Randall,

Thanks for the parameters.

This and your previous comments have gotten me=20
thinking about seismometer performance in=20
general.  I would like to propose that the=20
fundamental parameters of a seismometer design=20
are 1) the amount and spectrum of its=20
internally-generated noise, stated in terms of=20
ground motion, and 2) its clipping level.  The=20
frequency response and sensitivity are really=20
secondary since, given appropriately low noise,=20
the digital output data can always be deconvolved=20
to appear to be coming from an instrument with a=20
response which is constant at all frequencies,=20
and then scaled, filtered and otherwise digitally=20
manipulated to duplicate the signal as it would=20
be coming from any other instrument having a=20
differing but known frequency=20
characteristic.  That in general is how the=20
internal noise spectra of different types of=20
instruments can be properly compared.

Regarding tilt measurements, I feel that this=20
might a good example of the old saying "One man's=20
signal is another man's noise" (no gender=20
preference implied).  To the seismologist looking=20
for weak teleseisms, tilt effects on his=20
horizontal channels would likely be considered to=20
be noise, while to someone studying tilt and=20
other long term effects, an earthquake might well be his noise.

Regards,
Brett


At 03:58 PM 2/8/2011, you wrote:
>Brett,
>     My two channel instrument (orthogonal=20
> pendulum pair) is operating at 24 bits on each=20
> channel.  The upload setting (for gif=20
> transmission) to Larry=92s server is the value=20
> 100 for each of channel 1 (E-W) and channel 2=20
> (N-S).  The integrated output from each of=20
> these is using the value 500.   The integrated=20
> channels are being bandpass filtered (single=20
> pole, each corner frequency)=ADrange 10 mHz=20
> (period 100 s) to 500 mHz (period 2 s).  If you look on
>http://www.seismicnet.com=
/currentseismicity.html=20
>at the Macon signals corresponding to the raw=20
>data for E-W and N-S, you will see that there is=20
>no high pass filtering acting on them.  These=20
>show the present tilt offset (un-zeroed to d.c.=20
>output) of -23000 capacitance to digital=20
>converter (CDC) counts in E-W and 4500 CDC=20
>counts in N-S.  In other words, for the=20
>calibration constant of 2.5 billion CDC counts=20
>per radian, the E-W pendulum is offset by -9=20
>micro-radians and the N-S pendulum offset is 2=20
>microradians.  It is difficult to set up these=20
>pendulums to better than one microradian=20
>offset.  With the high-pass filter this is of no=20
>consequence.  But I want to look for=20
>displacements with characteristic periods of=20
>hours, days and months.    The VolksMeter does=20
>this nicely in the mode I=92m using --functioning=20
>as a tiltmeter with no low frequency limitations=20
>other than the stability of the pier on which it=20
>stands.  The pier is a monolithic 4 ft x 4 ft x=20
>2 ft rectangular parallelipiped with a bottom=20
>cylindrical piece (dia. 20 in) that goes 20 ft=20
>into the ground.  The structure is high-strength=20
>concrete in a reasonably isothermal enclosure,=20
>and isolated from the building slab vibrations=20
>(such as from air conditioning).  Only with a=20
>decent pier can thermal-induced, diurnal cyclic=20
>tilt changes of aggravating type be avoided.


Randall,


Thanks for the parameters.


This and your previous comments have gotten me thinking about seismometer
performance in general.  I would like to propose that the
fundamental parameters of a seismometer design are 1) the amount and
spectrum of its internally-generated noise, stated in terms of ground
motion, and 2) its clipping level.  The frequency response and
sensitivity are really secondary since, given appropriately low
noise, the digital output data can always be deconvolved to appear to
be coming from an instrument with a response which is constant at all
frequencies, and then scaled, filtered and otherwise digitally
manipulated to duplicate the signal as it would be coming from any other
instrument having a differing but known frequency characteristic. 
That in general is how the internal noise spectra of different types of
instruments can be properly compared.


Regarding tilt measurements, I feel that this might a good example of the
old saying "One man's signal is another man's noise" (no gender
preference implied).  To the seismologist looking for weak
teleseisms, tilt effects on his horizontal channels would likely be
considered to be noise, while to someone studying tilt and other long
term effects, an earthquake might well be his noise.


Regards,

Brett




At 03:58 PM 2/8/2011, you wrote:

Brett,

    My two channel instrument (orthogonal pendulum pair)
is operating at 24 bits on each channel.  The upload setting (for
gif transmission) to Larry=92s server is the value 100 for each of channel
1 (E-W) and channel 2 (N-S).  The integrated output from each of
these is using the value 500.   The integrated channels are
being bandpass filtered (single pole, each corner frequency)=ADrange 10 mHz
(period 100 s) to 500 mHz (period 2 s).  If you look on


http://www.seismicnet.com/currentseismicity.html at the Macon signals
corresponding to the raw data for E-W and N-S, you will see that there is
no high pass filtering acting on them.  These show the present tilt
offset (un-zeroed to d.c. output) of -23000 capacitance to digital
converter (CDC) counts in E-W and 4500 CDC counts in N-S.  In other
words, for the calibration constant of 2.5 billion CDC counts per radian,
the E-W pendulum is offset by -9 micro-radians and the N-S pendulum
offset is 2 microradians.  It is difficult to set up these pendulums
to better than one microradian offset.  With the high-pass filter
this is of no consequence.  But I want to look for displacements
with characteristic periods of hours, days and months.   
The VolksMeter does this nicely in the mode I=92m using --functioning as a
tiltmeter with no low frequency limitations other than the stability of
the pier on which it stands.  The pier is a monolithic 4 ft x 4 ft x
2 ft rectangular parallelipiped with a bottom cylindrical piece (dia. 20
in) that goes 20 ft into the ground.  The structure is high-strength
concrete in a reasonably isothermal enclosure, and isolated from the
building slab vibrations (such as from air conditioning).  Only with
a decent pier can thermal-induced, diurnal cyclic tilt changes of
aggravating type be avoided.