I can't help wondering whether the compensation schemes used in any amp have more effect than the underlying topology. You can over compensate and end up strangling the open loop gain which would destroy performance of both VFAs and CFAs.
PS. It turns out you can compensate a CFB with TMC + some lead lag (as per edmonds scheme in his PMP/MCP amps).
Edmond - that explains that classic big V8 sound. And that manifold is a thing of beauty!
PS. It turns out you can compensate a CFB with TMC + some lead lag (as per edmonds scheme in his PMP/MCP amps).
Edmond - that explains that classic big V8 sound. And that manifold is a thing of beauty!
Jan,
Please show me where there is a scientific test that demonstrates that an amplifier that can deliver 1 ppm at 1KHz (a supposed VFA) rather than 10 ppm ( a supposed CFA) sounds better.
There is none and your proposition therefore does no stand up to scrutiny.
Humans cannot even detect the difference between 0.1 % and 0.05% in controlled tests. So, I fairly to see what difference 1ppm or 5ppm will make.
Yes my language was a bit sloppy.
I should have said that the lower the non-linearity, the more transparent and neutral the amplifier is.
'Sounds better' is a very personal/preference kind of opinion and has not much value in trying to design the most transparent and neutral amplifier.
jan
jan
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?
No one can distinguish between amplifiers of whatever topology in a controlled listening test if they all generate, say, less than 1000ppm distortion across the audio band before clipping.
Indeed, Cherry in his Electronics World articles suggests that no one can hear 33% third harmonic distortion at 15KHz. (Send me mail and I'll send you the pdfs)
However, so-called "CFAs" are inferior to VFAs in audio applications for sound technical reasons.
Whether it is Cherry or anyone else --- one very simple but important element in the person's threshold of detection is the masking of the source of sound (speaker) distortion. You might guess that your threshold would be lower if the speakers distortion was lower.
One advantage to having personal access to test equipment is to listen to and creat various types and levels of distortion.... like what does only odd or even harmonics sound like and what are my own thresholds. What does just 2H sound like compared to only 3H Etc. All sorts of things can be learned instead of relying on old data or someone else's conclusions.
In my case, I use electrostatic speakers/headphones of very low distortion. As a blanket, unqualified, statement - i can detect somewhere between .01 and .1% harmonic distortion. So that becomes my bench-mark for design. Now IF I want to be sure myself and no one else might be able to detect such distortion, then i design for 10 times better or .001 (-100dB).
But having said that, it really is the entire chain of gear we are listeing to...and furthermore it is the entire recording and playback - every stage the siignal went thru... not just one piece of gear/amplifying stage... that we are listening to. So, with that in mind, we need very, very low distortion electronics to have an entire recording/playback "system" that sounds transparent or real and less like Hi-Fi.
Thx-RNMarsh
One advantage to having personal access to test equipment is to listen to and creat various types and levels of distortion.... like what does only odd or even harmonics sound like and what are my own thresholds. What does just 2H sound like compared to only 3H Etc. All sorts of things can be learned instead of relying on old data or someone else's conclusions.
In my case, I use electrostatic speakers/headphones of very low distortion. As a blanket, unqualified, statement - i can detect somewhere between .01 and .1% harmonic distortion. So that becomes my bench-mark for design. Now IF I want to be sure myself and no one else might be able to detect such distortion, then i design for 10 times better or .001 (-100dB).
But having said that, it really is the entire chain of gear we are listeing to...and furthermore it is the entire recording and playback - every stage the siignal went thru... not just one piece of gear/amplifying stage... that we are listening to. So, with that in mind, we need very, very low distortion electronics to have an entire recording/playback "system" that sounds transparent or real and less like Hi-Fi.
Thx-RNMarsh
Last edited:
Yep, Sub-ppm at 20K from the bare theoretical schematic ensures that in the band where it really counts (800Hz - 3KHz) all TIM, IM and whatever more products are totally stomped into the noise floor because the openloop gain from the VFA is exceedingly higher than any CFA in this audio band of interest. Thanks JCX
I never had any preference for CFB topology per se. I only noticed through the years that some amps sound much better than the others and only very late found out that as rule those amps are CFB. I am not prejudiced about specific topology, all I want is good music reproduction. I became interested in diy electronics because it's impossible to buy so much finished products to examine differences in topology, it's much cheaper to build relatively simple amps and try to find some answers. If it wasn't for this site I would remain in doubt if my perception about CFB is right. But it became obvious that I am not alone in this respect. And experiences of other people are incredibly similar to my own.
