On Tue, 17 Sep 2002 18:38:33 -0600 John & Jan  writes:
> Hi Tom,
> 
> I'll bite.  How is it done?

Some time ago I was pondering weak and weary over
many a bit of forgotten lore, when I nodded, nearly
napping, suddenly there came a tapping, a rapping
upon my cranial door. 'Twas only this, and nothing
more:

I always liked the sensor illustrated in some old
Scientific American article (forgotten lore), but felt
that using the plates as antennas rather than capacitors
was a drawback for too many reasons to put forth here.
In any event, the force balance approach recovers
many of the potential drawbacks because it (ideally)
minimizes excursions so that reasonable linearity can
be retained in an inverse-square environment, and the
acceleration data is easily integrated to displacement.

But it occured to me that if one were to use the plates
as capacitors in a pair of tuned circuits rather than as
antennas, it would be a simple matter to "lock" these
two tank circuits in phase to a very high degree of
precision using a couple of counters (just like a freq
counter) and apply any offset into a force feedback
loop. One could, of course use moving cores in
inductors as well, and that was what really got me
thinking. If you refer to my modest little arcticle that
was posted on the PSN web site a couple years ago,
(Nice little seismograph for children) you will note
that I used a "solenoid" type magnet and coil design
that worked quite well for several years. But the
"child" is now a hopeless Internet IM'er beyond
salvage, so the seismometer was once again available
for malevolent modification.

The first experiment involved hoisting it through the
open roof of my laboratory during a thunderstorm
while Jacob's ladders snapped away next to the
bubbling flasks...however the sensitivity seemed a
bit reduced when thus suspended. Next, I played
around with an SCR exciter and some capacitors
and found the coil could be resonant at around
17 Khz with the magnet core half inserted. I then
threw together a simple "Twin Tee" oscillator and
observed the freqency excursions with my frequency
counter and determined the sensitivity to be useful.
Next, I made a crystal controlled "divide by N"
frequency standard and tuned the detector to match,
and added a buffer/limiter to the detector oscillator
to convert the signal to a TTL square wave. I also
capacitively "decoupled" the detector tank circuit.
Finally I fed both signals to a pair of Max/Min
10 bit counters to race each other to the finish.

Now, how it operated was thus. When both counters
are at maximum they are both reset. Whichever
counter reaches maximum first applies a voltage
(integrated) opposite that of the other counter to
a current amplifier that applies DC to the coil to
restore equilibrium. That same DC was, of course
used as acceleration data and integrated in my
little adding-machine paper recorder.

That original experiment breadboard is currently
hacked up in preparation for phase two, which is
to use two coils and an iron bar with two oscillators,
thus avoiding the fixed reference and allowing all
temperature effects to be nulled out, as well as the
precision tuning required to synchronize with the
crystal standard. Depending upon how far you
want to divide your oscillators down, you can also
adjust the freqency response and "sampling" rate
to an incredible degree while achieving displacement
sensitivity that is difficult to calculate.

I, for one, have never been one to waste time with
theoretical performance and instead simply like to
see how the traces compare...and these have been
at least as good as any I have made previously, with
the added benefit of not having to constantly adjust
the stupid thing other than mechanical leveling for
zero offset current from time to time. Even so,
"steady state" offset is essentially DC and is filtered
out and not a problem with my recorder, so it is
only the non-complementary response to larger
excursions that is noticed in the traces when the
seismometer is tilted too much.

I guess I'll crawl back in my bunker for now and
keep finishing off my remaining Y2K supplies...

Tom
__________________________________________________________

Public Seismic Network Mailing List (PSN-L)