" - it is wrong in theory, practice, custom measurements, evidence from AP measurements - which include DIM for a few decades now
there are no studies I've seen that show sub ns "PIM", -100 dB TIM is any more or less audible than "AM" -100 dB IMD products - both of which high loop gain circuits are quite capable of
so no psychoacoustic evidence either
cult members can enjoy their belief system - but I will set the record straight when you keep conflating "high feedback" with "bad TIM" - its bad EE, shows lack of understanding of theory, refusal to face every line of evidence
way wrong - read Cordell's papers - on his site - high negative feedback is not correlated with TIM - not "a cause"
compensation, particularly Miller dominant pole and low input diff pair bias are possible problems - but we now degenerate the input pair, run more current than the input noise optimum - trading a small extra current noise with BJT for lots more input stage linearity and current to drive the VAS/ Miller C
its not just Bob, who only debugged Otala's incorrect conflation of GBW and "open loop bandwidth" - and then built hardware to measure TIM in Otala's own terms, and an amplifier with high feedback that measured ultra low TIM, DIM-30 in the noise floor
but early "converts" Walt Jung, Marshall Leach initially embraced Otala's prescription - wrote about it, showed circuits - and then both later wrote that they realized the "high open loop bandwidth, flat loop gain" prescription was not necessary to suppress TIM - that high feedback was possible to design with low TIM
Cherry weighed in at the time - against special interpretation of TIM, just another distortion mechanism that can be designed for
there is no theory supporting Otala's "school" - it is wrong in theory, practice, custom measurements, evidence from AP measurements - which include DIM for a few decades now
there are no studies I've seen that show sub ns "PIM", -100 dB TIM is any more or less audible than "AM" -100 dB IMD products - both of which high loop gain circuits are quite capable of
so no psychoacoustic evidence either
cult members can enjoy their belief system - but I will set the record straight when you keep conflating "high feedback" with "bad TIM" - its bad EE, shows lack of understanding of theory, refusal to face every line of evidence
Last edited:
TID
https://community.klipsch.com/forums/storage/3/1107724/01162904.pdf
One thing that leads on from my previous statement that I have never tried. If lets say we prefer a SE valve amp why not take a very low distortion transistor amp and add it as a 1:1 buffer. This is fraught with problems, it still might have some value. If the 3D quality is maintained regardless then it is an aberration of distortion. If it is like I suspect a MP3 style presentation then we learn something. The problem is that a valve amp driving the input of another amp is a different load. It still might prove something. Something that maybe I don't want to learn.
https://community.klipsch.com/forums/storage/3/1107724/01162904.pdf
One thing that leads on from my previous statement that I have never tried. If lets say we prefer a SE valve amp why not take a very low distortion transistor amp and add it as a 1:1 buffer. This is fraught with problems, it still might have some value. If the 3D quality is maintained regardless then it is an aberration of distortion. If it is like I suspect a MP3 style presentation then we learn something. The problem is that a valve amp driving the input of another amp is a different load. It still might prove something. Something that maybe I don't want to learn.
With regard to the above text, I really wonder who's the cultist here - but never mind that now.
I think I made it quite clear that TIM in the sense Otala showed it was more or less a thing of the past. But do refresh your memory, take a look at how amps were made until that text was published, and compare it with how similar price class products are made. About 1 in a 100 had degenerated input differential amps, today it's really hard to find one like that, that's using BJTs, FETs are another matter.
In those days, many Miller compensation caps were way out of any proportion to the currents feeding them, after that, meaning now, this is a rare case indeed.
If you weren't such a hard believer in Bob, you might pause to ask yourself a few rather obvious questions, like what's harder, to come up with a completely new theory, or to "debunk" it? If he was so smart, why didn't he come up with his work before? Why is he leaning heavily on the proving of another wrong rather than proving himself to be innovative?
Ultimately, you said it yourself - "we now". Fine, on basis of whose work "we now" ... ?
As far as I am concerned, and that's all I will say because I really don't want to get into another pointless debate on whose is bigger, Otala may not have been right all the way, we know he wasn't, but he did made a heck of alot of people sit down and start thinking along somewhat different lines, treading paths they didn't tread before.
Regarding Bob Cordell, I already said here and I'll say it again - my hat off to him for his book on power amplifiers, that undertaking alone deserves much praise. For the rest of his work, which I have not read yet, I expect will be well worth reading and considering, which I intend to do as soon as I get the builders out of my home, which can't be soon enough.
That's all I have to say on this matter. You have the right to believe in whatever you choose to belive, but please do be civil enough to extend the same courtesy to those with a different opinion, without comments on what they do or do not know.
Last edited:
TIM can be alleviated without changing loop gain, therefor feedback is not the cause.
There. I fixed it
Jan
That is a statement I can easily agree with. Especially since I did not say it DOES cause TIM, I said that it COULD cause TIM. Before the problem was known and alleviated.
Jan, you of all people must have a version of the original IEEE paper somewhere in electronic form, please post a link so that the fedayeen Cordellians can actually read what they are blasting.
I actually do have Otala's 1st "TIM" paper, I am a EE, not a Cordell, Cherry "fanboy" - I have argued Bob around to my view of some EE issues, still differ in places
I can be so "harsh" precisely because "TIM" has a EE Signal Theory description - no hearing acuity needed - only ability to follow circuit theory, math, logical arguments, measurements
as EE knowledge is "constructed" Otala's prescription as the only path to low TIM is as dead as any idea can get
I can be so "harsh" precisely because "TIM" has a EE Signal Theory description - no hearing acuity needed - only ability to follow circuit theory, math, logical arguments, measurements
as EE knowledge is "constructed" Otala's prescription as the only path to low TIM is as dead as any idea can get
Last edited:
Transitory Intermodulation Distortion, as it is properly called, was not discovered by Otala and Curl. This is one of the greater fallacies still making rounds on the internet. Slew rate limiting, was first described during the 1920's but that that time the cause was not well understood. It was first discussed by Roddam in the 1950's when it was assigned the name TID after the cause was isolated to compensation. Otala and Curl seemed to believe that it was caused by NFB alone, and that higher negative feedback induced more of their elusive distortion. Upon reading the original texts, one will find mention that the proposed TIM was caused by the slowness of the feedback loop. This in itself contradicts the law of drift velocity and is pure pseudo science. It could be said that the Otala TIM paper was a case of poor semantics, but in any case, it was a case of poor mathematics. The paper never should have been published, due to several flaws in the methods and analysis. However, elsewhere, such as in Japan during the 1960's, it was already understood to be derived from the miller stability compensation pole.
In other words, it had nothing to do with the degree of feedback at all. Even so, papers in the audio press continued to be published wherein the authors were blindly convinced that NFB was at fault. In light of this, any intuitive measuring will readily highlight the cause and where it does and does not exist. Today, the accepted term among regulated engineering bodies is transitory intermodulation distortion, ie TID, and pertains to radio frequency amplifier design. The effect is easily demonstrated with mathematics modeling and a scope. I was recently reading a lengthy modern article on the subject of TID and SID by an licensed engineer, and it was enlightening to see that TIM has be put to rest in the scientific community.
TIM cannot even be sourced in most peer reviewed journals and libraries, and the theory is not accepted for good reason. No properly designed modern amplifier can exhibit TIM/TIM in the absence of slew-rate induced distortions, SID, since it is the slew limiting that causes the asymmetrical tapering of the 90-180 and 270-360 quadrants of the waveforms that supports the intermodulation of higher frequencies on the lower frequency fast-rising carrier. People who still believe in TIM existing in all amplifiers with a sufficiently high Fo have failed to update their credentials with current education on modern engineering principles. For a properly designed modern amplifier, TID could only exist at radio frequencies. Today, the term TIM only exists in audio marketing.
In other words, it had nothing to do with the degree of feedback at all. Even so, papers in the audio press continued to be published wherein the authors were blindly convinced that NFB was at fault. In light of this, any intuitive measuring will readily highlight the cause and where it does and does not exist. Today, the accepted term among regulated engineering bodies is transitory intermodulation distortion, ie TID, and pertains to radio frequency amplifier design. The effect is easily demonstrated with mathematics modeling and a scope. I was recently reading a lengthy modern article on the subject of TID and SID by an licensed engineer, and it was enlightening to see that TIM has be put to rest in the scientific community.
TIM cannot even be sourced in most peer reviewed journals and libraries, and the theory is not accepted for good reason. No properly designed modern amplifier can exhibit TIM/TIM in the absence of slew-rate induced distortions, SID, since it is the slew limiting that causes the asymmetrical tapering of the 90-180 and 270-360 quadrants of the waveforms that supports the intermodulation of higher frequencies on the lower frequency fast-rising carrier. People who still believe in TIM existing in all amplifiers with a sufficiently high Fo have failed to update their credentials with current education on modern engineering principles. For a properly designed modern amplifier, TID could only exist at radio frequencies. Today, the term TIM only exists in audio marketing.
Last edited:
I'll take a look, but I think Bob has also some stuff on his website.
But I have no illusion that people who have a different view will actually read those papers. They know where they are, and don't need me to point them out.
