Like you, I have not yet read the long - and probably plesio-academic - diatribes above (although, I will do so later, in pursuit of being a thorough and thoughtful researcher).
But a couple of thoughts about virtual shorts. First, you write:
> have not seen a configuration without negative feedback that gives you the "virtual > short." I would certainly be surprised if you proved one existed :-)
Well, you are probably aware of the class of operational amplifiers which go by the name of Current Feedback Amplifier, which I happen to have pioneered in the late 1960's while at Tektronix (I'm now working with Analog Devices, in my remote design center in Oregon). I directed the development of ADI's first fully-complementary BJT process and designed ADI's first CFA there, the AD844 - in about 1978-ish.This part is still in the catalog, paying the rent more than 35 years later - Data Sheet attached.
That's just background, to center-up the focal point.....
which is. that the inverting input of a CFA is a virtual short (without feedback) and the second input determines the voltage to which this short is referenced - just like a standard op amp. If you read the Data Sheet you will see the benefits of the type of amplifier, so I won't explain them much further. Of course, all this isn't a sales pitch, but a purely academic discussion.
The CFA front door leads to a current-conveyor, whose invention is often attributed to follow-ons who took the time to write papers about this topology, but which was in fact an (unpatented, unpublished) invention of mine, also at Tektronix, in 1965. It was precisely because the lateral PNPs of that time were poor performers that led me, after moving to ADI, to promote a genuine, high-performance complementary bipolar IC process. The AD844 was designed on that first, somewhat primitive, CB platform, which, surprisingly, we keep running in wafer fab (at Wilmington, MA) all these long years later. Now, we enjoy designing on stunning CB processes, all generations of which are dielectrically isolated by building the IC layers on glass (silicon dioxide).
The CFA sprang out of a personal desire on my part to see whether there was anything to be gained by changing the character of an op amp by giving it a current-mode input - at least on one of the two input nodes. And it led to a lot more CFAs.
So that's about it for making virtual shorts without feedback. It's actually very easy.
the notion of a virtual short at the input of a standard op amp is dangerous, mostly because it's a fantasy. Yes, the voltage swing at the input is that at the output scaled down by something called A, the "open loop voltage gain". The problem is that this parameter is frequency dependent, and should generally be thought of as A(s). Its value can be quite small even at moderate frequencies, For example, imagine one's op amp has a unity-gain frequency, F1, of 10 MHz. Many op amps use a dominant pole HF stabilization paradigm in which the open loop gain at a signal frequency of FS is simply F1/FS, eventually limited at very low frequencies to a large value for A.
So in this example, suppose FS is 100 kHz; then the small-signal open loop gain would be only 10 MHz/100 kHz = 100. Thus, a +/-10V swing at the output should, according to rudimentary theory, result in an input swing of +/-100 mV.
This is far from being a "virtual short" !!
In fact, it's worse than this, because in many cases (using simple transconductance input stages) the amplifier will be suffering from slew-rate limitations, causing this input voltage to be very nonlinear. Of course, the whole thing is useless when it gets into this condition. Amplifiers can get around this problem in various ways (using a highly-degenerated input transconductance - which comes automatically in an appropriately-designed CMOS op amp).
Even then, it is very hard to meet the 2000 V/us slew rate of the AD844 CFA. Or better still, the 4100 V/us of the AD8000, having a unity-gain frequency of 1.5 GHz. And it's only $1.70.
We've come a long way since the uA741....
We have already done some humble attempts to explain the secret of CFA:
You can tell if we have succeeded...
Question to Barrie Gilbert:
In the past, I have seen many, many contributions (filters, oscillators) dealing with "Current-feedback-amplifiers (CFA)" - and in nearly all cases the authors were using the AD844 as a CCII.
As far as I know AD is the only manufacturer of a CFA that can be also used as a CCII. Why is this so? It seems that a CFA with access to the high-impedance node is a very versatile device and some other manufacturers should also be interested.
I have read through ths discussion, starting as "Virtual Ground" but evolving far adrift, including the multiple postings over several years, regarding the "Truth about CFA", and even the "disturbed_common-collector_stage", as written by Cyril and Lutz, and referencing other postings and topics.
I believe it was Barrie who wrote that our best analysis is " only a partial model of a real, living and breathing circuit."Today, I have spent the morning perusing this Derivative, and also "the long - and probably plesio-academic - diatribes "written prior. All this, as best I can.
I sense an "either" vs "or" goal. Eiher "Voltage" or "Current" as the controlling entity.
From Philosophy, I have come to see the possibility that "neither" inifinity nor zero actually exist.
From Math, I have come to see that we have "agreed" on methods of handling infinity and zero.
In practice I have come to see that both Voltage "and" Current are always present as a system.OHM's Law requires that "both" are always parts of a system.
What I find missing in these postings is this. The discussion is about a "system" which is dynamic, and bounded by the passing of "Time". * "Ground" only has meaning in a Servo-System , controlled by components and feedbacks, in a time framework. * "CFA" and "VFA" only has meaning in a Servo-System , controlled by components and grounds, in a time framework. So, I think that if we assume that everything in these circuits happens in an imaginary "instant", then we have un-explainable and conflicting phenoms. Ohm's Law is always present ( E=IR ) and requiring that an entity is never zero nor completely in control.
This presentation, and Barrie's ever-insightful perspectives, along with Cyril and Lutz earlier and multiple postings, have been a very interesting, and demonstrates the cooperative interplay of thinking, amongst some very talented engineers. Certainly the whole story has not been told,
* I think the term "TransImpedance" was a good invention as it describes the complex dynamic "Time Bound" system.* I think the term "Virtual Ground" was a good invention as we may never see a "Real Ground".
I wonder if this discussion will evolve into a book.
> wonder if this discussion will evolve into a book.
The answer is yes; the question is When?
And the answer to that is: As soon as I can find
the time to compress a lifetime of invention into
a few hundred all-too-short pages dripping with
genuinely new and mind-illuminating ideas. As
I see it, a new book must say new things.
This is a work in progress.
Its objective - amongst others - is to generate a
refreshing breeze of fresh air, blowing away all
the academic stuff and nonsense of the sort we
are seeing in this, and other, recent threads. Its
foundations are firmly built on commonsense and
glorious intuition. Mathematical analysis is used
cautiously and is limited to simply what might be
needed at any juncture; I am determined that this
work will not lapse into idle peacock displays of
I am truly dismayed by how simple - and even
rudimentary - ideas at the entry level of circuit
design are being treated as if profound truths
that require lengthy debate. To be frank, it has
been as much as anything else involvement in
these RG discussions that have convinced me
that a groundbreaking presentation of analog
circuit principles is needed, to clear the fog.
My big problem is time. When I write a book, I
spend endless time going over each chapter
in review and refinement. I'm told that this is
not necessary - just finish the job! - but that is
not how I see it. For me, any work of this kind
must deliver completely transparent concepts
fully rid of all taints of academic peacock-play.
BITS Pilani, Hyderabad
University of Pune
Instituto de Pesquisas Energéticas e Nucleares
Alexander Technological Educational Institute of Thessaloniki
Technical University of Sofia
Technical University of Denmark
University of Adelaide
Lutz von Wangenheim
The University of Tennessee Health Science Center