PSN-L Email List Message

Subject: seismometer sensitivity
From: S-T Morrissey sean@...........
Date: Mon, 16 Aug 1999 17:32:23 -0500 (CDT)


Regarding the questions of the sensitivity of various seismometers,
particularly compact 1-hz and shoter period phones:

There is no unique answer to the question, since they are all
manufactured to meet a wide range of data and recording requirements.
THe big variable is the coil resistance, which is kept low (hundreds
of ohms) for galvanometric recording or long field strings, but is 
made high (5500 ohms for a typical L4-C) for amplified applications
like telemetry. THe upper limit is where the resistance is so high that
the coil cannot be used for damping because it limits the current. Some
seises (long period) have a 100,000 ohm coil for the signal and a separate
500 ohm coil for damping.

The other variable has been the use of improved magnets; some manufacturers
have recently switched to niobium-based alloys, which can provide over
5x the output from the same package.

So the best bet is to get the data for EACH phone from the manufacturer.
Most provide a label with at least the coil resistance (which you can measure,
of course) and the generator constant of either the main coil or the
calibration coil. Most have tables or graphs of the output for various
coil resistances; these are accurate to about 20%. Some show the output
level at various percentages of critical damping (this is not the same
as the open-circuit output), and even give the damping resistance (which
has to be corrected for the parallel input resistance of the amplifier).

Most geophones are sealed so that directly measuring the generator constant
by a weight-lift test cannot be done. If the calibrator coil constant is
given, this can be used to determine the output of the main coil. THen
the actual input to the amplifier with the damping resistor in place can
be determined. Years ago, we calibrated 50 or so L4-Cs and found that the
nominal output (270v/m/sec for a 5500 ohm coil) varied by 20%, so we used
a series/shunt method (per. J. Eaton of the USGS) to standardize all of
them to a damping of 0.7 and an input to the amplifier of 100V/m/sec.
The subsequent shake-table calibration showed agreement within a few percent.

Obviously a shake-table provides an immediate answer to the sensitivity
question as well as the response of the amplifier and data system. But
they are rare, so weight-lift calibration can provide absolute results
IF there is access to the moving mass. Like for the HS-10-1 with the external
calibration coil and astitizing spring (to trim the period), the test 
weights (100 to 500 milligrams) can be applied directly or via 45 degree
threads (the explanation is long and detailed) for a horizontal.

Then there is the further question of the sensitivity needed to detect
a given quake. Obviously, "cranking up the gain" is fun, but not if it
just amplifies the freeway 1 km away. But for example: with the L4-Cs
damped and trimmed to 100v/m/sec, we found that amplifier gains of 60
to 72 db (x1000 to x4000, depending on the site noise) were adequate to
record a Mb=2.0 at 20km in the New Madrid region. In the west (CA and NV),
site noise can be much less, but attenuation can be much more (up to 10x),
so amplifier gains of 84 db (x16,000) might be needed for the same result.

So what does this mean for YOUR geophone/damping/amplifier. If the L4-C
above had a net output of 100V/m/sec, and is amplified 1000 times, 
the recorder is getting a signal of 0.1 volt/micron/sec or 0.1 millivolt/
nanometer/second..  In terms of displacement, a nm/sec is 1/(2*pi)*nanometers 
at a period of one second. If your amplifier noise floor is 10mv, your
displacement noise sensitivity is 16 X 10^-9 meter for a 1-second wave.

Assuming that you can see a 1-second waveform at 10 times the noise level,
what size earthquake might be detected at this level? Using  a standard
magnitude scale "Ms = logA/P + 1.66*log(distance) - 0.18", and using a 
distance of 1 degree (about 100 km), and A is the amplitude in nanometers,
(10x the noise is 160 nm), we get a Ms magnitude of 2.02, which is a quake
very near the threshold of sensation. If the event is 2 degrees away, it
will be a magnitude 2.5. But say that you forget the preamp, but your
digitizer noise is only 1mv, so your detection level is 16000 nanometers
(16 microns), the quake at 1 degree will be a magnitude 4.02.

So even with a high-output sensor like the L4-C, a pre-amp is necessary.
4.5 hz phones will need up to 10x more gain, especially if the amplifier is
"countoured" to boost the 1-hz output.  Last year I scanned the old pencil
drawing of the pre-amp that I use, along with a photo of it to my web site
http://www.eas.slu.edu/People/STMorrissey/index.html
...: stmmisc.html" PSN INFO ... SLU Seismic Network
I have since re-drawn it with the latest details, but it is not yet posted. 
I can send the new schematic to any SASE, and if Larry wants to provide it,
I can send him the artwork (which is just that; it would probably have to
be re-drawn for a numerical control PC board shop). With a few compromises,
it should cost around $100. With micropower amplifiers, it has a current
drain of about 0.1 milliamp, so it will run for several years from a 
pair of 6-volt alkaline lantern batteries.

A final curiosity regarding the data from an S-13 from a Ms5.7 at 200km
(I'm guessing the distance). Using the magnitude formula gives a value
for logA/P = 5.38, so if P = 1 second, the p-p ground amplitude is about
240043 nanometers, or about 0.24mm. The peak velocity (at 1 hz) is 1.5mm/sec.
The standard S-13 has a 3600 ohm coil with a generator constant of
629 V/M/sec. When damped critically with 6300 ohms, the output is 400V/m/sec.
or 0.4 volt/mm/sec. So a peal velocity of 1.5mm/sec. is 0.6 volts, which
will saturate any amplifier with gain much more than 10. But the seis
has an air gap length, or p-p coil movement, of 1.9 mm before it "hits
the stops".

Regards,
Sean-Thomas

_____________________________________________________________________

Public Seismic Network Mailing List (PSN-L)


[ Top ] [ Back ] [ Home Page ]

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