K7ITM wrote:
. . .
The analogy may not be prefect, but I think it's a lot like the
usefulness of the idea of a "virtual ground" at the inverting input of
an op amp. But it's a virtual ground only under specific conditions:
strong negative feedback is active, and the non-inverting input is at
(AC, at least) ground potential. For it to be a useful concept
without too many pitfalls, the person using it has to be aware that
the conditions that make it a good approximation don't always hold.
Similarly for a "virtual short" on a line.
. . .

Let me relate a story. . .

Years ago at Tektronix, I transferred to a different group. Across the
aisle was a very bright engineer, fresh from school with a Masters or
PhD degree -- I don't recall which. I recall that his advanced education
had specialized in nonlinear control systems, very much a mathematically
complex and challenging field. His entirely academic background had been
very different from mine, so I was often enthralled by his attempts to
reconcile reality with the idealized world he had, until very recently,
occupied.

One day I found him muttering, trying this and that, until he asked for
some help with his test setup. He was driving an inverting op amp
circuit with a square wave, and he was seeing sharp pulses at the
"virtual ground" summing junction with his 'scope. He had tried moving
his probe grounds, replacing the op amp, bypassing, and everything else
he could think of, to rid his display of this obvious erroneous
artifact. The voltage at the summing junction, he explained, should
always be zero, since it's a virtual ground. Those spikes shouldn't be
there.

I tried to explain to him how a "virtual ground" was created: An input
signal initially generates a voltage at the op amp input. The op amp
responds by sending an inverted signal back to the summing junction
which adds to the initial voltage to produce very nearly zero at the
input. I explained that it can never be zero, but at best is the op amp
output voltage divided by its open loop gain. But more to the point, the
op amp isn't infinitely fast, so it takes some time for it to respond to
that initial voltage or any changes. And during that lag, the summing
junction voltage can move a great distance from zero. So the spikes are
occurring during the time it takes the op amp to respond to changes in
the input stimulus.

Well, he didn't get it. He just couldn't make the transition from the
idealized, infinitely fast and infinite gain op amps of his academic
models to the real things he had to work with. And looking back on it, I
think the basic problem was that he never really fully understood just
how that virtual ground came about even in an idealized world.

After a number more frustrating and unresolved collisions with reality,
he wisely quit and got a teaching job. I'm sure he did well in the
academic world.

Those many models we use daily to keep calculations, concepts, and
analyses manageable are just that -- models. It's imperative to
constantly be aware of the range over which those models are valid, and
alert to any situation which might make the model invalid. People
solving routine problems can, unfortunately, often get along for a long
time without realizing the limitations of their models, and can be
lulled into a belief that they're not models at all but reality. But in
the environment where I've spent most of my time, this carelessness
leads you very quickly into places which can be very difficult to get
out of.

Roy Lewallen, W7EL