It's hard to explain something to someone when his feeling-well depends on not knowing it.
But hey, why not: http://www.cordellaudio.com/papers/another_view_of_tim.shtml
http://www.linearaudio.net/images/onlinearticlesPDF/volume1bp.pdf
About 'letting everyone have their opinion'. Fine with me, but let them then say 'this is my opinion only'.
Too many times people present their opinion as 'THE TRVTH (tm)'.
It's pretty stupid in my book to continue to stubbornly say 'this is how I see it', without any reasoned back up, in the face of facts that show otherwise.
Jan
But I have no illusion that people who have a different view will actually read those papers. They know where they are, and don't need me to point them out.
It's hard to explain something to someone when his feeling-well depends on not knowing it.
But hey, why not: http://www.cordellaudio.com/papers/another_view_of_tim.shtml
http://www.linearaudio.net/images/onlinearticlesPDF/volume1bp.pdf
About 'letting everyone have their opinion'. Fine with me, but let them then say 'this is my opinion only'.
Too many times people present their opinion as 'THE TRVTH (tm)'.
It's pretty stupid in my book to continue to stubbornly say 'this is how I see it', without any reasoned back up, in the face of facts that show otherwise.
Jan
Last edited:
DVV:
I'm not an ideologue on the TIM issue. I really focus on what I find works and the huge gap between reported observations and reliably testable, perceivable errors is a huge stumbling block. TIM is an interesting and manageable issue today and something that not many bother to test for anymore.
There is a pretty short list of perceived acoustic sensitivities that must be addressed. Frequency response (an A-B's between a SET and a solid state amplifier will violate this if the tube amp does not have a low output impedance). Distortion above .1 %, getting all distortion products below -80 dB is not difficult today. Noise, really insidious and hard to deal with in a system. Time stability (wow, flutter jitter etc.). Get all of those below known academically accepted thresholds and then start working on the more obscure stuff.
The high bandwidth, low distortion, low noise stuff I showed you are efforts to address these issues. Getting them right will most likely get many of the other issues right as well.
I'm not an ideologue on the TIM issue. I really focus on what I find works and the huge gap between reported observations and reliably testable, perceivable errors is a huge stumbling block. TIM is an interesting and manageable issue today and something that not many bother to test for anymore.
There is a pretty short list of perceived acoustic sensitivities that must be addressed. Frequency response (an A-B's between a SET and a solid state amplifier will violate this if the tube amp does not have a low output impedance). Distortion above .1 %, getting all distortion products below -80 dB is not difficult today. Noise, really insidious and hard to deal with in a system. Time stability (wow, flutter jitter etc.). Get all of those below known academically accepted thresholds and then start working on the more obscure stuff.
The high bandwidth, low distortion, low noise stuff I showed you are efforts to address these issues. Getting them right will most likely get many of the other issues right as well.
there is no other leg to stand on though
in the ~40 years since Otala's 1st TIM I have seen no controlled listening tests showing TIM is orders of magnitude more audible than "AM" IMD product - as would be required for flat open loop gain to be a good idea on TIM grounds
and then only as a patch with a underbiased, undegenerated bjt diff pair dominant pole strawman "high feedback" amp
"TIM" is just a subset of the usually measured IMD - the part with a 90 degree phase shift between the upper and lower products
"TIM" appears to have no explanatory power, never the less it is in fact avoided in modern quality audio amp design - the ppm distortion crowd's amps don't have % IMD products suddenly popping out short of clipping
in the ~40 years since Otala's 1st TIM I have seen no controlled listening tests showing TIM is orders of magnitude more audible than "AM" IMD product - as would be required for flat open loop gain to be a good idea on TIM grounds
and then only as a patch with a underbiased, undegenerated bjt diff pair dominant pole strawman "high feedback" amp
"TIM" is just a subset of the usually measured IMD - the part with a 90 degree phase shift between the upper and lower products
"TIM" appears to have no explanatory power, never the less it is in fact avoided in modern quality audio amp design - the ppm distortion crowd's amps don't have % IMD products suddenly popping out short of clipping
Then please consider this, and understand that this my opinion and experience only.
Of all the amplifiers I have ever heard in my own room, in my own system, for a period no shorter than 14 days, purely statistically speaking, I have shown a clear preference to the tune of approximately 10:1 for amplifiers resigned to have a relatively wide open loop bandwidth and relatively smaller global NFB factors.
Of those amps I own, 7 (3 from H/K, 2 from Marantz, and 1 each from Philips and Sansui) are leaning towards that idea (though not all equally so), and only one (Karan Acoustics KA-i180) is a high global NFB design. That is the only one I did not buy, but got as a present, for which I am grateful because I think it sounds really good.
