Some dont know what it is and how it works..... And, the two books ref here for DIY'ers dont cover it... and dont seem to even want to... while the rest of the world is deep into it. If people would just read up on it first and then go to design and build here.... you could get your wish.-RNM
Yes , a huge peak with a low ratio... 5-10X (FB -R)
20 - 30X ... flattens right out. My design is normally 26X (good).
OS
-RNM
I never even heard of it a few months ago 😱 .
But now I know it will work from junkbox parts and still perform
quite well.
YES ... (below) 😀
OS
Attachments
I'm not convinced that that description of whether something is a CFA is sufficiently definitive. The key, I believe, is that there is a continuum of behavior between CFA and VFA.
As I said in a previous post, how much of the feedback current flows into the emitter of the input stage, as opposed to into the feedback shunt resistor, defines where in this continuum the amplifier falls.
When the amplifier is well into the CFA end of the continuum, the value of the feedback resistor defines the ULGF, largely independent of the closed-loop gain, which will depend on the value of the shunt resistor. This is because, in the CFA mode, most of the feedback current flows into the emitter, thus controlling the loop gain.
What I said above is not necessarily at variance with #5127, but I think it is more definitive, especially in regard to the continuum of behavior of an amplifier between CFA and VFA.
Cheers,
Bob
Hi Richard,
Please be more specific about which terminology I am using that you seem to be taking issue with.
Cheers,
Bob
I am commenting on Wily as well and some comments applied to his remarks on what is or isnt a CFA. If the shoe fits, its for you.
What you have articulated here is exactly correct... but more so for some topologies than others..... regarding the continuum. The Classic CFA is designed from the start to operate only by controlling current... thus parameters and characteristics. Now the questionable one is the 'Simplifed' CFA... or the one I publish first which was able to be direct coupled, is the complimentary push-pull. That one has caused problems because it could behave in both ways, easily.
Note I used 100/1K in the feedback and low values all around for low noise and high speed/BW. It behaved as what is now named CFA or Current-Mode of operation. but you cannot tell only by the topology alone or just that the input Z for a port is low.
THx-RNMarsh
Last edited:
Long May This Thread Continue...
This CFA subject is certainly emotive...
But what ever its advantages or disadvantages, its a sad thing that the main amplifier design books pretty much avoid the subject. It would appear that the diamond input is the easiest (most forgiving) for design beginners to deal with. Or at least this is my experience. Only now after several "CFA"s can I begin to deal with "VFA"s and get usable results.
P.S. The thing I find the funniest is that it was Self's "Audio Power Amplifier Design Handbook - 5th (Page 134)" Edition books that got me into CFAs.
This CFA subject is certainly emotive...
But what ever its advantages or disadvantages, its a sad thing that the main amplifier design books pretty much avoid the subject. It would appear that the diamond input is the easiest (most forgiving) for design beginners to deal with. Or at least this is my experience. Only now after several "CFA"s can I begin to deal with "VFA"s and get usable results.
P.S. The thing I find the funniest is that it was Self's "Audio Power Amplifier Design Handbook - 5th (Page 134)" Edition books that got me into CFAs.
This discussion on what CFA is and is not is in full swing again. Interesting! When I looked around for clear definitions, I found a whole bunch of app notes with closed loop gain derivations (TI, Intersil, ADI, Natsemi, etc). I could not find one that clearly explained the differences simply and most importantly, intuitively. The fact is that we are still at it. I wrote a short article that in 99 % of cases makes accurate I identification possible. Read the article, apply the two tests and two pointers and lets MOVE ON!!!
http://hifisonix.com/cfa-vs-vfa-a-short-primer-for-the-uninitiated/
http://hifisonix.com/cfa-vs-vfa-a-short-primer-for-the-uninitiated/
Last edited:
Sorry, this is incorrect. That would be true only if Rf>>Rshunt, that is, for large closed loop gains. For low closed loop gains, which is what most of the CFA industrial applications are focused on (see DSL drivers, etc...) the effect of Rf on the closed loop gain cannot be ignored. If you break the feedback loop, the inverting (low impedance) input is loaded by Rshunt||Rf and the output is loaded by Rshunt+Rf (which can be usually ignored). Therefore, the open loop gain (and open loop unity gain frequency) is controlled (in such applications) by both Rf and Rshunt, which is a fundamental property of a CFA.
How the feedback network loads the open loop amplifier is in any EE feedback theory book (presented with each of the 4 types of feedback).
Sorry, your short article was already proven as incorrect, by providing counter-examples.
Amen. Bonsai -
Its done very well here and been done for decades in industry. Explained well enough IMO. Plenty of other written works out there for the finer point, details and description.
Check!
When I saw this was going back to before the beginning, I thought... OK. This is done. Its all there. Time to do something else.
Back onto designing high power Current-mode operating amps. 🙂
THx-RNMarsh
Its done very well here and been done for decades in industry. Explained well enough IMO. Plenty of other written works out there for the finer point, details and description.
Check!
When I saw this was going back to before the beginning, I thought... OK. This is done. Its all there. Time to do something else.
Back onto designing high power Current-mode operating amps. 🙂
THx-RNMarsh
Last edited:
Making a distinction based on the behavior of the ULGF is dubious because that does not reflect DC topology. What about an amp that behaves like a VFA at DC and a CFA at RF?
I think distinctions based on non-DC behavior should be considered topologies of the stability compensation, which can be substantially independent of DC topology.
AFAIK there is no need to use ULGF to determine whether it's a CFA or not.
I think distinctions based on non-DC behavior should be considered topologies of the stability compensation, which can be substantially independent of DC topology.
AFAIK there is no need to use ULGF to determine whether it's a CFA or not.
ULGF (unity loop gain frequency) or OLUGF (open loop unity gain frequency)? For ULGF, you are right, for OLUGF you are wrong.
The DC vs. AC behavior was used in this thread, by yours truly, in one example to show that Bonsai's short article description is incorrect.
[42] D. F. Bowers, “The so-called current-feedback operational amplifier...technological break-trough or engineering curiosity?,” in Proc. IEEEInt. Symp. Circuits and Systems, Chicago, IL, 1993, pp. 1054–1057.
Everybody should read this article. Not because it supports my view(s), but because it really demystifies the picture behind the CFA marketing stamp.
Everybody should read this article. Not because it supports my view(s), but because it really demystifies the picture behind the CFA marketing stamp.
If you are referring to my write-up, ULGF is not used as the determinant of the topology type.
Bob Cordell wrote "I'm not convinced that that description of whether something is a CFA is sufficiently definitive. The key, I believe, is that there is a continuum of behavior between CFA and VFA."
In discrete power amps, you can certainly comp a CFA so that it has the same ULGF as a VFA, but to do do means you don't exploit some of the performance attributes of a CFA.
We also had a lot of discussion on this thread and OS and Damir have shown some very high performance (I.e low distortion) designs. But, if you raise the OLG, in most cases that will require a trade off wrt to ULGF in order to preserve adequate phase margin.
It seems to me that with a CFA you need to make a conscious choice to go for simplicity if you want to the benefits of speed and bandwidth. The trade off is higher distortion - but IMV that's just one of 3 or 4 important parameters - it's not the most important.
In discrete power amps, you can certainly comp a CFA so that it has the same ULGF as a VFA, but to do do means you don't exploit some of the performance attributes of a CFA.
We also had a lot of discussion on this thread and OS and Damir have shown some very high performance (I.e low distortion) designs. But, if you raise the OLG, in most cases that will require a trade off wrt to ULGF in order to preserve adequate phase margin.
It seems to me that with a CFA you need to make a conscious choice to go for simplicity if you want to the benefits of speed and bandwidth. The trade off is higher distortion - but IMV that's just one of 3 or 4 important parameters - it's not the most important.
Thanks jcx
I'll take a look tonight and add them to the references.
I gather from one or two of the titles that the debate was (is?) alive in academic circles.
Last edited:
I was just thinking aloud since I've seen the AC behavior brought up a lot in CFA discussions.
What if you took a BJT VFA, and increased Rf/Rfshunt so that the feedback resistance was much higher than the base resistance of the negative input? Since Ie and Ib are proportional, this would behave like a CFA. So it's a CFA isn't it?
I'm not referring to anything, just asking questions based on the discussion.
What if you took a BJT VFA, and increased Rf/Rfshunt so that the feedback resistance was much higher than the base resistance of the negative input? Since Ie and Ib are proportional, this would behave like a CFA. So it's a CFA isn't it?
I'm not referring to anything, just asking questions based on the discussion.