PSN-L Email List Message

Subject: noise and sensitivity
From: sean@...........
Date: Mon, 4 Dec 2000 15:31:33 -0600 (CST)

Some more comments on long period noise.

First,  I think that everyone on the list would be better able
to understand the problem if you started a query about your own
seismometer by briefly describing it.

eg: "I am operating a Lehman with a 15 second period."
If you are asking about amplification, etc, describe the coil/
magnet constant, damping, and amplifier gain. These would help
others relate to your enquiry.
eg: "The 1000 ohm coil has a constant of 25 Volts/m/sec and is 
damped at critical with at 1500 ohm resistor, etc."

Speaking of damping: dash-pot (oil, air) damping was used only on
completely mechanical seismometers like the Wiechert. Once electromagnetic
generator coils and galvanometers were used, resistive damping was
used, although some copper vane/eddy current arrangements were still 
used to avoid loss of the weak output if the early iron magnets.

With restive damping, the EXACT amount can be calculated, rather than
trying to fit curves and over-swing ratios, which generally involves
visible motion of the mass which is actually not characteristic of the
micron to sub-micron motion of real quakes, since the linearity
approximations become small arcs, both vertical and horizontal
for a horizontal "garden gate" seis. A resistive damping value about
equal to the coil resistance gives a broad overdamped response that
will make the seis sensitive to P and S waves as well as much longer
surface waves. THis will make the seis sensitive to both near field
(lt. 500 km) events as well as teleseismic events. However, this
causes a loss of half the signal voltage: the output is proportional
to [Rd / (Rc + Rd)], where Rd is the damping R, and Rc is the coil R.

However, a loss of 6db (50%) is no problem for modern amplifiers,
whose gain switches usually select in 6db (x2) steps. All the long
period (Sprengnether S5100) seismometers that I currently operate have
500 ohm coils damped with a 500 ohm resistor. This results in a semi-
broadband response (a broad arc of about 20db (x 10) amplitude) from 
1 second to 60 seconds for a 15 second Tn seis.

Someone mentioned having the seis overswing of more than one cycle.
This is not damping at all, and results in a response that is very
peaked at the natural period, and greatly reduced elsewhere. It
will be sensitive mostly at the mechanical period, like for teleseismic
surface waves (and local tilt noise), with little sensitivity to
everything else. A critically or, better, an overdamped seis is much
less sensitive to tilt noise, since the causative transient will 
usually pass before the seis has responded very much.

Regarding recording sensitivity: if you aren't seeing the 6-second
microseism background, you won't see many quakes. Of course this
varies by your location; most are induced by low pressure storms
off the northeastern continental shelf, and propagate through the
eastern USA upper crust with little attenuation. Today at St. Louis
they are running about 5 millivolts P-P from the VBB-STM seis with
an output of 1600 Volts/m/sec, or 1.6 mv/micron/sec. So the amplitude
at 6 seconds is about 3 microns. During storms they may be 10 times this.
This is with a digitizer of 200 mv full scale and 0.1 mv least count.

So setting the amplifier gain depends on your coil constant and the
range of your digitizer. If the constant is 50 V/m/sec, and the amplifier
gain is 100 (40db), this is 5 millivolts/micron/second. A 10 volt full
scale 16 bit digitizer has a least count value of 0.3 millivolts. So
the sensitivity is 16 counts/micron/second, and a 3 micron P-P 6-second
microseism will be about 50 counts P-P. Of course, if your coil constant
is only 5 V/m/sec, the microseisms are only 5 counts, and you might
have to use 10x more gain to see them with any resolution.

Regarding noise on windy days. With an un-enclosed vertical, this
is mostly mass buoyancy in the barometric pressure variations. With
a long period horizontal, the noise is due to tilting because of 
differential loading of the surface of the earth due to the gradients 
in the barometric pressure that are causing the wind. Wind blowing on 
trees makes both vertical and horizontal surface waves but only in the 
very near surface, which is one reason why installing the sensor at 3 m
or greater depth is a big improvement.


Public Seismic Network Mailing List (PSN-L)

[ Top ] [ Back ] [ Home Page ]

Larry Cochrane <cochrane@..............>