|
Figure 1 shows the
filtered output of a 6000 meter depth sensor used to detect
earthquake-generated tidal waves (Tsunamis).1 The real signals
are resolved to 1 mm of water (1 part in 6 million) and clearly
show the signature of the Tsunami which is only several
centimeters magnitude at the deployed depth of thousands of
meters.
Data from an evaluation by Hutt, Holcomb, and Agnew2 of
high quality sensors for use in atmospheric seismic studies are
shown in Figure 2. Noise of the Paroscientific transducer is
much lower than all other sensors.

Figure 2. Microbarograph Comparison
(Click here
to enlarge)
Noise levels as a function of
frequency are generally expressed as spectral densities. Plots
of this type are used to determine whether a sensor can measure
a desired signal. The goal is to have the sensor noise levels
much smaller than the expected real signals at all frequencies
of interest. In the tests by Hutt and Agnew, our transducers had
power spectral density noise levels a factor of 100 lower (20
db) than the next best transducer.

Figure 3. Noise vs. Record Length
The ultimate resolution
achievable with a transducer is limited by its noise level.
Typically, the rms noise increases for longer data records
because of sensor drift and because temperature and other
environmental contributors to noise tend to vary more over a
longer period of time. Typical rms noise levels for our
transducers are shown in Figure 3. For records shorter than
about 1 hour, the rms noise level is less than 1 part per
million. The rms noise rises slowly with record length, reaching
approximately 10 ppm for records several years long.
Applications where absolute
accuracy and stability are paramount include pressure
calibration systems, transfer standards, meteorological
measurements of barometric pressure for weather forecasting, sea
floor depth survey equipment, and long-term deployments in
remote locations. Please click here
to see our long-term barometer stability test. |