You are on page 1of 8

Initial Testing of the TitanSMA Accelerograph at the Albuquerque Seismological


May 2012

A. T. Ringler1 and C. R. Hutt1

U.S. Geological Survey, Albuquerque Seismological Laboratory

The authors did some initial tests of the Nanometrics Titan SMA acceleorgraph at the
Albuquerque Seismological Laboratory (ASL). This unit is a small accelerometer, what
appears to be similar to the Nanometrics Titan, packaged with a 24-bit digitizer (Figure
1). The unit is easily set up and has physical dimensions that are comparable to an
accelerometer that doesnt contain a digitizer. Various parameters on the unit (e.g.
sample rates and sensitivity) can be controlled through a computers web browser. The
user can calibrate the instrument with sine waves or steps through the web browser.
Finally, it is also possible to retrieve mini-seed data using the web browser. We were
provided with instrument sensitivities along with a pole/zero model that describes the
response of the instrument.

The TitanSMA comes with three connections cables (network, power, and GPS). It is
also possible to install digital audio files and have them replayed through the calibration
coil. However, we did not pursue these tests and restricted our attention to what we feel
are the most important common tests: instrument noise, static sensitivity, and integration
tests (Hutt et al, 2009).

Basic Usage
We initially set up this instrument in the ASL surface vault (Figure 1). We found the unit
relatively painless to setup and retrieve data from. We were also able change the
sensitivity of the unit without difficulty. The unit has four contact points to the physical
ground, which is different than most seismometers and accelerometers. However, we did
not observe any issues with the additional foot. It is also possible to remove two of the
feet and replace them with a central foot allowing for a standard three configuration.
This was the setup we used for the noise tests.

Figure 1: Initial setup of the Titan SMA at ASL.

Static Sensitivity
In order to compare the sensitivity of the instrument with the manufacturers reported
sensitivity we did static sensitivity tests (commonly known as a flip test) while the unit
was in its 2 g full-scale mode (4 counts per g). This involved placing the instrument in
a box and flipping it on its various axes and recording the acceleration of gravity. The
results of this test are shown in Figure 2. As can be seen by the legends in Figure 2 all
components had static sensitivities to well within 1% and likely limited by our ability to
orient the instrument. Our error estimates in the sensitivities are 99% confidence
estimates (2.5 assuming a Gaussian distribution).

Figure 2: Time series of static sensitivity tests for the TitanSMA. The sensitivities in the
legends denote our estimates along with the 99% confidence interval error.

Instrument Noise
In order to characterize the noise of the TitanSMA we performed noise tests on the
instrument in the ASL underground test vault. We did noise tests on the instrument in
both the 4g (Figure 3) and 1/8g (Figure 4) full-scale configurations. These two
configurations correspond to the least sensitive and most sensitive settings, respectively.
We did not perform any coherence analysis to estimate the noise. The instrument was
able to record microseisms in the 1/8g full-scale mode. This suggests future noise tests
would benefit from a more careful procedure using multiple sensors. For each noise test
we used a visually quiet 1 hour segment of data. In either case, both configurations
showed that the accelerograph had very low noise levels for a strong motion unit. To
fully characterize the self-noise of this unit in the 1/8g full-scale mode more careful
testing would need to be performed, closer to a typical broadband self-noise test.
Figure 3: Noise levels of the TitanSMA accelegrograph while in 4g full-scale mode.
Figure 4: Noise levels of the TitanSMA accelegrograph while in 1/8g full-scale mode.
Separation between spectra at periods greater than 10 seconds period could be a result of
test conditions. The fact that these instruments are able to record microseisms suggests
coherence analysis techniques will be required to fully characterize the noise.

Integration Tests
In order to test the instrument under varying levels of ground motion we looked tested the
TitanSMA (configured for 2g full-scale) on the ASL horizontal Anorad shake table. Our
test consisted of shaking the instrument with a linearly increasing amplitude, up to near
full-scale (approx. 1.9 g), on the shake-table. We then double integrated the TitanSMA
output and compared it to the displacement feedback output of the Anorad table (Figures
5 and 6).

In order to correctly compare the tables output with the TitanSMA we detrended both sets
of data, removed the mean, applied a wide-band band-pass filter, deconvolved the
TitanSMA output with its instrument response (frequency domain water-level method),
and also applied two cumulative trapezoidal integrations to get to displacement. Both the
North-South and East-West components showed similar results from the integration tests.
We found the sensitivity to be well within 1% on both components (relative to the tables
output). We found the root-mean square residual between the tables displacement output
and the double integrated accelerograph output to be around -44 dB, which is at the
advertised limit of the table and likely further limited by our ability to align the
instrument. Finally, we found the phase delay between the table and TitanSMA to be
approx. 0.002 seconds which is half a sample interval, again suggesting our test setup
was the limiter.

Figure 5: Double integration test of the North horizontal component of the TitanSMA.
We show both the shake tables output (blue), the integrated output of the TitanSMA
(black), as well as the residual between the table and the accelerograph (red). The
envelope of the time series is a result of the tapers used in deconvolution.
Figure 6: Double integration test of the North horizontal component of the TitanSMA.
We show both the shake tables output (blue), the integrated output of the TitanSMA
(black), as well as the residual between the table and the accelerograph (red). The
envelope of the time series is a result of the tapers used in deconvolution.

Our limited set of tests of the TitanSMA suggests that this instrument is very easy to use
and is able to produce strong motion acceleration data with little effort on the part of the
operator. The units ability to produce mini-seed data allowed us to do a number of tests
that suggest the TitanSMA is also capable of produce very high quality data within the
advertised linear range of the instrument. We were able to reproduce the sensitivities
advertised by Nanometrics to within 1% using both a static sensitivity tests as well as the
derived sensitivity from our more broadband integration tests.
We did not attempt to verify the response using the instruments ability to playback
digital recordings through the calibration coil. This would allow the user to verify the
response of the instrument using a source produced by the user of the instrument. This
would be valuable for future testing.

In summary we find this unit easy to use and capable to producing very high quality
strong motion acceleration data. The relatively compact size of the unit makes it a very
nice addition to the accelerograph market. Our limited number of tests suggest that the
unit operators within the manufacturers specifications. Of course, more rigorous testing
will be required to fully verify the additional specifications.

We would like to thank Nanometrics for providing us with a TitanSMA for testing. We
would also like to thank Mark Robertson for his help with GPS cables.

Hutt, C. R., J. R. Evans, F. Followill, R. L. Nigbor, and E. Wielandt (2009). Guidelines
for standardized testing of broadband seismometers and accelerometers, U. S.
Geological Survey Open-File Report 09-1295, 62 p.