From: firstname.lastname@example.org (Tom Bruhns)
Subject: Re: How to measure C+stray to 0.01pf
Date: 3 Sep 2002 14:35:42 -0700
References: <3D713180.969035D2@science.net> <email@example.com> <firstname.lastname@example.org>
NNTP-Posting-Date: 3 Sep 2002 21:35:42 GMT
email@example.com (Adam Seychell) wrote in message news:<firstname.lastname@example.org>...
> Agilent have seem to make the impossible possible. The Agilent 4285A
> Precision 75KHz - 30MHz LCR meter for example claims 10aF resolution !
> Can somoene explaing the purpose of such a ridiculous reading ?
Well, let's see. In the work I mentioned in my earlier posting, we
worried about drift on the order of 25 parts per million or so. Since
we were dealing with nominally 3pF capacitors, that would amount to
about 75aF. So yes, I can imagine some folk would care.
A bigger trick (than simply the resolution) is to make things stable
enough to actually use that resolution. I didn't say it in my first
reply to Sally's posting, but it's worth noting: a C0G cap is
generally rated at +/-30ppm/C tempco. If I recall correctly,
somewhere in its normal temperature operating range, it typically goes
through zero tempco. For a cheap part, that's pretty darned good.
Bob Pease, in one of his National analog design seminars, made mention
of that fact. If you want a resistor that does better, you'll pay
more dearly for it. If you want a precision current source with very
high stability, you're likely to pay quite a bit more for it. Voltage
sources with part per million stability are not cheap. So to measure
capacitance with very good stability, one must think carefully about
this sort of thing. Assuming the capacitance you want to measure is
really stable, how do you keep your measuring circuit similarly
stable? If you have lots of money to throw at the problem, there are
ways, but if you want to keep things cheap, how do you do it? For
"pretty good" stability, a bridge formed from low-drift capacitors
might not be too bad.