Hi Petr,
I'm the same guy. In my publications before the mid 2000's I used Robert Cordell, as many of my publications were in connection with my professional career in things like integrated circuits and fiber optic systems. Same for my audio publications in those earlier years, such as in JAES. Now I use the less-formal "Bob" for all of my audio endeavors.
Cheers,
Bob
This has been a great place to understand what a CFA circuit looks like and learn its pro's and con's. I/we learned that the distortion can be as low as VFA and other areas could be improved as well... such as PSRR. In this understanding has come better understanding and improvements in VFA as well, as a byproduct.
THx-RNMarsh
THx-RNMarsh
Richard,
This has indeed been a good thread. As with many threads, it has touched on related topics, and even strayed a bit, but we have learned from those adventures as well. It has also been a fast-moving thread, and I must admit that I have at times had trouble keeping up with it 🙂.
Cheers,
Bob
I hope very much that this means Guru Cordell has been persuaded to do more than damn CFAs with faint praise in the next edition of his book 😎
A very good thread, indeed. Some kind of a reference. I hope it will help CFAs to get out of the ghetto where this topology full of qualities seemed to be. And tease some readers to try to build and *listen*, because ... it is all about.
Last edited:
Now that this thread is finally slowing down, can we recap and summarize for those not wishing to read over 600 pages? Some general questions:
Where is the sweet spot for fully complementary bi-polar pairs (ie: which parts)?
If we were to characterize the output attribute of CFA designs, we would assign which percentages to:
1. soundstage
2. timbral accuracy
3. PRAT
Are CFA amplifiers generally cheaper to produce than other topologies?
Those are my recap question suggestions and if there is a consensus on any of them or if you have better ones to ask, please reply here.
Where is the sweet spot for fully complementary bi-polar pairs (ie: which parts)?
If we were to characterize the output attribute of CFA designs, we would assign which percentages to:
1. soundstage
2. timbral accuracy
3. PRAT
Are CFA amplifiers generally cheaper to produce than other topologies?
Those are my recap question suggestions and if there is a consensus on any of them or if you have better ones to ask, please reply here.
I agree - no cheaper to make than VFA's.
As to the sound and subjective assessment, questions like these can be contentious 😉
However, I have built two big VFA's and three CFA's.
I am extremely pleased with the CFA's. They sound wonderful.
As to the sound and subjective assessment, questions like these can be contentious 😉
However, I have built two big VFA's and three CFA's.
I am extremely pleased with the CFA's. They sound wonderful.
Awesome thread ...
Overcoming CFA shortcomings helped to push the VFA's ahead , as well.
Both can sound wonderful, given the right application.
OS
Overcoming CFA shortcomings helped to push the VFA's ahead , as well.
Both can sound wonderful, given the right application.
OS
1). It doesn't take very much hack to get a giant soundstage with the CFA, which is something you'll have to fight for with the classic-miller-VFA.
2). Timbral accuracy, is an epic fight and possibly a failure mode for the CFA, but really easily done with the classic-miller-VFA (most especially at the power circuit which can be either unorthodox or expensive--your choice).
3). PRAT, such as epic forceful breeze really slammin dynamic audio is done with the CFA, and is something you'll have to fight for with the classic VFA, but actually it is quite doable with the VFA and sounds a hella lot better without problem #2.
SO, for initial satisfaction the win goes to CFA; however, for long term utility, the win goes to classic-miller-vfa for useful pleasant tone. I don't think that the news comes as a surprise. Of course we'd like to do better. The good news is that somebody did!
. . .
Given that it takes an awful lot of effort to bend/kink the laws of physics, I, personally believe that the win goes to a specific CFA--the SSA amplifier, which delivers the benefits of a CFA, but surprisingly With the pleasant useful tone of a VFA.
I suggest that we have a closer look at the SSA.
Nico used some fortunate parts and latfets to pimp his SSA and then endorsed it.
Biggun used some constant current power stability circuits to pimp out an otherwise ordinary SSA, and then endorsed it.
That SSA has been compared to my premium 90W~127W chip amplifier effort, and beat mine. Well, I had taken every step to assure it wasn't easy to do that. Nevertheless, I'll admit it is possible. I had NOT experienced that pleasure in person, but I'll admit it could be done. Let's further scrutinize the CFA based SSA amplifier.
