Member
Joined 2009
Paid Member
But I do use it as a description at times, quite convenient it is. Anyhow, we should get back to amplifiers .... only it's Friday and it's been a long week ...
You ever seen this ?
Eddie Izzard- Death Star Canteen - YouTube
Darth Vader: No, I will not need a CFA. I do not need a CFA to hear you. I can hear you without a CFA, with the power of the Force, which is strong within me. Even though I could hear you with a CFA if I so wished. For I would hack at your power supply with the thin bit until the electrons flowed across the canteen floor.
Canteen Worker: No, the feedback impedance is low. You'll need a CFA.
Darth Vader: Oh, I see the impedance is low. I'm sorry. I did not realise. Ha ha ha ha … oh … CFA for the … yes. I thought you were challenging me for the fight.
Last edited:
Hi Richard,
I've got emitter resistors in the current mirror, I hope you noticed that, only I have 1k and 220 ohm instead 1k and 100 ohm.
What do you think of my high impedance + input, does it could work in reality?
BR Damir
That is how I was able to build just about EVERY class AB VFA in existence.
I use the same modular method for my PCB's (below)
That template will be the HK680 triple .... I will "transplant" anything I
want into the blank space for a "front end".
I also have CCS's , multipliers ,Vbe's and regulators as separate files. Just
put them together as "building blocks" - finished board in hours !
One question ? Is the PeeCeeBee a CFA ?
OS
Attachments
Oh GG Waly, could you please substantiate your claim that
for us unwashed. Some have suggested Stochino but please give us your wisdom & expert opinion of his efforts. 😱
And though we are unworthy, if you have better, more elegant, simpler versions of the mythical beast than Stochino's crude & over complex sh*t, please deign to enlighten us.
_______________
Just to remind yus unwashed, I have no torch for VFA or CFA but a big one for simplicity. I base this on a limited number of Double Blind Listening Tests to Lipsh*tz & Vanderkooy bla bla which suggest Simple Sounds Better.
If 2 amps have the equivalent THD, slew, overload & recovery bla bla but one is much simpler than the other (eg Bonsai's nx-amp vs Stochino ) I can use the money, time & effort saved to claim .. I mean have ... my mains cables specially made from the tresses of virgins and insulated with solid Unobtainium. 🙂
The reason why I hope Waly will chime in on VFAs with zillion V/us is I'm hoping he will come up with similar performance without Stochino's mind boggling complexity.
_______________
For those who haven't bothered to read this whole thread, it is now obvious that the evil myths spread by the VFA fans are false. There doesn't seem to any level or item of VFA performance that can't be equalled or bettered by a simpler CFA.
Loadsa mythbusting in this thread.
That's not to say CFAs are plain sailing or easy but that's cos there is much less info on how to do good CFAs. No equivalent to the Self & Cordell books. In fact this thread might be the biggest repository of CFA knowledge.
Alas there is also much noise from evil pseudo gurus 😱
_______________
For those who want to actually build & listen (what a novel idea!) to a good CFA, the obvious candidates are Bonsai's nx-amp or if you want something bigger, dadod's 200w-mosfet-cfa-amp
Andrew's (Bonsai) is simple, fully worked out, tested and makes a good comparison with Stochino. In particular, he gets 'equivalent' performance with much less complexity AND exemplary overload and recovery without extra clamps etc that are essential with Stochino.
Damir's (dadod) is a new design which attempts to incorporate much of the new knowledge from this thread.
_________________
Waly, how about you post a practical or even SPICE world design using your undoubted expertise and knowledge.
I say this totally without prejudice ... but it pains me that you use your intellect & experience just to deride other people's efforts instead of making useful contributions.
That's the sort of behaviour I expect from certain pseudo gurus who only post to promote their egos. But one hopes for better from a budding true Guru.
__________________
.. as it seems to be expected [deleted: 47 pgs of rant about CFA vs VFA .. and yus beach bums are all idiots & deaf .. who dunno what a true CFA is ... ]
Last edited:
2 CFA boards with positive comments from Bonsai and AlexMM and others have been posted on the 200w-mosfet-cfa-amp thread. NO ONE wants to build either the 100 or 200 watt versions.
SO OStripper if you want or need some any kind of help, even though I do not think you do, I have Sprint layouts available for both. Maybe a CFA "BUILDERS" thread can be started and let the sim and word warriors sling arrows and such at each other while some Builders get some real world electrons flowing under some solder and flux.
