Almost fully incorrect, just another biased opinion about CFA vs. VFA. I will be back later, meantime, only one comment: pole splitting is a side effect of Miller compensation, not a compensation method, and is not specific to VFAs, can be taken advantage of in CFAs as well. Yet another reason why shunt compensation is suboptimal in CFAs and VFAs.
No, and it's not different from a VFA.
Maybe, but the stated causality is absolutely wrong.
True, as old as Miller compensation, and what would be the "new" way, mind you? Don't bother to reply.
Standard Operating Procedure: you put up a bunch of wrong arguments presented as facts, then you slowly take them back when challenged. Do I need to quote your posts in which you praised CFAs as providing more loop gain, and hence lower distortions compared to VFAs?
Whatever, those who understand know where the truth lies, anyway. Those who don't, there's no logical argument that will change their minds. Sometimes I really wonder why I assumed the role of garbage collector on this forum. People, and likely myself, would be so much happy if I would give up. It's fun to a point, but that's wearing out.
Last edited:
Sad but true. People often look for any argument that would confirm their existing belief rather than look for arguments to build a position on.
Like the argument 'you can't subtract a voltage from a current' which is factually true, but hasn't anything to do with the discussion. But it is great to shore up your believe of course.
At a certain point you learn to let it go
Jan
At a certain point you learn to let it go
Jan
Jan I've lost count of the times you listened to a well thought out argument and changed your mind about some long held belief. This is rare, in a fews days we will probably see Otala brought up again as irrefutable gospel. That chart paper plot held up as some kind of audio Rosetta Stone.
OT: Scott: house concert @ Axis of Evil HQ, 6 Oct. I'll be there.
/OT
Jan
Me too.
Waly you do seem to get excited over things. It is you who is misquoting and taking things out of context. Again, an attempt has been made by various people to have a discussion on the subject, and again we end up slinging personal insults around. Just like the first CFA/VFA thread. I find it difficult to understand why this is such an emotive subject, or why there is a rush to dismiss those that want to discuss it.
I stand by my technical points - there's no retraction or backing off on any of them:-
Load any amplifier and the OPS pole will migrate down in frequency and up in magnitude. If the pole lies enough below 0dB, little compensation needs to be applied to ensure it does not become a problem as you load it. This is behaviour of a classic low loop gain, wide bandwidth CFA. A VFA, with higher OLG and LG's, is different (unless it is sub-optimal and the loop gain is uncharacteristically low) because the OPS pole lies close to the ULGF and the loop gain is therefore also close to 0dB. Load it and you quickly have enough phase shift to cause ringing or oscillation ergo, you have to compensate for it. I am not saying CFA's are better - just pointing out the differences in behaviour between classic versions of either topology, OK?
I stand by my technical points - there's no retraction or backing off on any of them:-
Load any amplifier and the OPS pole will migrate down in frequency and up in magnitude. If the pole lies enough below 0dB, little compensation needs to be applied to ensure it does not become a problem as you load it. This is behaviour of a classic low loop gain, wide bandwidth CFA. A VFA, with higher OLG and LG's, is different (unless it is sub-optimal and the loop gain is uncharacteristically low) because the OPS pole lies close to the ULGF and the loop gain is therefore also close to 0dB. Load it and you quickly have enough phase shift to cause ringing or oscillation ergo, you have to compensate for it. I am not saying CFA's are better - just pointing out the differences in behaviour between classic versions of either topology, OK?
That's better, Mr. Bonsai, but still confused. It is the open loop output impedance which, together with the capacitive load, determine an extra pole (so NOT the OPS pole) and hence a phase shift, and the position of this extra pole DOES depend on the compensation method; Miller based compensation again shines here, and shunt compensation is suboptimal. The whole trick is to have this pole far in frequency (usually more than one decade) so it is not below the ULGF when closing the feedback loop.
Once again, this has nothing to do with the CFAs or VFAs stability with capacitive loads. A common way of improving a capacitive load stability is increasing the circuit's noise gain, without changing the signal gain, thus reducing the frequency at which the loop gain reaches 0dB. It is true that a VFA would benefit more from this technique (because there's more loop gain), but again this doesn't prove any correlation with the CFA or VFA topology, it's just another trade of bandwidth vs. loop gain.
Perhaps you could say that shunt compensation in a CFA would deliver more resilience to capacitive loads, but that's a very poor excuse to waste otherwise perfectly good loop gain for this.
BTW, claiming that a VFA is a long tail pair, plus a VAS plus an output stage is a gross oversimplification.
"That's better, Mr. Bonsai, but still confused. It is the open loop output impedance which, together with the capacitive load, determine an extra pole (so NOT the OPS pole) and hence a phase shift, and the position of this extra pole DOES depend on the compensation method; Miller based compensation again shines here, and shunt compensation is suboptimal. The whole trick is to have this pole far in frequency (usually more than one decade) so it is not below the ULGF when closing the feedback loop.
