Dave,

Unless you have a very wide range digitizer (like 24 bits), any 
visible motion of the seismic mass should be grossly overloading
your system when you are operating at a decent sensitivity. This 
could only be done properly if a carefully designed attenuator is
installed between the seismometer and amplifier. We even need to
use such an attenuator when using our hydraulic shake table with
L4-C seismometers for telemetry stations. Even though the table
range is 0.1 to 100 microns, we use a 40db (100:1) attenuator to
keep the data from clipping in the amplifier. We cannot directly
see the 100 micron (0.1 millimeter) motion at 1 hz, but we can feel 
it at 10 hz. (We use a 1000:1 optical lever indicator and an LVDT
to monitor the table displacement.)

Determining the calibration of the sensor coil only requires a simple
multimeter, which is essential for almost any seismometer work,
and a calibrated weight set. You can buy a 1 gram to 500 gram set
from McMaster for $52, or just 5 weights, 1 gram thru 20 grams,
(all you should need for a seismometer) for $15. This will allow
the sensitivity of the coil and magnet to be determined by several
methods, most commonly by applying a force with a weight and then
restoring the resulting offset with a current. The resulting 
sensitivity of Newtons/Ampere is the same as the velocity output
of Volts/meter/second. For example, if 1 gram is balanced (by using
a potentiometer and battery thru the meter) by 1 milliampere, the 
sensitivity is (0.001 kgram / 0.001 Ampere)*9.8 m/sec^2 = 9.8 N/A 
or 9.8 V/m/s. The potentiometer can be a 10k ohm volume control 
from Radio Shack, used with an AA battery and clip leads. 

The normal background noise for a seismometer like a Lehman are the
6-second microseisms, usually caused by storms off the east coast.
Away from the immediate shore (100km) these run 2 to 4 microns peak-
to-peak, but can be ten times that during a hurricane or a Nor'easter.
For a seismometer sensitivity of 100 V/m/s, an amplifier with a gain
of 100 is needed to raise the signal from the microseisms to about 20
millivolts. This would provide about 100:1 signal to noise if the least 
count of a 10 volt, 16-bit digitizer is about 0.30 millivolts.

Also note that how well you record a quake depends on your relationship
to the source mechanism. Earthquakes are rarely point sources of energy
that propagate evenly in all directions. They are called "double couple"
sources because the energy radiates predominately in two directions.
For the common right-lateral fault, like the San Andreas or New Madrid
systems, the fault on the east side moves north, and the west moves
south. Thus there is much less energy radiated east and west. The 
converse is the situation for a north-south thrust fault: the east moves
up, or the west moves down, radiating less energy North and south. This
is why magnitude determinations can vary by half a unit from an array
of stations. And why sometimes we record much less than we expect.

The usual way to study the local background noise and the relative
sensitivity of a station and site is to compare the power spectral
density to the USGS noise models. This is a bit more complex, but I
could review it if necessary. There are some examples on my web site.

If you need me to repeat the rest of the calibration info, let me know.

Regards,
Sean-Thomas
__________________________________________________________

Public Seismic Network Mailing List (PSN-L)