If you ask me to define in as few words as possible the advantages of CFB, I would say: sound is free from some unknown (to me) form of distortion that causes relatively quick listening fatigue present in all VFB amps. (I noticed that some are deliberately overcompensated to mask this annoying property) VFB amps are not transparent enough to me even when THD is vanishingly small. I believe that the ear is by far more reliable and relevant a measuring instrument than any lab equipment. You may think what you want but I once reported to Ben Duncan my experience with mains fuses - I can hear the difference in sound when the same mains fuse is inserted in fuse holder in opposite direction (the reason to avoid them). I can not understand why is human ear so much degraded in relevance by the people that are supposed to design audio amps. Where is this exclusive fascination with lab instruments coming from?
That this is total nonsense was scientifically demonstrated a long time ago.
More total nonsense. The current through a mains fuse is AC, so a fuse can no more be directional than the sun be made of cheese.
To try and get to the heart of the matter I looked up some material from Analog Devices, who might just know a thing or two about such things,
. Jung's "Op Amp Applications Handbook" directly compares and contrasts the two topologies and states:Seems pretty straighforward to me ...
Yes, again that 'magic' range between 60 and 80dB down pops its head up, I've seen that to be the dividing area so many times ...
People often don't appreciate how attenuated 60dB actually is - everyone should listen to a track at normal volume, 0dB, and the same again at -60dB level - a few might get a shock ...
...and yet, virtually every audiophile forum on the web is full of the oh-so-opinionated "trust you ears" types.
The end goal is to create something that you think sounds good to yourself. however, one man's gold is another's poison.
Each person is different. Although I can't see why you can't match these "sound qualities" to measurements. It must be possible to attach measurements to these qualities. These measurements must not be limited to THD alone. You should be able to predict the style of the sound with a given of measurement results.
Each person is different. Although I can't see why you can't match these "sound qualities" to measurements. It must be possible to attach measurements to these qualities. These measurements must not be limited to THD alone. You should be able to predict the style of the sound with a given of measurement results.
mcd99uk,
The problem is that we always seem to have these test protocols that are for individual components, not the entire chain. The testing that is done at the end of the chain, after the speakers is not as exacting as it is with only the electrical components alone. Yes we can check for FR and Phase and group delay and other similar tests. But is this enough in the way of testing or are we missing things at this critical final step in the testing methodology? I think that we probably are, but until we find those missing tests requirements we hopelessly will fight over what is it that stops us from that perfect reproduction that we are so interested in finding.
The problem is that we always seem to have these test protocols that are for individual components, not the entire chain. The testing that is done at the end of the chain, after the speakers is not as exacting as it is with only the electrical components alone. Yes we can check for FR and Phase and group delay and other similar tests. But is this enough in the way of testing or are we missing things at this critical final step in the testing methodology? I think that we probably are, but until we find those missing tests requirements we hopelessly will fight over what is it that stops us from that perfect reproduction that we are so interested in finding.
Interesting how the thread still manages and how admins are still allowing this, after so many years of fruitless...
That's it and that's all.
We need no more CFA vs VFA. Both are great topologies and both will exist forever. That Mr. Self does/doesn't add a particular topic to his book matters little when one has found his ideal topology, by himself, through trying, buying, making, measuring and listening... knowing more about his likes/hates, knowing more about himself.
Let's be at peace. We are billions, we are different, but we're still human, we're still building and buying amps, listening music, and loving our lovely amps for the pleasure they bring.
Yes, again that 'magic' range between 60 and 80dB down pops its head up, I've seen that to be the dividing area so many times ...
People often don't appreciate how attenuated 60dB actually is - everyone should listen to a track at normal volume, 0dB, and the same again at -60dB level - a few might get a shock ...
Which is even higher if you are trying to detect harmonics thru speakers with much higher distortion. But its a worse case threshold for my ears/system.
Same for freqs.... put on a signal generator at just 8Khz...... its really really high compared to most of the music freqs you listen to.
The most important issue with distortion is not just a single amp/tone threshold. But the total System distortion. If you think .1% is a threshold or higher.... and every piece of gear the signal went thru from mic preamp to speaker input added an OK, 0.1%... it would sound like Hi-Fi we have now by the time the signal/music got to your ears at a very-unquestional- much higher level than threshold.
If you ever want to start hearing msuic uncorrupted and truely transparent.... you have to start with extreamly low distortion all the way thru each stage. Until that is realised.... its -> Just Rock N Roll. But I like it. (hi-Fi).
Thx-RNMarsh
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.