CFA Topology Audio Amplifiers

Jump to Latest
#5,381
:cop: previous warning issued http://www.diyaudio.com/forums/solid-state/240712-cfa-topology-audio-amplifiers-215.html#post3794563

This thread is a constant moderation chore. It has been closed on a number of occasions in the past. EVERYONE. Please stick to technical content. If you disagree with something, back it up with technical reasons. Flippant comments designed to rile people up will likely result in bin time. you have ALL been warned.

I'd also suggest that everyone re-reads this post http://www.diyaudio.com/forums/solid-state/240712-cfa-topology-audio-amplifiers-96.html#post3686204
 
#5,382
logician said:
I would have thought that the answer was by now at least somewhat apparent to all or even ubiquitously self evident; although the definite relevance of extremely high slew rates could still be called into question.

Looking at the matter of slew rate in audio amplifiers from my own personal perspective as a professional designer of instrumentation for semi-conductor functional and parametric testing systems then the answer to to the connundrum of why such a preference may exist seems obviously not to lie directly in the harmonic and temporal content of the media itself i.e. music; but moreso perhaps in the many additive and subtractive compound misdoings within the messenger i.e. those circuits and components which perform the amplification thereof.

By this what I mean is that given the need for audio amplifiers to employ at least RF input filtering then; the advantages of a higher internal slew rate are primarily of benefit in the amplifiers correction of itself rather than being of any real relevance to the actual input and output signals per se. Even a cursory examination of the spectral content produced by these internal amplifier circuit non linearities can be seen to give rise to higher frequency components beyond those present at the input.

As it is well known that below a certain threshold of distortion there is almost no correllation whatsoever between percieved sound quality and conventional single frequency THD test results and with the same holding true for IMD and even multi tone burst testing.

My first question is: How realistic is it for us to expect that any single solitary slew rate parameter would be in any way a better indicator?

Therefore, my second question becomes; not what slew rate should be targetted, but: How do we determine what internal slew rate target should be set for any given internal circuit configuration?

P.S. My thanks to all the contributors of this very interesting thread
Click to expand...

:bulb:
Thank you, logician. I have same thought, but because my English is not good I can not describe it as clear as you.
 
#5,384
We have added to the standard descriptors of an amplifier -- THD, Freq response and S/N, and output Watts ---such things as PSRR and Zo and CMRR and others because they do not, in and of themselves, tell us anything about the signal quality but all together we get a better picture of the design. SR adds to that in the same way. I note also that we tend to think of each of these descriptive data as higher or lower is better... But realize there are limits to that as well. So lets continue to make the best possible current-mode of operation amplifier to the max... onward and upward.

Has anyone developed a higher power CFA/CMA amp or has just begun?

THx-RNMarsh
 
Last edited:
#5,385
john curl said:
Many people here do not seem to know that we did EXTENSIVE testing of slew rate differences in amps in the 1970's. Many technical papers were published on the subject. I even spent a month in a lab testing slew rate differences for a paper where we actually wrote a standard for TIM that is now added to many pieces of test equipment today.
We did listening tests, as well as listening tests, even double blind tests by medical staff. Just look up the results, OR listen for yourself. Of course, all else being equal, higher slew rate gives lower TIM.
Click to expand...
Would you like to give us more details or links to these "double blind tests by medical staff"?

I hope they say more than your article on vinyl slew rate which you posted on another forum. That showed less than 6dB more slew than Baxandall and essentially confirmed the Shure data though you used MC cartidges and took into account clicks & pops.

Yes, Otala & his cronies published many technical papers, practically all of which have been shown to be snake oil adverts by Bob Cordell & others. They show nothing that had not already been known by Baxandall & other competent designers from much earlier.
________________

But to get back to CFAs ...

CFAs provide
  • easy zillion V/us slew
  • anecdotal reports that they sound better than VFAs

IF doing a a bandwidth limititaion DBLT with a CFA amp provides a different result from ALL the DBLT on bandwidth limitation carried out so far, it would be the first concrete evidence of the utility of zillion V/us slew and CFAs.

This is about the simplest possible type of DBLT to organise though it is still MUCH more difficult to do than most people realise. Alas, we can't do this over the www cos the number of deaf pseudo Golden Pinnae who will cheat so they may be though true golden pinnae. Besides we are testing amps.
_______________

Here are the FACTS about ALL DBLT on bandwidth limitation carried out in the previous & present Millenium (at least a dozen to my knowledge). Ask Jan Didden who has participated in one this century for a list.
  • In ALL properly conducted DBLTs on Bandwidth Limitation, of those who can tell the difference (ie true golden pinnae), ALL preferred bandwidth limited to between 15k-20kHz. This result is so consistent over more than 3 decades , different amps & speakers, that it cannot be coincidence.
  • ALL the tests also show that ALL those who claim to prefer unlimited BW in sighted tests are unable to reliably repeat their preference blind. ie they are deaf.
These are not anecdotes but Experimental Fact.

