A small correction to my previous post. Rf and Rg are in the feedback loop but not corrected by feedback... rather are reference dividers...hence are worthy of being Dales or other audio grade devices.
It isn't clear if you already are aware of this, as such relates equally to 10x probe function, so I have tried to clarify this in the attachment. In the case with the open switch the division of Cf and Cg is adjusted by Cf to be equal to the division of Rf and Rg. This occurs when the time constant of Rf-Cf equals Rg-Cg. Hence when the switch is closed the low frequency and high frequency division is matched and an output square wave results as in Fig. 2. The dashed lines indicate a discrepancy in the equity of division.
In the case of an amplifier the high frequency closed loop gain is magnified equal to DC closed loop gain. Increasing high frequency closed loop gain can improve the stability margin in relation to open loop gain as preventing oscillations. As seems likely obvious to you Cf permits changes as specific to the leading edge of signals as part of psycho-acoustic tuning that seems of your intent. Square wave response isn't necessarily what you want in consideration of a complete system.
For what other reasons would I give myself to such strange practices that have been considered perverse at first sight by at least two contributors of this thread?
;-)
Last edited:
CFA case, uncompensated VAS's FFT: f1:125 MHz, f2:250 MHz, f3:500 MHz, measured with Tex on the amp's output. Such thing possible in VFA?
With fast amplifiers, you do need to keep the inverting input node capacitance to ground low. You will note this is discussed in both VFA and CFA application notes. Ditto the capacitance across (especially) Rf to prevent peaking problems in CFA or BW issues in VFA. For this reason, trying to build fast high performance amplifiers with leaded components is usually difficult.
Last edited:
The discussion is very dense. I recall that the proposals for the basic definition of feedback of amplifiers were interesting but not so easy to formulate.
I do not recall that a definition was proposed for current feedback. Did I miss it or was there not any ? Have you a link to dissipate my memory lapse ?
Yes, for the first proof of concept of an actually commercially viable ADSL channel. The first test beds had an nonviable BOM.
At the top RF2 sets the BW, how does a VFA achieve that? RF2 sets the BW because the current in it matters. In a VFA the relative current levels don't matter.
Or because the open loop gain (or h21, if you prefer the two port-formalism) depends on RF2. Then you don't necessary need to consider the currents and can treat the amplifier as a shunt output, serial input feedback topology and use the hij hybrid parameters. No need to invent a "current feedback" concept, although the "RF2 current view" is correct and convenient for analysis.
https://www.diyaudio.com/forums/sol...back-amplifiers-semantic-171.html#post5638115
You remember the story about Lucy, Charlie Brown and the football, right? Well, this thread is the football and you and I are Charlie Brown. Guess who Lucy is?
OK put this on datasheets, and YOU answer the phone. 🙂 Besides two port formalism treats only small signal behavior.
Last edited:
I forgot to mention a while back when SPICE was being discussed that one of the treasured artifacts that I have lost was a complete analysis on paper (those yellow engineering pads with the blue lines) of the uA741 by Alberto Bilotti, probably at the behest of Dave Fullagar. A massive sea of maths that few except Dr. Bilotti could muster. His closed form solutions of thermals on a die are beyond most mortals (except maybe Russians). 😉
If I could further comment the IC guys like Widlar, Fullagar, Solomon, Williams, etc. rarely if ever presented in academic formalism. In fact "Intuitive IC Op Amps from Basics to Useful Applications by Thomas M. Frederiksen" was more the norm.
I did catch Barrie Gilbert once in an S parameter faux pas on noise.
Last edited:
Of course; that’s precisely the reason why the “current feedback” concept was invented. It’s concise, intuitive, easy to put in a data sheet and easy to expand in application notes. The fact that the CFA concept collides with the academic curriculum in the last century is obviously not a big concern. Side note, although the “current feedback” concept is old as dirt, I never heard of it (and I happily survived) while I was an EE student in the late 70’s. This, in despite the fact that the simplest CFA (two transistors with input emitter feedback) was a standard seminar topic.
Large signal behavior, I don’t really see the slew rate related to a general CFA concept, but more related to a specific circuit topology (common base inverting input) that by convention we call “CFA”. I don’t think it’s possible to get the same slew rate performance, within the same power budget, while maintaining the application flexibility, other than with what data sheets are calling “CFA”.
Just curious, has anybody ever seen a decompensated CFA op amp? It may sound strange, but having two degrees of freedom could have some advantages, like sliding the range of closed loop gains where the bandwidth is constant.
Last edited:
In a rigorous sense can that exist? You can't have less than 1.0pF or so on an empty pin so at some value of RF it would be stable (technically not decompensated). With a VFA it's easy again, you know, they are different.
Don’t know. I know close to nothing about CFA op amp internal compensation schemes (if any systematic approach exists).
I have that book here right next to me! The math is at the right level for me, but importantly, it’s interesting to see how a good apps engineers mind works: they make it look easy and builds confidence.
I bought my first copy in IIRC in the very early ‘80’s and lost it in all my many moves (23 times in 34 years - of which 20 were for my job) but managed to get another copy off Amazon about 2 years ago.
I used to have all the Linear Technology’ app books as well . . . Long gone regrettably.
We know it's the difference between CFA and VFA but the fundemental operation is still controlled by voltage.
The concept was never heard before 1975-1980 despite most amplifiers had not a differential input before the same date.
The inverting input is not operating in common base, its Vbe being not controlled by the current through it.
Ccomp from TIS to ground, MIC, ‘light MC’, TPC etc
But as Scott notes, in a high speed IC, if you start hanging discrete components off the internal nodes you will have to deal with parasitics all over the place.
Best just to comp an IC CFA with Rf like it says in the data sheet and be very careful with layout and feedback node and component capacitances. Ditto Rf and Rg self inductance, power rail decoupling etc.