SO OStripper if you want or need some any kind of help, even though I do not think you do, I have Sprint layouts available for both. Maybe a CFA "BUILDERS" thread can be started and let the sim and word warriors sling arrows and such at each other while some Builders get some real world electrons flowing under some solder and flux.
Last edited:
Member
Joined 2009
Paid Member
HI Damir --- Yes, I did notice.... just pointing out the development and thinking process on it and CFA. I am impressed with your quick learning and SIM application to learn more about this. In fact, several people around here seem to be wizards with SIM programs. I spent too much time managing programs/projects to get to do much SIM but I followed it and have several software programs as they developed better and better along with the PC. I have the latest MicroCap now. The only thing that stops me from doing more and more and more -- was I got old somewhere along the way and retired. Ooops! What a drag that is. But, it is a lot of fun still. More new adventures.
Thanks to everyone here for such stimulating thoughts and comments. I'm sure there is more to come. I'm enjoying it a lot. ... Lets also go build some CFA and listen to some music .... pcb's ready!!
Thx-Richard Marsh
Last edited:
I saw Dadods big 200 watter. Looks well thought out ... but too many devices.
I just looked at the NX , It looks like my first attempt (100 posts ago ).
Actually , is that a diamond buffer I see ? OP stage is different , as well.
I never looked at the NX before - but it is similar to my first sim (below) ... which was put
into doubt ... as to whether it was a "true CFA" 😕 .
BTW - excellent documentation , Bonsai - easy to read/comprehend.
The far simpler Pee Cee Bee looks like a better bet.
I think builders would want -1. scalable 2. durable 3. not "picky" (devices-wise)
4. ALL through- hole
I have seen the comments " call CFA .. "JOE" or "controversial feedback amp"
I'll ask again .. are the NX and peeceebee real CF amplifiers ?
Much of this thread is pure silliness - that's why I am "skipping around" , I
just want the info needed because I am "goal oriented". If the "JOE" amp
sounds/performs good - so be it , whatever it is called.
It is a shame DIY'ers would would rather build poorly designed projects with
catchy names instead of better designs like the NX and others. Dadod's design
just scares the builder with TOO many devices.
OS
Attachments
The nx and sx have both been built and are in service ( I'm listening to the sx now). They are both CFA.
PMI was arranging GBs for the nx, and Jims audio has them on ebay
(I have no commercial affiliation with either)
PMI was arranging GBs for the nx, and Jims audio has them on ebay
(I have no commercial affiliation with either)
OS you may want to look at Alexander comp rather than the MC you are showing above. Single cap rather than 2 needed. The both nx and sx use it.
It seems you ignore me, anyway pecebe is VSSA, look to VSSA thread and see. There are couple of hundreds GB builders of VSSA, all reported no problems, claimes the sound is better than most VFA.
Again, there's no reason for out of the feedback buffer in front of any amp, soundwise is better to avoid it, tested.
Start with simplest CFA front end, put CCS instead input bias resistors and than proceed from there.
Regards L.C.
I'm sure that superb amplifiers can be made with CFAs, but I must admit that I have never built a CFA power amplifier, and I have some difficulty understanding that CFA has any fundamental advantage over VFA or the other way around.
Perhaps due to my lack of experience with CFA, my impression is that it may be more straightforward to achieve high performance with the VFA topology. Note that straightforward is not necessarily the same thing as level of complexity. Indeed, we do not all count complexity in the same way. I believe there is more to it than counting small-signal transistors.
I think the slew rate thing is a red herring. I do tend to be an advocate of high slew rate, but I do not believe that slew rate achieved by current-on-demand where an early stage may go out of class A is an advantage to sound quality. I like good slew rate margin because it is normally reflective of margin against high-frequency nonlinearity.
At the same time, VFAs can be designed to have way more than adequate slew rate without resort to a current on demand CFA topology. The use of Miller Input Compensation (MIC) in my MOSFET EC amp enabled a 300 V/us slew rate in a 50-watt amplifier. Bottom line is that I don't believe the slew rate argument supports CFA as a better topology (this is NOT to say that CFA is inferior to VFA).
I still have trouble getting beyond the basic limitation of the output stage being the thing that levels the playing field between CFA and VFA. In a really well-designed version of either topology, the output stage would seem to govern the ultimate distortion performance achievable. I could be wrong, but I don't see magic in the CFA topology that allows us to get more stable feedback around the output stage in a CFA than in a VFA.
I get the impression from this thread that CFAs may have more issues with matching than VFAs. Maybe I am wrong. But I can say that my own way of looking at things takes points away for increased need for matching or adjustments.