But as far as I know (certainly the ones I've been involved with), these tests were conducted using VFAs with good but not zillion V/us slew.

Note also that, unlike the DBLTs on 2nd harm. dist. which, when detectable, divide the true golden pinnae into those who prefer and those who don't, the DBLTs on bandwidth limitation are unanimous.
________________________

In the 60's, the Gold standard for THD was 0.1% from the last of the great valve amps. Today, we know that under certain condition, certain people, on certain types of signal can detect certain types of THD at 0.01%.

It may be that slew has a similar overkill ratio for any given bandwidth.

If the Bandwidth DBLT using CFAs comes out with a different result (ie if those who can tell the difference prefer that CFAs are NOT bandlimited) this makes simple CFAs very attractive as 'rather better than 0.01% THD' and zillion V/us slew is almost the archtype performance of a modern simple CFA.
 
Last edited:
#5,386
Gentlemen, SR math is so simple.
Let's assume 400W/8ohm amplifier. It needs 56.57Vrms output voltage to get the required power, whis is 80Vp (peak, amplitude). 20kHz sine at 80Vp has SR = 10V/us. In case we want to have amplifier 10x faster, to get enough margin in speed, we need SR = 100V/us.

So we have a request for 1.25V/us for 1Vp output voltage. Quite easy, is not it? Regardless the fact that no music signal would contain 20kHz frequency at full swing. We are pretty above any reasonable requirement.
 
#5,387
RNMarsh said:
Has anyone developed a higher power CFA/CMA amp or has just begun?

THx-RNMarsh
Click to expand...

Yeah - I found Accuphase (and all the schema's) 😀

Accuphase Laboratory, Inc.?M-6000
1200W/1R CFA - they basically use Bonsai's NX with a big row
of power MOSFET's , precision CCS's , boosted rails ... other
"embellishments" ...

The E-210A is nearly exactly like my CFA , as is the new 4200 ..
it's a EF3 w/3 pairs of BJT. It even uses Bonsai's FET output
protection switches.

So ... If anyone just wants to buy instead of build ...and hear the
"typical CFA" ... this is the company.
 
#5,390
ostripper said:
Yeah - I found Accuphase (and all the schema's) 😀

Accuphase Laboratory, Inc.?M-6000
1200W/1R CFA - they basically use Bonsai's NX with a big row
of power MOSFET's , precision CCS's , boosted rails ... other
"embellishments" ...

The E-210A is nearly exactly like my CFA , as is the new 4200 ..
it's a EF3 w/3 pairs of BJT. It even uses Bonsai's FET output
protection switches.

So ... If anyone just wants to buy instead of build ...and hear the
"typical CFA" ... this is the company.
Click to expand...

Wrong, they dont use Bonsai s NX, if anything Bonsai uses their M 6000, and then not even close, what about MCS. Neither do they use Bonsai s fet protection switches, its probably the other way round.
Also your CFA is nearly exactly like their E 210 and 4200, not the other way round.

Accuphase amps technology predate your or Bonsai ideas by more than 13 years. :rofl::rofl::rofl:
 
#5,391
manso said:
Wrong, they dont use Bonsai s NX, if anything Bonsai uses their M 6000, and then not even close, what about MCS. Neither do they use Bonsai s fet protection switches, its probably the other way round.
Also your CFA is nearly exactly like their E 210 and 4200, not the other way round.

Accuphase amps technology predate your or Bonsai ideas by more than 13 years. :rofl::rofl::rofl:
Click to expand...

Manso, as you are the one to know CFA technology(at least in audio amps) the best of all here in this forum(my opinion), tell me if any of commercial CFA amps use OIC(Output Inclusive Compensation or pure Cherry). D. Self conclusion is that it is not possible to make stable amp with OIC.
BR Damir
 
#5,392
dadod said:
Manso, as you are the one to know CFA technology(at least in audio amps) the best of all here in this forum(my opinion), tell me if any of commercial CFA amps use OIC(Output Inclusive Compensation or pure Cherry). D. Self conclusion is that it is not possible to make stable amp with OIC.
BR Damir
Click to expand...