Once again, this has nothing to do with the CFAs or VFAs stability with capacitive loads. A common way of improving a capacitive load stability is increasing the circuit's noise gain, without changing the signal gain, thus reducing the frequency at which the loop gain reaches 0dB. It is true that a VFA would benefit more from this technique (because there's more loop gain), but again this doesn't prove any correlation with the CFA or VFA topology, it's just another trade of bandwidth vs. loop gain.
Perhaps you could say that shunt compensation in a CFA would deliver more resilience to capacitive loads, but that's a very poor excuse to waste otherwise perfectly good loop gain for this.
BTW, claiming that a VFA is a long tail pair, plus a VAS plus an output stage is a gross oversimplification. "
No confusion whatsoever.
I think we all understand that if a pole migrates down if frequency with a capacitive load, the output impedance of the source is non-zero. It's is common knowledge that amplifier output Z is non-zero.
Shunt comp is sub-optimal and I've certainly not promoted anywhere.
I am saying CFA's are more load tolerant if they are low loop gain, wide loop bandwidth and the OPS pole lies well above the ULGF i.e. gain magnitude well below 0 dB. That's not a blanket statement that CFA's are more capacitive load tolerant and VFA's. If you start increasing a CFA OLG so it more resembles a VFA, all bets are off of course - I've said that numerous times.
For audio applications, why increase noise gain under any circumstances other than when using a decompensated opamp? In a discrete power amp that approach is sub-optimal. You have access to all internal nodes so you should be able to optimize the comp design to get the best out the amp by maximising the loop gain - VFA or CFA. The only reason I could think you would do this is to overcome instability issues brought on by a capacitive load - but then a better solution is to isolate the load from the amp with an output inductor.
There's plenty of professional literature out there that treats a VFA as an [LTP+VAS+OPS] for basic stability analysis - See Bruno Putzey's 'The F Word' and James Solomon's 'The monolithic opamp . . . ' for examples.
Once again, this has nothing to do with the CFAs or VFAs stability with capacitive loads. A common way of improving a capacitive load stability is increasing the circuit's noise gain, without changing the signal gain, thus reducing the frequency at which the loop gain reaches 0dB. It is true that a VFA would benefit more from this technique (because there's more loop gain), but again this doesn't prove any correlation with the CFA or VFA topology, it's just another trade of bandwidth vs. loop gain.
Perhaps you could say that shunt compensation in a CFA would deliver more resilience to capacitive loads, but that's a very poor excuse to waste otherwise perfectly good loop gain for this.
BTW, claiming that a VFA is a long tail pair, plus a VAS plus an output stage is a gross oversimplification. "
No confusion whatsoever.
I think we all understand that if a pole migrates down if frequency with a capacitive load, the output impedance of the source is non-zero. It's is common knowledge that amplifier output Z is non-zero.
Shunt comp is sub-optimal and I've certainly not promoted anywhere.
I am saying CFA's are more load tolerant if they are low loop gain, wide loop bandwidth and the OPS pole lies well above the ULGF i.e. gain magnitude well below 0 dB. That's not a blanket statement that CFA's are more capacitive load tolerant and VFA's. If you start increasing a CFA OLG so it more resembles a VFA, all bets are off of course - I've said that numerous times.
For audio applications, why increase noise gain under any circumstances other than when using a decompensated opamp? In a discrete power amp that approach is sub-optimal. You have access to all internal nodes so you should be able to optimize the comp design to get the best out the amp by maximising the loop gain - VFA or CFA. The only reason I could think you would do this is to overcome instability issues brought on by a capacitive load - but then a better solution is to isolate the load from the amp with an output inductor.
There's plenty of professional literature out there that treats a VFA as an [LTP+VAS+OPS] for basic stability analysis - See Bruno Putzey's 'The F Word' and James Solomon's 'The monolithic opamp . . . ' for examples.
So it's no longer about the OPS pole. Still, wrong causality. A pole migrates when, and because, closing the feedback loop, not because the non-zero output impedance, EE101
Yes, you are saying this, but did not provide a shred of evidence.
I guess you mean CLG, but then again, can't read your mind of what you are trying to convey here.
Says who? Maybe, or maybe not, I have not seen any pro or con evidence. All audio amplifiers are decompensated before compensating, and compensating for the common reactive loads and adding an extra margin for capacitive loads by increasing the noise gain could be beneficial, I don't know.
Not to ruffle more feathers, but if you want to get to professional literature you should probably visit a library and look slightly beyond a Linear Audio article. If I won't be such a lazy **** I could provide some references, but I'm sure you'll manage without my help.
I have not seen any argument to support that it has nothing to do with the discussion. Feedback relies on an amplified difference of two voltages. The amplifier circuit needs an active device to sense it. In its most basic form, this device is a single valve or a singe transistor where the voltage difference is between the input of the device and its voltage following electrode.
A push pull inverting input on which most people here seems to concentrate is nothing more than an improvement of the basic architecture
By the way, what is the date of the first negative feedback applied to a cathode ?
They think by cliché, associating drawings they recognize with names, following currents by fingers.
Your analysis is not exactly right, but maybe we should call a CFA IPS a Floating Common Base IPS and everybody would be happy?
Cheers
S
Try ignore list, it works almost like a charm! Almost, because closed messages still appear in the thread.
