good for you OS, so now we can go on building that amp of yours with options for VFA or CFA on the same platform for an easy A/B comparison....
at least this thread bore fruit after all...
good for you OS, so now we can go on building that amp of yours with options for VFA or CFA on the same platform for an easy A/B comparison....
at least this thread bore fruit after all...
I did not mean that in a sarcastic way , Bonsai's site material was fantastic.
The "controversy" was just the push I needed to get the CFA right.
The "hairpulling" of trying to get certain VFA performances out of the CFA
created a "feedback loop" to even greater VFA performance
. (If anything , the CFA exercise was a "workout" )
Both are easy now ....
OS
I did not mean that in a sarcastic way , Bonsai's site material was fantastic.
The "controversy" was just the push I needed to get the CFA right.
The "hairpulling" of trying to get certain VFA performances out of the CFA
created a "feedback loop" to even greater VFA performance
(If anything , the CFA exercise was a "workout" )
Both are easy now ....
OS
Better understanding of each topology is a great result! I am looking forward to the results of listening evaluations by many DIY'er who will build the designs shown here and on individual sites that extent those designs.
Such evaluations have already been done on consumer audio equipment for decades with both topologies,... CFA and VFA.
Such consensus has been building which is why we now have two camps and this subject being discussed. Now with CFA designs and learning to over-come weaknesses is all to the good for making better products and for DIY'ers.
BUT --- I am not sure everyone is clear about why IC makers and Industry as a whole even have separate types (CFA and VFA) in their design bag of tricks. So, something is still missing and we dont get it, yet. Or do we? Can this be explained here?
Thx-RNMarsh
Last edited:
I have gone back into the loop gain models I posted earlier
1. Input stage bias current on CFA set to 560uA; VFA LTP current left at 10mA as per the original and the CFA input stage stand off resistors reduced from 15 Ohms to 1 Ohm. This was done as part of the process to reduce the CFA TIS current detailed in (2) below
2. VAS/TIS current on both models set to 20mA (actual difference is 100uA higher on VFA) - too much fiddling needed to get them exactly the same; the VFA VAS current was 12mA - a little low it could be argued for an apples to apples comparison. The CFA TIS degen resistors were increased from 22 Ohms to 33 Ohms - again, so that the 2nd stage current could be aligned.
3. Output stage bias corrected to 75mA and 77mA - close enough for our purposes and not necessarily targeting lowest distortion since I am not considering re'
4. Verified that all transistor models used between the two are exactly the same (I just copied the VFA devices across to the CFA in the circuit)
VFA comp'd for 1 MHz ULGF (MC Cdom set to 125 pF)
VFA phase margin is 85 degrees
CFA ULGF is 2.2 MHz - same PM (delta in PM at 1 MHz is 1.5 degrees).
The additional LG at HF on the CFA is 7dB
I'll defer to jcx's succint explanation that the delta in PM is due to the HF poles in the small signal section in the VFA - the LTP and mirror. If you look at the way the loop gain drops off on the VFA beyond the ULGF, and the phase margin decrease, this makes perfect sense. The open loop plot is just as revealing, where the UG for the VFA is now 9.2 MHz and the CFA 13.4 MHz. But, the phase on the VFA has already dropped off considerably - the difference at OL UG being ~90 degrees - i.e. huge. - in favor of the CFA
If you look at the plots (including the ones I posted the other day), this latest result does not materially change the outcome and the conclusions, namely, that there is more gain and phase margin in a CFA that allows the loop to be closed at higher frequencies, or, if it floats your boat, closed at VFA like frequencies, but with more GM and PM.
None of the above is an endorsement of one over the other - I like 'em both.
1. Input stage bias current on CFA set to 560uA; VFA LTP current left at 10mA as per the original and the CFA input stage stand off resistors reduced from 15 Ohms to 1 Ohm. This was done as part of the process to reduce the CFA TIS current detailed in (2) below
2. VAS/TIS current on both models set to 20mA (actual difference is 100uA higher on VFA) - too much fiddling needed to get them exactly the same; the VFA VAS current was 12mA - a little low it could be argued for an apples to apples comparison. The CFA TIS degen resistors were increased from 22 Ohms to 33 Ohms - again, so that the 2nd stage current could be aligned.
