"Tim" wrote in message
news
...
But that's the sort of thing that you need to ask. I once went to a
presentation by Litton of their fiber optic gyro product. The
engineering rep gave a technical talk, starting with the pitch "This
wants to be a gyro, an accelerometer, a thermometer, a microphone, and
many other sensors all rolled into one -- we had to do some work at
compensation." (For those of you scratching your heads -- it's a
system's engineer's in-joke. Every sensor you try to make responds to
_everything_ -- it's your job to try to figure out how to build it to be
inherently insensitive to everything else, or to independently measure
the other factors and null them out. It's why dedicated systems
engineers are bald).

--
Tim Wescott

I encountered that problem when I built a fast picoammeter to measure the
millisecond-scale relaxation time constant of dielectric absorption in FR4,
wire and reed relay insulation. It detected any sounds in the room including
the ultrasonic alarm system, which we didn't know about and anyway should
have been well above its bandwidth. The Feyndishly brilliant engineer
calculated that the cover panel had to move only about the diameter of an
atom to produce the capacitive coupling we measured. I fixed it by creasing
the panels to increase their fundamental resonance.

jsw