With a 7:1 ratio, I'd say I have a certain preference, no matter who writes what.
I think before TID was thought up there was a mystery. Why did so many amplifiers sound awful. A revision of TID seem to support it rather than dismiss it. Mostly we get into how many angles can fit on a pin head type arguments and the wheel continuously reinvented. Angles on pin heads was apparently a real intellectual debate. Times change, do people ?
It is highly doubtful that the Quad 405 can suffer any form of TID. It is a bland and a nothing special type of amplifier. Apparently this is nothing to do with feed-forward error correction and just very low current delivery. A parallel bridged 405 II starts to correct that. Highly recommended for Gale 401 speakers.
I have often speculated that a 405 with MOS FET outputs would work well. The gradual turn on might be very good. I would like to make it with a proper class A stage to replace the Quad economy version. If MOS FET I would remove it's protection current limiters. If correctly set up is is very nearly a zero distortion amp. I think I read Quad were 20 db away from correct set up. They did it to produce an amp no worse than 303 which was quicker and cheaper to build. The feed-forward was discredited because they did not offer the best it could do. In all fairness 2 ready built modules were £50. They could be fitted and running in 10 minutes.
It is highly doubtful that the Quad 405 can suffer any form of TID. It is a bland and a nothing special type of amplifier. Apparently this is nothing to do with feed-forward error correction and just very low current delivery. A parallel bridged 405 II starts to correct that. Highly recommended for Gale 401 speakers.
I have often speculated that a 405 with MOS FET outputs would work well. The gradual turn on might be very good. I would like to make it with a proper class A stage to replace the Quad economy version. If MOS FET I would remove it's protection current limiters. If correctly set up is is very nearly a zero distortion amp. I think I read Quad were 20 db away from correct set up. They did it to produce an amp no worse than 303 which was quicker and cheaper to build. The feed-forward was discredited because they did not offer the best it could do. In all fairness 2 ready built modules were £50. They could be fitted and running in 10 minutes.
Yes, you did show me, and I thank you for that. You and I seem to be in good sync, since I learnt from you a lot in a very short time, and that really, deeply pleases me. I love to learn.
We discussed this, and I agree with what you said above. So long as that 0.1% is related to the worst case, or some personally set limit of an amp. Although, truth be told, I prefer it to show more THD then clipping.
I especially agree with your last sentence - it's not just this or that, it's how it works as a whole that counts.
And I don't mind admitting I have been bushwacked a few times by amps which, looking at their overall topology and the parts used, had absolutely no right to sound that good - yet it did. Initially, I would never have thought it possible, but the fact it did way better than expected made me sit down and examine my own expectations. Sure enough, when you don't expect but actually look and think, the explanation will come to you.
I am anything but an ideologue of TIM, as I have followed that particular issue somewhat, so I know that in its original form it's a thing of the past. However, the funadmental logic of Otala/Lohstroh's amp did, and still does appeal to me as the right way to go about it, and I refer to myself only, of course. Add to this my most positive feelings about H/K products, all of which are also made on basis of that logic, and it's easy to see why I feel that way.
You know, I never claimed, nor did I ever hear/read you claim that Otala was 100% right and is a god walking among us mere mortals.
Whoever proposes something truly new will by default be not so right and quite wrong in some aspects of his work in its early days. Heck, Columbus expected India and he got America, which was not even named after him, but after Amerigo Vespucci.
Even then, there were many who wanted to burn him at the stake. None of the would be burners stopped to think: but what if he's right, even for the wrong reasons? Would we be having this debate if he had not done it?
No matter what anyone says or writes, it is a fact which is proved easily by simply looking over the schematics of the models of the day, that right or wrong, those IEEE papers forever changed the way we do amps, because at the very least, they tell us no matter what we do, we at least should know what not to do.
At least Otala and Lohstroh are here to know they are being called names. Helmholtz was not nearly as lucky, Peter Walker made the Quad electrostatic speaker exactly 100 years after Helmholtz put forth the theory of his resonator.
Of course, the worst deal ih history was handed over to that bearded Greek philospoher, who basked his feet in the waters of the Aegean in the port of Ephesus over 2,000 years ago, and postulated that everything is made up of atoms. He must have grown an extra long beard even in Heaven waiting for science to prove him right.
Last edited:
I took the trouble to look him up and talk to him:
http://www.linearaudio.nl/linearaudio.nl/images/pdf/Jan_Lohstroh_interview(1).pdf
Jan
- Status
- Not open for further replies.