Those reviewers are trusted (I know them personally)--On audio quality, it was SSA for the win.
Mine still costs slightly less and is easier to build; however, if you're aiming for the ultimate, it was a specific CFA amplifier, the SSA.
HOWEVER, if we're talking about a random CFA amplifier, you'd better use it for a radio amp and never let it amplify audio band.
*this is my opinion because I have none else to give*
- About miller in a CFA, put-it between the out of power devices (the point where the feedback is taken) and the base of the VAS. You will gain a digit in distortion in comparison to collector/base position.
- With no filter in, chose the value of this compensation cap in order to have the most extended *and flat*, (no peak in the response curve) at HF.
- Then, set an input filter just low enough to have no overshoot on little level square waves.
- Adding an input "damped" diamond stage to the VSSA brought-me a huge improvement, a step up in quality, specially in the bass reproduction.
- If you use the input low pass filter to dump the diamond, IE between its two stages, the FC of this filter will not be affected any more by the impedance of the source, another advantage. And the "nervous" character of the diamond is killed.
-This input diamond will be less expensive than the two caps shorting the CSS output for AC, and you will not have to worry any more about their quality and influence on the sound character.
Can be done with a servo to get rid of any offset problem, fine tuning and their adjustable resistances. In-out of the servo don't change anything you can notice in listening.
If you design this servo for SMDs parts, you even can save me more surface on the board and forget its presence.
www.esperado.fr - VSSA with Diamond input and DC servo
http://www.esperado.fr/vssa-diamond/dvssa.php
(Just sims, make your mind in real life ;-)
Just, i find CFAs are easier to achieve the high slew rates i love, and i like their sonic "expansive" and naturally fluid character. (in opposition to the "compressive" behavior of LTPs)
Of course, they have their own inconveniences, mostly PSRR side.
I'm curious to get your feeling, once you'll have build the best ones from both typologies between all your designs. Don't forget to build a CFA and compare, you may chose this one for your daily use ;-)
- With no filter in, chose the value of this compensation cap in order to have the most extended *and flat*, (no peak in the response curve) at HF.
- Then, set an input filter just low enough to have no overshoot on little level square waves.
- Adding an input "damped" diamond stage to the VSSA brought-me a huge improvement, a step up in quality, specially in the bass reproduction.
- If you use the input low pass filter to dump the diamond, IE between its two stages, the FC of this filter will not be affected any more by the impedance of the source, another advantage. And the "nervous" character of the diamond is killed.
-This input diamond will be less expensive than the two caps shorting the CSS output for AC, and you will not have to worry any more about their quality and influence on the sound character.
Can be done with a servo to get rid of any offset problem, fine tuning and their adjustable resistances. In-out of the servo don't change anything you can notice in listening.
If you design this servo for SMDs parts, you even can save me more surface on the board and forget its presence.
www.esperado.fr - VSSA with Diamond input and DC servo
http://www.esperado.fr/vssa-diamond/dvssa.php
(Just sims, make your mind in real life ;-)
Of course, OS.
Just, i find CFAs are easier to achieve the high slew rates i love, and i like their sonic "expansive" and naturally fluid character. (in opposition to the "compressive" behavior of LTPs)
Of course, they have their own inconveniences, mostly PSRR side.
I'm curious to get your feeling, once you'll have build the best ones from both typologies between all your designs. Don't forget to build a CFA and compare, you may chose this one for your daily use ;-)
Last edited:
http://www.diyaudio.com/forums/solid-state/193923-simple-symetrical-amplifier.html
Where all the L.C. project, VSSA and First One had begun. The story of all his first attempts ;-)
I see, I have tried the SSA and the VSSA types a few times, but frankly the diamond input versions are far better.
I like your compensation scheme, it's like nested feedback , with at short HF loop around the OPS
I like your compensation scheme, it's like nested feedback , with at short HF loop around the OPS
Thank-you. In fact, i don't know exactly why such a difference. I believe the high current capacity of the output stage makes the load of the compensation cap at HF negligible, creating less distortion than when it is an additional load for the collector of the VAS, very limited in current at high frequencies ?
But you can sim the both: it works.