Re: [Phys-l] Kirchhoff's laws, or not (was: arcing ...)

[Stefan Jeglinski <jeglin@4pi.com>]

> Both of Kirchhoff's laws
> are flatly contradicted by the Maxwell equations, and
> both are routinely violated in practice.

As a catch-all term, I assume you are referring to the "radiation" 
aspect? Or am I completely missing the point? (a common happenstance)


>   2) Ground loops are a familiar example of a gross
>   violation of K's voltage law.

I have thought of this also as a radiation issue. In short, 
radiation(s) induces [presumably] unwanted current(s) that exhibit as 
noise on top of the "wanted current(s)." K's voltage law is violated 
in the same way as that for a current loop that has no voltage drop?

In the case of the copter, I acknowledge your earlier observation 
that the situation is AC rather than DC. It is easy enough to wave my 
hand and say that the copter is merely an antenna, but the numbers 
need to work out too - I am uncertain as of this writing, and more 
interested anyway in the following:


> This is important in the real world.  Anybody who does
> any kind of low-noise instrumentation probably needs
> to deal with violations of Kirchhoff's laws on a daily
> basis.  This includes physics-lab instrumentation as
> well as other stuff such as medical instrumentation.
>
> There's a lot more that could be said about this if
> anybody's interested.

I am, driven by this (admittedly wordy) anecdote:

We once inherited a competitor's hardware from a customer that 
replaced theirs with ours. We noticed a funny bulge in a cable and 
took it apart, noticing that the shield had been broken and then 
spliced with a 100-ohm resistor. With some investigation, we found 
that they built all their cables that way. Later, when designing a 
new piece of instrumentation, we had a pesky noise problem that we 
had never encountered. I can provide a schematic if necessary, but I 
roughly describe it as a detector flanged to a vacuum column, in 
which the detector was isolated from the column. We provided the 
ground potential via the shield from our instrumentation box. At 
least on the DC level, we asserted that there were no ground loops. 
On an AC level, this can be hard to be certain of. Our previous 
designs had not encountered such a noise issue, but this one included 
a 3rd-party ADC card as opposed to an ADC card we had previously 
designed and built ourselves. Despite having the schematics and the 
ear of the 3rd-party mfgr, and a reasonable belief that there 
"should" have been no problem, there was. On a hunch, we separated 
the shield right before the signal came into the 3rd-party card, 
spliced it with a 100-ohm resistor, and... noise gone.

OK, so our discussion went along the lines of "we must have broken 
some ground loop." But we couldn't *identify* a ground loop, and if 
you "disconnect" a ground loop and reconnect it again, why is it not 
still a ground loop? Analysis: the 100-ohm resistor can be thought of 
as a large impedance compared to the shield impedance, but I'm not 
convinced that's effectively an infinite impedance. But if you think 
of that as *breaking* the ground loop, you also have to think of it 
*isolating* part of your shield (ie leaving it to float).

Ground loops have always been confounding beasts - AC matters as well 
as DC (eg impedance vs pure resistance), and they find clever ways to 
exist when you presume they don't.


Stefan Jeglinski, often mystified