Michael,
Then can you factually and effectively state why it is a mistake to use this topology in the current opamps manufactured and used currently? I am asking in all seriousness, is there some reason that this should not be done in an operational amplifier, is there some other reason that this is acceptable in that application?
Then can you factually and effectively state why it is a mistake to use this topology in the current opamps manufactured and used currently? I am asking in all seriousness, is there some reason that this should not be done in an operational amplifier, is there some other reason that this is acceptable in that application?
My thinking is that there is no conceivable reason for using so-called "CFAs" in audio applications because traditional VFAs are demonstrably superior in this application in all respects.
Just consider Scott Wurcer's beautiful AD797 or the venerable but excellent NE5534. No so-called "CFA" can approach the performance of these devices in all relevant parameters.
As for discrete "CFA" power amplifiers, I fail to appreciate why one would bother designing such things at all.
Just consider Scott Wurcer's beautiful AD797 or the venerable but excellent NE5534. No so-called "CFA" can approach the performance of these devices in all relevant parameters.
As for discrete "CFA" power amplifiers, I fail to appreciate why one would bother designing such things at all.
Read my post.
I was rebuffing Mikeks erroneous claim that the front end stage in CFA's runs in class AB. Nothing to do with rise time claims
Nevertheless, the 5GHz Ft transistors don't determine the rise time and slew rate in discrete CFA or VFA power amps.
Random thoughts after a few months away from this forum (frustrated):
Why for discrete CFA power amps does one accept significantly lower open loop gain?
Just because the CFA input stage is class AB does not mean it operates in class B over the intended range of inputs.
Chasing ppm THD with a simulator (though not my cup of tea) has lots of merit and has brought to this forum many design subtleties.
Why so much negativity?
Design ideas, improvements, time saving lessons learned, and just having to rethink things we thought we knew come form all posts so consider being grateful. Its a hobby, your creation should have personality (however you see fit) it doesnt have to be "better".
Thanks
-Antonio
Why for discrete CFA power amps does one accept significantly lower open loop gain?
Just because the CFA input stage is class AB does not mean it operates in class B over the intended range of inputs.
Chasing ppm THD with a simulator (though not my cup of tea) has lots of merit and has brought to this forum many design subtleties.
Why so much negativity?
Design ideas, improvements, time saving lessons learned, and just having to rethink things we thought we knew come form all posts so consider being grateful. Its a hobby, your creation should have personality (however you see fit) it doesnt have to be "better".
Thanks
-Antonio
Nonsense Michael. Pure delusion on your part regrettably.
You may get this in a discrete CFA design if you've set you total feedback impedance too high, which is analogous in VFA to setting your LTP tail current too low. Secondly, most sane designers precede their amplifiers with a low pass filter that should account for this and prevent the input stage (VFA or CFA) from overloading.
You did no such thing old chap. A complimentary push-pull so-called "CFA" input stage does, in fact, operate in class A until the current demands of the compensation capacitor at sufficiently high frequencies and voltage swings cause it to revert to class B. Hence such an input stage is said to operate in class AB.
You may lead a horse to water...
Where on earth did you get this totaly delusive analogy from?
My thinking is that there is no conceivable reason for using so-called "CFAs" in audio applications because traditional VFAs are demonstrably superior in this application in all respects.
Just consider Scott Wurcer's beautiful AD797 or the venerable but excellent NE5534. No so-called "CFA" can approach the performance of these devices in all relevant parameters.
As for discrete "CFA" power amplifiers, I fail to appreciate why one would bother designing such things at all.
On pure distortion, VFA may win just and only just (single ppm differences). At mid to lower output levels there is nothing between them.
On bandwidth, rise time (small and large signal) CFA has the edge.
On stability, see Waly's comments. To get demonstrably better performance out of VFA, you need to go to TPC or TMC. If you don't, the only way is very high loop gains. Nothing wrong with that, but It lacks finesse in my book.
On circuit simplicity, CFA has the lead as well.
It's not as good on PSRR, but there's a trick to easily match or exceed VFA . . .