Damir, I can honestly not say if it were posible to use those compensation schemes. Either I would have to build such a working amp or someone here could do it. Anyway it seems plausible from the simulations presented. If its D Self s conclusion he may may be wrong, KgrLee says he used it in his designs although those were VFA.
That no one has ever used such schemes in CFA amps Im pretty sure. Not Krell, not Accuphase, Myriad, NAD, Marantz, Sony, well I can probably mention another 25 or so manufacturers which have used CFA design.

Heres an incentive, build your amp and lets find out. 🙂
 
#5,393
PMA said:
Gentlemen, SR math is so simple.
Let's assume 400W/8ohm amplifier. It needs 56.57Vrms output voltage to get the required power, whis is 80Vp (peak, amplitude). 20kHz sine at 80Vp has SR = 10V/us. In case we want to have amplifier 10x faster, to get enough margin in speed, we need SR = 100V/us.

So we have a request for 1.25V/us for 1Vp output voltage. Quite easy, is not it? Regardless the fact that no music signal would contain 20kHz frequency at full swing. We are pretty above any reasonable requirement.
Click to expand...
Yes, exactly, and everything more is sporting ambition.
Is it not, for example, more important to look if CFA has maybe advantages in the stability against VFA?
 
#5,397
You can make two amps that have identical small-signal behavior but that have different slew rates. If you succeed, it should only matter if the amp is being driven into slewing.

However most amps today use a topology that doesn't behave this way, and for them, small-signal behavior will always correlate with slew rate in one way or another.

So I think it is a bit misguided to judge based on slew rate alone. Different topologies will respond differently to slew rate in terms of small-signal behavior. So two amps of different topologies with high slew rate may give different results.

One thing I see almost everyone doing is posting scope shots of amps driven into slewing, and considering this a metric for audio quality. We know that probably none of the amps on this forum are driven into slewing by normal use. It's the small-signal behavior that matters. You can have an amp that gives a perfect square wave while slewing but shows step response aberrations in small-signal.

If CFB really does sound better, we might not know exactly why yet but if we just take the time to look through all the small-signal characteristics affected by an increased slew rate for different topoliges, that may give us a good clue. Also consider the attention in power decoupling and other areas that a high-slew design forces designers to make. If you just applied the high-slew decoupling scheme to a normal amp, what's the effect?
 
#5,398
manso said:
Accuphase amps technology predate your or Bonsai ideas by more than 13 years.
Click to expand...

I/we know many others have been using CFB amp topology and operation for a long time and have been pushing Bob C to include them in his excellent coverage of amp design. It is one of the many interests here in that type of circuit operation and it has been a trip to find out how they work and how they sound.

You might notice that there has been world-wide acceptance of its sound... which appears from reviews to have a characteristic which many people, myself included, know to be more real/accurate to music. IMO. So, here we have a chance to see for ourselves by designing, building and listening. Initially, at least, the results favor the CFB again. It's pretty amazing to me.

So, I recently went out and bought a modest stereo power amp of CFB type -- a Marantz MM7025. Measured and listened. It really starts to sound like real, natural, musical instruments playing. I am going to put it back in and listen longer for greater details.... like reverb trail-off etc. The micro details. Its poorly made, unfortunately.... traces must be .1 mil and a lot of cost cutting which will affect life/reliability.

So, now I want to build a power amp from one of our designers here but at 250W/8 and wondering if any one here has started one of that power level. Otherwise, i'll buy one and tweek it... like an Accuphase or Marantz mono-blocks or? It's really exciting to know this about commercial amps moving in this direction.... even 13 years ago...which is still long after the topology was in an even more advanced state in IC's.


THx-RNMarsh
 
Last edited:
#5,399
keantoken said:
So I think it is a bit misguided to judge based on slew rate alone. It's the small-signal behavior that matters. You can have an amp that gives a perfect square wave while slewing but shows step response aberrations in small-signal.

If CFB really does sound better, we might not know exactly why yet but if we just take the time to look through all the small-signal characteristics affected by an increased slew rate for different topoliges, that may give us a good clue. Also consider the attention in power decoupling and other areas that a high-slew design forces designers to make. If you just applied the high-slew decoupling scheme to a normal amp, what's the effect?
Click to expand...

Hi,

Some attention was needed to cover SRate in the CMode operation. It is a distinguishing feature of the topology. Could it be responsible for its sound? Dont know. But surely, it cannot be thee single number/characteristic alone. I think we all understand this. After flushing this out, a design goal is maybe twice the old standard... 2v/usec. Myself, I like to have more margin so would go 3-4v/usec/peak volt output. But thats just me and my intuition speaking.

The interiour small signal behaviour might give some interesting clues to behaviour/operation and would be a nice direction to learn more about. maybe along the lines of Logician, a few pages back?


THx-RNMarsh
 
Last edited:
You must log in or register to reply here.
Top Bottom