Regarding suspension resonances in seismometers:

Doug has very nicely explained the origins of these spurious responses.

THey are essentially due to the fact that while a spring's behaviour
can be carefully described when it extends in the normal direction,
when the spring elements move other than in the normal direction, 
particularly at right angles to the main tension axis, the overall
effect is to shorten the spring in a cyclic fashion depending on the 
geometry involved. The clearest example is with a long coil spring as
in long-period vertical or "LaCoste" type suspensions. THe tension of
the spring supports the boom and mass, which have a characteristic period
determined by the usual equations. But if something causes the center
of the spring itself to move laterally, this will raise the mass, and
as the spring mass itself oscillates laterally, this spurious resonance
will be imparted to the man mass motion. Often, the oscillation of
the spring is much weaker than the main suspension behavior, so the 
large main mass will integrate the spurious resonance into a one-sided
deflection of the output. Often this integrated response is WITH IN the
passband of the data of interest, and cannot be eliminated. Fortunately,
it usually requires significant acceleration to excite it, so is not a
problem except for near field events (or noise), where strong-motion
sensors are useful for recording true ground motion.

A typical example of this is from the old 15-second long period vertical
that I am operating as a 600-second VBB in a vault under the stairs
of our building; when classes change, the troops romping down the stairs
excite the spring resonance, which causes one-sided long-period downward
(since the mass is pulled up) bumps in the record.

This is one of the reasons that leaf-spring suspensions are used in
broadband sensors; the spring has much less mass and is much more rigid
than the coil spring, so its resonances are outside the VBB passband of
the sensor.

For compact vertical sensors, like geophones, they can be a problem for
modern digital recording. The manufacturers are usually mum on these; 
the most notorious are the resonances of the 1-hz vertical L4-C at 16 
and 22 hz. THe resonances can easily be found with a shake table, shaking
a vertical in a horizontal direction. But a calibration coil can also
be used to excite them; with sufficient input the inherent non-linearities
of the suspension will cause the resonances to be excited.

Since they are a mechanical problem, little can be done other than knowing
where they are and narrowing the data passband to aviod them, or possibly
specifying the level of acceleration that is needed to excite them. 
Some efforts at visco-elastic damping have been attempted; we even had
a seismometer that was filled with a liquid freon compond to damp them.

Regards,
Sean-Thomas

_____________________________________________________________________

Public Seismic Network Mailing List (PSN-L)