I'm not against CFAs, but I guess I'm not quite there yet in understanding why they would be preferred over VFAs.
Cheers,
Bob
Perhaps due to my lack of experience with CFA, my impression is that it may be more straightforward to achieve high performance with the VFA topology. Note that straightforward is not necessarily the same thing as level of complexity. Indeed, we do not all count complexity in the same way. I believe there is more to it than counting small-signal transistors.
I think the slew rate thing is a red herring. I do tend to be an advocate of high slew rate, but I do not believe that slew rate achieved by current-on-demand where an early stage may go out of class A is an advantage to sound quality. I like good slew rate margin because it is normally reflective of margin against high-frequency nonlinearity.
At the same time, VFAs can be designed to have way more than adequate slew rate without resort to a current on demand CFA topology. The use of Miller Input Compensation (MIC) in my MOSFET EC amp enabled a 300 V/us slew rate in a 50-watt amplifier. Bottom line is that I don't believe the slew rate argument supports CFA as a better topology (this is NOT to say that CFA is inferior to VFA).
I still have trouble getting beyond the basic limitation of the output stage being the thing that levels the playing field between CFA and VFA. In a really well-designed version of either topology, the output stage would seem to govern the ultimate distortion performance achievable. I could be wrong, but I don't see magic in the CFA topology that allows us to get more stable feedback around the output stage in a CFA than in a VFA.
I get the impression from this thread that CFAs may have more issues with matching than VFAs. Maybe I am wrong. But I can say that my own way of looking at things takes points away for increased need for matching or adjustments.
I'm not against CFAs, but I guess I'm not quite there yet in understanding why they would be preferred over VFAs.
Cheers,
Bob
Some parts of this debate look like this:
“According to legend, the radical medieval theologian and poet, Peter Abelard, once confounded his teachers by subjecting their received wisdom to a simple empirical test.
His scholastic masters had been arguing about exactly how many teeth there should be in a horse’s mouth. If they applied the principles of the classical philosopher Aristotle, they arrived at one number, but, if they relied upon the observations of another ancient sage, a different total suggested itself.
Backwards and forwards the argument raged until the young Abelard, frustrated beyond endurance, rose to his feet, and, calling upon his fellow students to follow him, marched down to the marketplace, where he simply forced open the mouth of the one horse after another – and counted its teeth.”
I wonder if someone can judge the quality amp design but he/she is never build the mathematical model or simulating or build the amp then measure and listen.His scholastic masters had been arguing about exactly how many teeth there should be in a horse’s mouth. If they applied the principles of the classical philosopher Aristotle, they arrived at one number, but, if they relied upon the observations of another ancient sage, a different total suggested itself.
Backwards and forwards the argument raged until the young Abelard, frustrated beyond endurance, rose to his feet, and, calling upon his fellow students to follow him, marched down to the marketplace, where he simply forced open the mouth of the one horse after another – and counted its teeth.”
Last edited:
Bimo post 2595
I wonder if someone can judge the quality amp design but he/she is never build the mathematical model or simulating or build the amp then measure at listen.
The other day I modeled output stage Patent US4595883_Kazuaki_Nakayama
As it turned out, this output stage operates without cutoff quiescent current of the output transistors and distorts when the output voltage of 30 V (peak) for resistive load only 0.005%. When I connected to it a model of the electrodynamic loudspeaker distortion it became 0.6% - increased by more than 100 times. So it's not so easy with quiescent current of the output transistors as it seems at first glance.
best regards
Petr
I wonder if someone can judge the quality amp design but he/she is never build the mathematical model or simulating or build the amp then measure at listen.
The other day I modeled output stage Patent US4595883_Kazuaki_Nakayama
As it turned out, this output stage operates without cutoff quiescent current of the output transistors and distorts when the output voltage of 30 V (peak) for resistive load only 0.005%. When I connected to it a model of the electrodynamic loudspeaker distortion it became 0.6% - increased by more than 100 times. So it's not so easy with quiescent current of the output transistors as it seems at first glance.
best regards
Petr
Alas, I was hoping this thread would persuade Guru Cordell to at least dip his toe in CFA waters and get some 'real life' experience with CFAs. 🙁
I confess to being anti-CFA in my previous life but this thread has certainly opened my eyes as to what's possible.
If we accept Stochino's design priorities (which I don't entirely), then Stochino vs nx-amp, I think, demonstrates CFA's sweet spot. Bonsai's design has zillion V/us, good THD, overload, recovery & stability coupled with deceptive simplicity.
