PSN-L Email List Message

Subject: vbb transient noise
From: S-T Morrissey sean@...........
Date: Sat, 1 Jul 2000 15:47:00 -0500 (CDT)

Sorry I have not responded sooner, but an emergency came up when a
contractor tore up the fiber-optic cable linking the processors at
the IRIS station CCM. I've ended up replacing about 100 meters of cable,
making new "pipe vaults" for the connections that have to survive 
flooding 10 to 12 meters deep. Hopefully the additional splice in the
2.4km cable will not push our optical power budget.

Regarding LP transient noise from your VBB vertical.  This is not an
uncommon problem with broadband seismometers operating at extended
periods, like in excess of 20 seconds. THe longer the operating period,
the more sensitive the sensor is to steps in displacement, which 
show up in the velocity output as one-sided lumps with about the
duration of the integrator period. I can only guess at some possible
causes, so I will pass on my experience for everyone's benefit.

The manufacturers of commercial broadband sensors actually suggest
being tolerant of these transients during the "settling in" period of the 
instrument, which may last several weeks as all the parts come into
thermal equalibrium and mechanical stresses dissipate. Of course,
the same is true for the enclosure, pier, vault, etc., even the 
lay of the cable to the seis is of concern.

Once an installed instrument has established itself as reliably quiet,
the sudden incidence of transient noise can be attributed to  either
the electronics of the feedback, the mechanics of the sensor, the 
immediate environment (covers, cable, vault), or the regional environment.
You mention seeing similar transients on a station in Norway. This would
be pure coincidence. As you know, the only transient-like or one-sided 
impulsive phases would be direct compressive arrivals of teleseismic quakes:
most of the energy from distant quakes that a broadband would record are
long sequences of super-imposed sine waves that would follow the direct
waves on most of the broadband sensors worldwide.

Assuming that the electronics worked OK when it was installed, the main
source of transients is dampness in the enclosure, which particularly
affects the high impedances of the integrator. Intermittent conduction of
contamination, like fingerprints, can be a problem. The STS-1 integrator
components are actually mounted on ceramic standoffs to deal with this.

Although seismometers are generally installed in dark, damp, moldy 
places, even in dripping caverns, no amount of dampness within the system
can be tolerated; the enclosure should be sealed, using vapor tight 
connectors (or seal them with silica gel around the inside terminals).
The enclosure(s) should be desiccated with  several units of diatomacious 
earth (cheaper than silica gel) and an indicator strip should be installed 
to indicate that the desiccant is active. (I shocked some people
by drilling 1.5" viewports in the covers of the Streckheisn electronics
enclosure so I could install the indicator inside them). The desiccant
can absorb about 1cc of water per unit; with 5 units in the 8"x10"x3"
box, it stays active for about 5 years in the dampness of a cavern.
In my opinion, there is little point using desiccant without the 
humidity indicator strips. (from Humidial Corp.)

The connectors should also be (or be made) weather tight. Some have
O ring seals, which are good only until the connector is dropped on
the dirt floor. If I have any doubt, I use stretched self-vulcanizing 
tape over-wrapped with vinyl electrical tape that can be cut away if/when
the connection has to be removed.

The seismometer enclosure itself should also be sealed and desiccated
if it is in a damp environment. The desiccant needs to be secured so
that as it expands it doesn't touch any moving part. If the seis
enclosure is tight enough to stay dry, it will also be spider and
waterbug proof.  Bugs can make interesting transients. I use wide, 
thin foam weatherstrip to seal the cover; it does breathe slightly 
along its length, which is preferable to absolute seals that might 
equalize barometric pressure through a pinhole with a little jet that 
blows the mass around.

Another problem has occurred with commercial seismometers that use
large capacitive transducers  with close (0.2mm) clearances. ANY
contamination, like an eyelash, will cause contact. A recurring 
problem has been peeling adhesives or coatings. These are often 
detected by strong local vibration, like a good stomp, that causes
a transient. For the speaker coil/magnet transducer, small magnetic
slivers on the poles can cause this. I use an inside-out strip of
masking tape to repeatedly clean the magnet gap as well as to ensure
that there are no dog hairs on the coil.

For the home-made seis, another problem can arise from the flexures,
especially if they are epoxied in place. They may be coming unglued
from improper cleaning at assembly, or some residual excess glue
might be in the flexing area. I haven't had this problem with the
later versions of the large vertical. Of course, any spontaneous
mechanical change of the moving system results in a transient; often
these are bidirectional micropositioning due to temperature changes.

You mention that you suspect changes in the slab under the seis, possibly
due to temperature or drying out. This is always possible, but unless
the cause is really obvious, like desiccation cracks, I would expect
slow changes rather than transients. If there are large enough dimensional
changes, the feet of the seis base may be moving on the slab; we always
use glass plates under the feet to allow for smooth differential expansion;
these are often epoxied to the pier or slab surface. Highly glazed ceramic
tiles will also do; I used them for the top surface of the pier here.

Other sources of strange transients from the environment can come from
slight movements of the covers, cables, etc. The cable should always
approach the seis laying on the pier, making S-turns or possibly a circle
around the seis.


Public Seismic Network Mailing List (PSN-L)

[ Top ] [ Back ] [ Home Page ]

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