I am not sure I understand. A complementary common emitter stage, which is what many "CFAs" really are, will always source and sink current to and from the feedback network, but I fail to appreciate why this should mean "no feedback".
By the way, Scott, what would you call the complementary feedback pair used by Douglas Self as the preffered output stage in his amplifiers?
Would you call it a "CFA" pair?
Last edited:
I think the old "Tigersaurus" used an output stage with a CFP and a gain network to need less swing at the driver stage. As for what Doug does I have no idea, never read it.
I am surprised you haven't read Douglas's work.
Anyway, the complementary feedback pair was around long before the term "CFA" came into use, and was always regarded as possessing 100% shunt (voltage) derived-series (voltage) applied negative feedback. Note its similarity to the so-called "CFA".
Last edited:
You didn't see my example, you typically can't use the "theoretical" extra loop-gain since the LTP must be degenerated or the comp cap increased or the VFA will not be unity gain stable. This extra loop-gain is imaginary.
I am sorry this is not true at all. I stated that for relatively low closed loop gains the so-called "CFA" reduces the forward path gain and, therefore, the loop gain.
This is not the case with VFAs: the two stage VFA with an LTP input stage will always permit greater useable forward path gain and therefore loop gain than the so-called "CFA", even for the same stability margins.
This is not the case with VFAs: the two stage VFA with an LTP input stage will always permit greater useable forward path gain and therefore loop gain than the so-called "CFA", even for the same stability margins.
Last edited:
I am surprised you haven't read Douglas's work.
No offense to Doug intended, if I wanted to do a discreet PA I would read Doug and Bob's books. To be honest many of the things in books have been trade secret for years, after all I have Barrie Gilbert, Paul Brokaw, and more available anytime I need. For me a really nice FET input version of the old Hafler 220 would be all I ever need. That I can handle by myself (so to speak).
Not fair, there are two stage CFA's, oh about $500,000,000 shipped.
Take a one stage VFA, LTP into folded cascode and compare.I'll leave you to this fight now.
Last edited:
Michael,
Somewhat confused by your answer, when you say an LTP generates far less even order distortion, do you mean for practical matching reasons?
I understand what you are saying about the higher LTP loopgain but my question was which topology had less distortion (in isolation, no other distortion mechanism's present). Much like an LTP can be degenerated to lower loop gain but increase local linearity.
Thanks
-Antonio
It seems to me that your prime objection to the CFA is the fact that its open loop gain is not as high as VFA.
However, you have forgotten that in most amplifiers (IC opamps included) it is the output stage that sets the upper ULGF. So, at LF the CFA indeed does have lower OLG, and hence loop gain when f/back is applied, but at HF it easily matches or exceeds conventional VFA amplifiers. To SW's point, you have to close the VFA loop so that you have sufficient stability margin, so to claim that VFA loop gains are higher overall is illusioury and does not consider the whole story.
From the designers perspective (and espcially ones like me that build, test and listen to my creations), I dont know that 70dB of loop gain at 100 Hz actually brings anything more to the final listening experience than say 50 dB. But, I won't get drawn into that deabate, since we all know its been thrashed to death on this forum.
In the meantime, the world will contine to call them what they really are: CFA's, and companies will continue to make 'em and sell 'em ($500 million worth from one of them) and I will continue to design them along with VFA's as well of course.
I will also step out of this discussion now. We have been down a country lane and we are headed absolutely no where on this one . . .
However, you have forgotten that in most amplifiers (IC opamps included) it is the output stage that sets the upper ULGF. So, at LF the CFA indeed does have lower OLG, and hence loop gain when f/back is applied, but at HF it easily matches or exceeds conventional VFA amplifiers. To SW's point, you have to close the VFA loop so that you have sufficient stability margin, so to claim that VFA loop gains are higher overall is illusioury and does not consider the whole story.
From the designers perspective (and espcially ones like me that build, test and listen to my creations), I dont know that 70dB of loop gain at 100 Hz actually brings anything more to the final listening experience than say 50 dB. But, I won't get drawn into that deabate, since we all know its been thrashed to death on this forum.
In the meantime, the world will contine to call them what they really are: CFA's, and companies will continue to make 'em and sell 'em ($500 million worth from one of them
) and I will continue to design them along with VFA's as well of course.I will also step out of this discussion now. We have been down a country lane and we are headed absolutely no where on this one . . .
The flaw in your analysis is the assumption that both the "CFA" and the VFA would need the same compensation capacitor. This, I am afraid, is a comparison of apples and aadvarks.
The customery "CFA" is a complimentary common emitter arrangement with shunt capacitive compensation at its output, while the usual VFA is a two stage arrangement with minor loop (Miller) feedback compensation.
The two arrangements cannot, therefore, have the same compensation capacitor for the same unity loop gain frequency.
Obviously, so you agree forward gain for pole-split VFA ~gm/Cc*s and the CFA ~1/RfCc*s for the same crossover (usually process limited) the forward gain is the same. Either you increase the Cc or reduce gm for the VFA both bad for large signal BW.
Furthermore by and large high frequency drivers have gone fully differential, now the CFA regains symmetry. The industry has spoken.
Another list you can compile, all the commercially successful general purpose IC op-amps employing high order nested feedback.
The usual single-stage so-called "CFA" will always generate less loop gain across the audio band, especially at low closed loop gains, than a competently designed two stage VFA for the same stability margins.
The notion that a so-called "CFA" provides more loop gain at high audio frequencies than a compentently designed VFA is laughable. I KNOW this isn't the case.
A LTP with a current mirror and properly degenerated generates appreciably less even order distortion than a single common emitter stage. This is a significant reason why the so-called "CFA" is inferior to a compentently designed VFA in audio applications.
Not fair, there are two stage CFA's, oh about $500,000,000 shipped.
Not shipped to the audio industry, i'll be bound?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.