For 1pp zillion THD20k, (which again isn't my priority) I would probably still start with a VFA but this is cos I'm familiar with how to do this with VFAs. IMHO, the problems with doing this with CFAs are more to do with the symmetrical topologies which most people seem to associate with CFAs.
I'm also not sure that the OPS is the remaining major THD factor. The use of 'pure Cherry', and to a lesser extent TMC, makes the evil xover type distortions less important than some of the evils of IPS/VAS (due to symmetrical topologies) ... even with the EF2 OPS I favour over Bob's beloved Locanthi Triples 🙂 YMMV but I see this with my sims at 50W 8R and his devices on my simple designs. This counts as a demerit against CFAs.
My problems in getting 1pp zillion with CFAs are to do with the limited gain of symmetrical IPS/VAS rather than xover OPS type evils. But this is of course dependent on exact circuitry & topology. They are not insurmountable and I've posted some possible solutions .. but untried in 'real life'.
My remaining caveats are to do with stability & repeatability of VAS current. Though LC reports stable results with his VSSA, I would want to get a good handle on this that you only get when you make loadsa amps yourself. At the end of the day if this requires difficult tweaking, I may have to mark CFAs down like Bob.
May I urge Guru Cordell to have a 'real life' go. I'm certain that once he starts, he will find good things to say .. even if he doesn't become a rabid CFA fan. 🙂
I confess to being anti-CFA in my previous life but this thread has certainly opened my eyes as to what's possible.
If we accept Stochino's design priorities (which I don't entirely), then Stochino vs nx-amp, I think, demonstrates CFA's sweet spot. Bonsai's design has zillion V/us, good THD, overload, recovery & stability coupled with deceptive simplicity.
For 1pp zillion THD20k, (which again isn't my priority) I would probably still start with a VFA but this is cos I'm familiar with how to do this with VFAs. IMHO, the problems with doing this with CFAs are more to do with the symmetrical topologies which most people seem to associate with CFAs.
I'm also not sure that the OPS is the remaining major THD factor. The use of 'pure Cherry', and to a lesser extent TMC, makes the evil xover type distortions less important than some of the evils of IPS/VAS (due to symmetrical topologies) ... even with the EF2 OPS I favour over Bob's beloved Locanthi Triples 🙂 YMMV but I see this with my sims at 50W 8R and his devices on my simple designs. This counts as a demerit against CFAs.
My problems in getting 1pp zillion with CFAs are to do with the limited gain of symmetrical IPS/VAS rather than xover OPS type evils. But this is of course dependent on exact circuitry & topology. They are not insurmountable and I've posted some possible solutions .. but untried in 'real life'.
My remaining caveats are to do with stability & repeatability of VAS current. Though LC reports stable results with his VSSA, I would want to get a good handle on this that you only get when you make loadsa amps yourself. At the end of the day if this requires difficult tweaking, I may have to mark CFAs down like Bob.
May I urge Guru Cordell to have a 'real life' go. I'm certain that once he starts, he will find good things to say .. even if he doesn't become a rabid CFA fan. 🙂
As with the "Badger" , I would leave "all options on the table" ... compensation
wise. 😀 ." Alexander comp." - I saw that on the NX and then read this - http://www.analog.com/static/imported-files/application_notes/58052492001115525484056221917334AN211.pdf.
Now that (it has been said) that these amps are really CF - I see light at the end
of the tunnel. 🙂
Oh ... no, You seem to be the one that started this whole CFA thing.
I took the steps you suggested (CCS) (below 1) . Added the cascode to
allow 70v rails and to have a massive choice of input devices (only 24V-Q2/4).
It is close to my known VFA (below 2 - comparison),
I might try to use Voltage regulators instead of current sources to close
the gap (20db). Still , not bad in most of the audio range.
But after extensive simulation the amp will work and not burn. The finer "tweaks"
(different compensation/currents) and assessment can come later with the real amp.
EDIT - nope .. CCS is best. Omitting either the CCS OR the cap multiplier
sends PSRR right to the "doghouse". Attachment 1 schema is best for PSRR.
OS
Attachments
I mean if someone can judge something that he/she doesn't know anything about it 😉.
I agree, simulation help a lot in design process.
The Stochino Ultra Fast Amplifier is complex on paper but quite simple to build. I have built atleast 2 versions on differently laid layouts. No problems with oscillations, despite the very high bandwidth. Sound to me the best of all VFAs I have built so far; must admit that I have not yet built a CFA, but close to building the NX.