3. Output stage bias corrected to 75mA and 77mA - close enough for our purposes and not necessarily targeting lowest distortion since I am not considering re'
4. Verified that all transistor models used between the two are exactly the same (I just copied the VFA devices across to the CFA in the circuit)
VFA comp'd for 1 MHz ULGF (MC Cdom set to 125 pF)
VFA phase margin is 85 degrees
CFA ULGF is 2.2 MHz - same PM (delta in PM at 1 MHz is 1.5 degrees).
The additional LG at HF on the CFA is 7dB
I'll defer to jcx's succint explanation that the delta in PM is due to the HF poles in the small signal section in the VFA - the LTP and mirror. If you look at the way the loop gain drops off on the VFA beyond the ULGF, and the phase margin decrease, this makes perfect sense. The open loop plot is just as revealing, where the UG for the VFA is now 9.2 MHz and the CFA 13.4 MHz. But, the phase on the VFA has already dropped off considerably - the difference at OL UG being ~90 degrees - i.e. huge. - in favor of the CFA
If you look at the plots (including the ones I posted the other day), this latest result does not materially change the outcome and the conclusions, namely, that there is more gain and phase margin in a CFA that allows the loop to be closed at higher frequencies, or, if it floats your boat, closed at VFA like frequencies, but with more GM and PM.
None of the above is an endorsement of one over the other - I like 'em both.
Last edited:
In the beginning ... there was "light"
.. actually the - http://www.ti.com/lit/ds/symlink/clc103.pdfCLC103.
They invented these for video/RF. VFA's are for audio/servo's
TI says, "VFA's for precision and general purpose applications" and ...
...." Although CFAs do not have the precision of their VFA counterparts, they are preciseenough to be dc-coupled in video applications where dynamic range requirements are not severe."
OS
RNMarsh (3986)
BUT --- I am not sure everyone is clear about why IC makers and Industry as a whole even have separate types (CFA and VFA) in their design bag of tricks. So, something is still missing and we dont get it, yet. Or do we? Can this be explained here?
To take just one company that produced both types of amplifiers Accuphase
CFA: E-210, E-212, E-406, P350, P-550, P-700
VFA: A-50, E-202, E-260, E-303, E-305, E-406, M-60, P-266, P-300, P-360, P-500, M-1000EN
As you can see only the third CFA.
If CFA had a clear advantage, they would produce more than VFA, or even would have given up production VFA
best regards
Petr
BUT --- I am not sure everyone is clear about why IC makers and Industry as a whole even have separate types (CFA and VFA) in their design bag of tricks. So, something is still missing and we dont get it, yet. Or do we? Can this be explained here?
To take just one company that produced both types of amplifiers Accuphase
CFA: E-210, E-212, E-406, P350, P-550, P-700
VFA: A-50, E-202, E-260, E-303, E-305, E-406, M-60, P-266, P-300, P-360, P-500, M-1000EN
As you can see only the third CFA.
If CFA had a clear advantage, they would produce more than VFA, or even would have given up production VFA
best regards
Petr
It's very clear in my view. High loop gain precision go for VFA. High speed wide band go for CFA
Recent advances in processes and circuit techniques mean the two overlap more than they did 10 or 15 years ago, but each still has it's place in the engineers toolbox IMV. And, sales 100s of millions of $ a year in CFA at healthy margins is justication enough. That's marketing for ya
Recent advances in processes and circuit techniques mean the two overlap more than they did 10 or 15 years ago, but each still has it's place in the engineers toolbox IMV. And, sales 100s of millions of $ a year in CFA at healthy margins is justication enough. That's marketing for ya
Petr, what is 'clear advantage' when it is about marketing ? Harmonic distortion numbers at 1KHz you can print in the performance chapter ? Subjective listening comments in specialized hifi magazines, often flawed because the company witch product the amp on the bench table is an important advertiser ?
I a was courageous and crazy enough to begin a Hifi company, i'm not sure what will be our first power amp. The one witch will generate the highest margin and sales.
Thinking to this, i believe i would try to offer the two topologies as an option in the same amp.
Obvious in my view too
The VFA Accuphase models you listed are all out of production.
All their current models are CFA.
So yes, Accuphase has "given up" on VFA.