Finally, I suspect due the the input stage structure, it's DC performance in discrete power amps is also better. DC coupled and no servo needed. Try that with a FET input, folded cascode design or blameless and you see it is difficult to do unless you AC couple the FB network and with FETs a servo is just about mandatory if you want decent DC performance.
So, in my view, CFA is easily a match for VFA.
But, I think I am the only one here that has designed and built both topologies recently and can make therefore a fair comparison. If I am to believe the naysayers who have done nothing more than perhaps read a few data sheets, then CFA is a trashy, cheap compromised topology. My experience tells me a totally different story.
So before mouthing off any more, I challenge you to build and listen. Then you'll see they are damn good.
Last edited:
In da 20th century, this was of great interest to me and I've conducted a series of Double Blind Listening Tests bla bla with my Listening Panel including some of the best ears in the business ie true golden pinnae.
My tests were done with 20kHz analogue brick wall filters. Vinyl with MC cartridges and 30 ips mastertapes wangled from various recording companies. Music & speech only as its easy to find artificial test signals that are easier to identify.
When you do proper tests of this sort, you quickly find out that
- most people can't tell the difference
- self declared Golden Pinnae including most HiFi reviewers, are even less (a LOT less) perceptive than Joe (or Jane) Public.
- OF THOSE WHO CAN TELL THE DIFFERENCE RELIABLY (ie true golden pinnae) .. ALL PREFERRED THE BAND LIMITED SIGNAL
This millenium, I was pleased to find those who have tried to do such tests properly (including Jan Didden) .. got the same results as my 1980 tests and those of the BBC & others ie
WHEN A DIFFERENCE CAN BE DISCERNED RELIABLY (by the true golden pinnae), THE PREFERENCE IS ALWAYS FOR THE BAND LIMITED SIGNAL.
I believe Jan's tests were with 15kHz brickwalls.
ostripper, can you tell us what source material you used for your tests? Specific recordings please. What speakers?
I have a yen to repeat my tests this century and hopefully get a different result.
I hope I don't have to explain to this august panel that the opinion of 'those who can't pick out the difference reliably' (which includes most (all?) self declared Golden Pinnae) is not really of interest.
Mike, why don't you list your 'in all respects' and post a simple traditional VFA design of yours that illustrate these.
I'm certain that there are enough CFA advocates that will relish the challenge of improving on your design 'in all respects' with similar simplicity.
Using Scott's AD797 as an example is rather unfair as .. with due respect to the CFA fans on this forum .. I don't think anyone of them would claim to be able to do better than Guru Wurcer.
Your observaions are right on the bulls-eye mark !!
[except being th only one who has deigned/built/listened to both topologies]
Thx-RNMarsh
Why can't you clear up the misconception instead?
(the bias or strip or whatever current you want to call it goes to zero on the off transistor regardless of the feedback resistors).
Thanks
-Antonio
Bonsai, how else would you explain the so-called "CFA's" propensity (with a complimentary push-pull input stage) to generate very high slew rates, well in excess of those supported by the stage's quiescent current?
In other words, where do think the relatively high currents (higher than the standing current of the input stage) come from to drive the compensation capacitor during slewing?
I'm trying to follow this need a little clarification. What causes the loading on the IPS for both VFA / CFB? Is it the miller compensation or shunt compensation?
If both input stage types are being loaded then what happens to a VFB when the loading increases?
Sorry for the basic questions. Just getting a little confusion in the mind.
If both input stage types are being loaded then what happens to a VFB when the loading increases?
Sorry for the basic questions. Just getting a little confusion in the mind.
Actualy Mikekiwanuka is right , a CFB can enter class B if the input
signal slew rate is high enough.
An exemple is your own NX amp when stressed with a fast square
wave that still is within its output level capability.
Notice the value of the input transistors peak current ,
in this case visible only for a side of the amp.
Attachments
Last edited:
Anyone who would bother to actually study a 'CFA' or at least the data sheet for an ic one, would know that. Jeez.
jan
It s a direct consequence of the ability for each side of the amp
to supply more than the quiescent current if necessary , the non
conducting half will then be switched off largely but , well , not
everyone is really interested in the functionalities of the topology
it seems.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.