To obtain a distortion profile similar to a asymmetric CFA while using a symmetric CFA input stage namely the diamond, use a current source instead of push pull vas similar to blameless vas. Dont fear, the currents in the upper and lower halves of the diamond can easily be matched. There have been a couple of CFA designs like this used commercially.
Member
Joined 2009
Paid Member
ah, but perhaps the point of using an asymmetric CFA is because the designer actually likes the distortion profile it generates.
I agree, there is much in common between the Singleton CFA input and the Diamond:
http://www.diyaudio.com/forums/solid-state/159780-biguns-botch-up-amplifier.html
I agree, there is much in common between the Singleton CFA input and the Diamond:
http://www.diyaudio.com/forums/solid-state/159780-biguns-botch-up-amplifier.html
Thats my point, using a diamond input you can obtain similar profile by using asymmetric vas with it.
They are afraid that they can not sale their amp 😀.
I agree. I like the amp's sound that have more slew rate and more bandwidth.
Last edited:
😀The only reason you think you like the sound of an amp. with "...more slew rate and more bandwidth" is because you've been told or otherwise informed that it has more slew rate and bandwidth.😀
For CFA using Current-Mirrors --> basic data comparing topologies;
View attachment Current Mirrors -BJT.pdf
Thx-RNMarsh
View attachment Current Mirrors -BJT.pdf
Thx-RNMarsh
Member
Joined 2009
Paid Member
I suspect it's not just Bimo who prefers the sound of more bandwidth - distortion is generally lower when you have high gain-bandwidth in a gnf amplifier even if slew rate per se is not an indicator of sound quality.
its not hard to get amps of any feedback type to be close enough to accurate that level matching, fast switching DBT is the only way to establish "sound"
until and unless some start posting that they have really tried "ears only" Blind testing with controls, multiple amps present at the same time with fast switching its all easily dismissed anecdote - no better or more reliable than the glossy mags florid "reviews"
and of course you then get the “problem” with controlled blind comparisons of amps – they are hard to tell apart – and trivial nulling of frequency response, output Z differences seems to make them indistinguishable “by ear”
until and unless some start posting that they have really tried "ears only" Blind testing with controls, multiple amps present at the same time with fast switching its all easily dismissed anecdote - no better or more reliable than the glossy mags florid "reviews"
and of course you then get the “problem” with controlled blind comparisons of amps – they are hard to tell apart – and trivial nulling of frequency response, output Z differences seems to make them indistinguishable “by ear”
Are we still designing CFA or drifting into audiophile subjective speculation...... ZZZZzzzz.
🙁
-RNM
🙁
-RNM
Can you summarize in what consists the great progress ? Unless I missed something in this thread, I have yet to see investigations on real CFA circuits. Note I am familiar with the complementary pair input concept since the publication of Dick Marsh's preamp published in Audio Amateur circa 1980.
"or audible" requires standards for assessing audibility
how many quotes do you think I could scrape from different posters in this thread seemingly asserting flat open loop bandwidth, slew rate, speed, CFA front end linearity by themselves are audibly superior? - presented as absolute as fact?
why should I accept such statements as in any way reliable, useful, actionable for audio circuit design? - you couldn't post "CFA" here in any thread at diyAudio for years after I joined without some saying CFA sucks for audio - by some assuring us they design high end equipment, they and their discerning clients hating "the sound" from application CFA in power amps or DAC I/V
Last edited:
It is opinion and speculation..... as to why some like the sound performance ...... as we are not doing a subjective review here. I take it as a-- could be true or not. Not needing or looking for 'proof'.
We are looking at the pro-con of CFA type circuits and how it differs for use in audio amps. SIM'ing them to see what distinctions there are or might be. The 'fact' that many people around the planet think such topologies sound better/different/ or, what ever, is reason enough to explore it as designers. Or, just to learn more about other topologies/design. What ever the motivation to participate is not important -- off track -->
We aren't going to debate or offer 'proof' to anyone of how it sounds or the like. Not for this forum/thread. But, that doesnt keep anyone from saying their piece about it. Just that it isnt the main focus or point here. No one is going to try to convince others of anything.... nor cares to try. Unless, it is SIM or measurements or data to figure out all the myths about the circuit topology... which has been 'proven' here and else where to be just that - myths. So... we continue on to the OPS and isolation and other issues towards completion of the topology - doing 'due diligence' along the way. After that, any one can go forward with variations and adaptions and all that good stuff to make their design meet any requirement they think is important to the application. With another tool in their box of design techniques. Hopefully we will do a more complete summary towards that end... we do such every so often already.
