diyAudio (
-   Solid State (
-   -   Is the CFB topology superior, and why? (

Elvee 28th August 2012 05:20 PM

Is the CFB topology superior, and why?
4 Attachment(s)

I would like to investigate about the actual/perceived/imagined superiority of current feedback, and possibly to pinpoint what exactly makes it superior (if indeed it actually is).

As a study subject, I propose to use a very simple, quasi-canonical form of a CFB amplifier, the CFP with gain.

The two first pics show the linearity and frequency performances of the circuit.

The interesting thing with the sim is that it allows the transformation of this ideal CFB amp into an identical ideal VFB stage.
That can be done with an arbitrary voltage source (the next two pics).

As could be expected, the linearity has improved thanks to to the increased loop gain, but somewhat more surprisingly the bandwidth also increases.

Elvee 28th August 2012 06:55 PM

1 Attachment(s)
Now that we have noticed the differences, we can begin to try to even them out, to try and see where the two versions diverge.

Let's see the effect of the loading on the feedback point, by adding an equivalent circuit, while retaining the VFB configuration: it begins to look more like the CFB incarnation, but do not be fooled by the graph: LTspice has automatically scaled it, and the 1GHz intercept is at -4dB instead of -25dB.

richie00boy 28th August 2012 07:23 PM

Isn't R2 in the wrong place?

Elvee 28th August 2012 07:46 PM


Originally Posted by richie00boy (
Isn't R2 in the wrong place?

I don't think so. This rather crude first tentative does have issues, but normally R2 is not part of them. I could be mistaken of course. This is an exploration trip, and I do not know where it will lead us

Elvee 28th August 2012 08:08 PM


Originally Posted by richie00boy (
Isn't R2 in the wrong place?

OK, I should have explained more clearly what I did exactly.
I thought the diagrams were sufficient, but that is not necessarily the case for people unfamiliar with spice.

I started with a classical CFB circuit; I then buffered the feedback point using a spice ideal buffer: a voltage source that simply copies its input, without loading or delay.
I then attempted to reproduce the behavior of the initial CFB circuit by loading again the feedback point with a (not so!) similar load.
That's where we currently are

Elvee 29th August 2012 05:10 PM

1 Attachment(s)
The first attempt to revert to the CFB characteristic by loading the feedback divider has failed.

The reason is simple: in the real CFB, the impedance is in fact different, because Q1's base is not at a fixed potential: it follows the signal, and thus bootstraps the emitter's impedance.

This in itself is quite interesting: it means the so called CFB is not really a pure CFB: it also displays attributes of VFB.
In order to have a really pure CFB, we would need to sufficiently reduce R1 to make R3 unnecessary.

So, back to our CFB-like VFB, all we need to do is to equalize the base potentials of Q1 and Q3.
And this time, it works:

Elvee 29th August 2012 06:15 PM

We can now try to confirm what we discovered: the canonical CFB is in fact a closet VFB circuit.
We just have to compare the magnitude of the currents in the sensor transistor, Q1 and the output divider: (the current into Q1 is 72µApp)

Yet, this very light loading is sufficient to cause significant errors, as we have seen earlier.

Elvee 29th August 2012 07:52 PM

1 Attachment(s)
It looks like I have forgotten to post the pic, here it is:

Nico Ras 29th August 2012 08:01 PM

Elvee, I am sorry to see that you are a little lonley, to some it is clear but maybe some of audience don't quite understand what you are proving or disproving.

I also think that newer folk is afraid to even ask what is being discussed not understanding the difference or what CFB actually tries to accomplish.

Jay 29th August 2012 08:34 PM


Originally Posted by Nico Ras (
maybe some of audience don't quite understand what you are proving or disproving..

Yes. Right now I see it like Elvee is comparing apple to an orange. Not sure whether he is trying to prove which one is more spherical than the other, or more yellowish... But I'm waiting... for the conclusion :D

All times are GMT. The time now is 11:16 AM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio