A few random thoughts.

I agree that the ball bearing/hardened plate is extremely low friction
and has well defined pivot motion, but I consider that it has a major
fault if used in a potential hi-G (read local earthquake) areas such as
California, and that is the possibility of the pivot slipping since
there is very little constraint to forces parallel to the pivot surface,
horizontal or vertical.  With a modern 24 bit A/D, you could shoot for a
dynamic range of about 16,000,000 to one, which somewhat equates from
noise floor to a 7.0 earthquake - bigger than I hope I ever experience. 
I may be way off base, but it seems to me the use of a 24 bit A/D over a
16 bit opens up the possibility of both a sensitive distance detector,
but also capable of recording local strong motion during such events. 
If true, then the possibility of the pivot jumping or sliding is
important to consider.

The answers that have been posted to my original question about
stability seem to fall in one of two camps - pivot related and local
soil condition (tilt) related.  I subjectively thought this to probably
be true, but it seems to be confirmed.  I've always been partial to the
Rolamite style bearings with regard to well constrained motion.  In
practice their loss should be no more than a foil hinge of equivalent
foil width.  With the additional benefit that the foil spring force is
balanced, so that doesn't enter into the equation, only the hysteresis
loss.  The big problem with them in this application is that if you trap
dust in between the bands and the rollers, they probably will act badly
from a seismograph pivot viewpoint.  However to provide the temperature
control, the seismograph is pretty much enclosed and could be made dust
proof, so this may be a viable approach.

Now for a way out concept - and be prepared to shoot it down!  A
possible solution for making a self adjusting Lehman or similar pendulum
system that can ignore slow soil tilt changes.  Imagine floating the
entire Lehman on a boat-like platform in water.  Two problems jump to
attention: 1) the damping of the 'boat' in the water is terrible and 2)
now how do we transmit the ground motion to the seismo?  The simplistic
answer is float the seismograph in Silly Putty (Dow Corning 3179
Dilantant Compound).  It is a 'dilatant" compound as opposed to a
thixotropic compound, i.e., it flows very slowly, but hardens with fast
shear/force.  (see 
http://home.earthlink.net/~gschenberg/puttyfaq.htm) 
So when fast motion occurs to the tub of Silly Putty, the seismo would
be carried along, just what is needed.  A variation to conserve the
amount of Silly Putty would be to have 'outriggers' on the seismo which
reach over the tub of water and stick into small amounts of Putty.  Then
the seismo motions would be coupled through the Putty into the
'outriggers' and activate the seismograph.   Now as the earth tilts over
periods of hours, the seismo self-levels, never requiring readjustment. 
However, realize that the amount of putty necessary does not have to be
very much if the 'boat' and the tub are very close fitting, you only
have to provide a gap that is larger than the tilt which in practice is
probably only a few tens of thousandths and tub height sufficient
to'float' the boat.  The putty SP is 1.14, slighty more than water.  So
build and test with water, then replace with putty, and you should float
just a schosh higher.  Just a wild thought - any takers?

(Incidentally the site quoted above, but at:
http://home.earthlink.net/~gschenberg/sillyput.htm
does bulk buys and will sel you Dow Corning 3179 Dilantant Compound in
poundage quantities at considerably better prices than buying Silly
Putty at the toy store.)

Regards,
Charles R. Patton
__________________________________________________________

Public Seismic Network Mailing List (PSN-L)