Thx-RNMarsh
Last edited:
Current-Mirrors
Did we include this already.... brain fade.
View attachment Current Mirrors -BJT.pdf
Thx-RNMarsh
Did we include this already.... brain fade.
View attachment Current Mirrors -BJT.pdf
Thx-RNMarsh
They're just one datapoint amongst myriads of them, so alone I don't think you should accept them as reliable.
Its the job of the scientist to formulate testable (i.e. falsifiable) hypotheses. What's your testable hypothesis for why Jocko's customers didn't like the sound of a CFB opamp (I seem to recall it was AD846) as I/V?
http://www.delta.dk/imported/senselab/AES125_Tutorial_T4_Perceptual_Audio_Evaluation_Tutorial.pdf outlines a methodology
I think starting with the psychoacoustics literature, quick engineering estimates we get the 1st question "is there an audible difference"
please do put out some numbers with references to psychoacoustic thresholds if you think there are good reasons to think the differences are clearly audible
if the calculated/expected/measured differences between circuit's performance is not clearly above published psychoacoustic thresholds, masking effects, ect. they suggest a informal or pilot study where controls are used with the simple question can A and B be audibly distinguished reliably in DBT
clearly this is more difficult if some take the position that “Unbelievers” won't hear differences – positive controls can be included but it seems lots simpler to just ask “Believers” to do the controlled DBT
acutally just about everything under discussion proabably falls under their flowcart's "expected magnitude of perceptual diffs.>Very small>Too many observtions needed - evaluation not feasable"
I think starting with the psychoacoustics literature, quick engineering estimates we get the 1st question "is there an audible difference"
please do put out some numbers with references to psychoacoustic thresholds if you think there are good reasons to think the differences are clearly audible
if the calculated/expected/measured differences between circuit's performance is not clearly above published psychoacoustic thresholds, masking effects, ect. they suggest a informal or pilot study where controls are used with the simple question can A and B be audibly distinguished reliably in DBT
clearly this is more difficult if some take the position that “Unbelievers” won't hear differences – positive controls can be included but it seems lots simpler to just ask “Believers” to do the controlled DBT
acutally just about everything under discussion proabably falls under their flowcart's "expected magnitude of perceptual diffs.>Very small>Too many observtions needed - evaluation not feasable"
Last edited:
Of course, I understand that only you in the world that can measure an amplifier 😎
My first question in response to this is 'audible to whom?'. ISTM to those customers I referred to, its audible or they'd not reject the product with the AD846 and accept the later, modified product. Or do you have an alternative explanation (i.e. testable hypothesis) for their rejection of the perceived 'CFB sound' ?
What's the problem forr?
There have been some great designs presented here that are easily comparable to VFA.
Again, I just do not understand the antagonism towards CFA on this thread.
Anyway, let's move forward. Where were we?
I think the amount of isolation of the output load from the VAS input was very interesting as just before we had been wondering about the choice of an output stage. B. Cordell early in this suggested a triple OPS. Christopher showed with his circuit example gave an isolation of a -60dB +. Different OPS might be evaluated for its isolation from VAS input stage.
If both signals were sine waves, the -60dB signal from the output would add/subtract and not increase distortion... just a very slight amplitude change which would go unnoticed with a voltmeter.
However, if the output waveform (10v) came back distorted and mixed with the input sine wave, there would be reprocussions - increased or different distortion.
Can this be shown in a SIM? Also if you got the -60-something of isolation to be -100dB then maybe it would measure better (notice I am staying away from pushing that button marked "audible') under real reactive, nonlinear loudspeaker load conditions..... and not just with sine waves and resistive loads. This useful to demonstrate and would be applicable to many topologies.
Thx-RNMarsh
I think the amount of isolation of the output load from the VAS input was very interesting as just before we had been wondering about the choice of an output stage. B. Cordell early in this suggested a triple OPS. Christopher showed with his circuit example gave an isolation of a -60dB +. Different OPS might be evaluated for its isolation from VAS input stage.
If both signals were sine waves, the -60dB signal from the output would add/subtract and not increase distortion... just a very slight amplitude change which would go unnoticed with a voltmeter.
However, if the output waveform (10v) came back distorted and mixed with the input sine wave, there would be reprocussions - increased or different distortion.
Can this be shown in a SIM? Also if you got the -60-something of isolation to be -100dB then maybe it would measure better (notice I am staying away from pushing that button marked "audible') under real reactive, nonlinear loudspeaker load conditions..... and not just with sine waves and resistive loads. This useful to demonstrate and would be applicable to many topologies.
Thx-RNMarsh
Last edited: