• NSF
  • Iris
  • Iris

Random binary tests


This note is intended to provide users with supplemental information that may be used to analyze random binary calibration tests routinely performed at IRIS/IDA GSN stations.   The procedures used to characterize the full system response of sensors deployed at Global Seismographic Network stations are described in Davis et al. (2005).   Responses are encoded into dataless SEED volumes and made available to users through the IRIS Data Management System (DMS).  Independent tests of the accuracy of these published responses are reported in the above paper and in Davis and Berger (2007). 

One important step in the system calibration process is to measure the relative response of the principal seismometers and any analog anti-aliasing filters shaping their output as a function of frequency. The procedure and theory of calibration is explained in detail in Berger et al. (1979).  A random binary signal is fed into the calibration coils of the seismometer, and the output, recorded for later analysis. In newer systems (those deployed post 2000), the input signal is digitized on a separate channel and retained for later analysis.  The shape of the sensor frequency response is found by fitting a perturbed function of the nominal system to the cross-spectrum of the output and the (known) input (Fels and Berger, 1994). The outcome is a representation of the sensor’s portion of the total system response in standard pole-zero format.

The input random binary signal is a sequence of step functions of identical amplitude but alternating polarity and characterized by a clock period T (see Berger et al. 1979, pg 272 for complete references).  At each interval T, the signal has a 50-50 chance of changing state.  Examples of several input series for clock periods used to calibrate older IDA instruments are provided with this document.  These series already have the FIR filter response convolved into the random binary series.  For users who wish to generate the original series without the FIR response, the random number seed file and a C program to read it are included as well.  (See File Notes below.)  The clock periods of the signals employed are listed in Table 1 and provided as a csv file.

Recordings of these random binary tests were originally withheld from distribution to avoid confusing users.  These time series have now been encapsulated into SEED and will be made available as a special data product by the DMS.  The random binary input to the coil varies with the version of data logger and with the run-time configuration of various parameters set when the operator initiates the test.  Although the field engineers note this information in trip reports, it is not tabulated here, and no effort is made to track the amplitude of the various input signals.  Below are plots of example time series for each data logger type and details of how they are generated.



At the core of the Next Generation System is the Quanterra Q330HR data logger.  For calibration tests on IRIS/IDA systems, the Q330HR DAS generates the random binary signal and also provides a record of the input to the sensor calibration coils.  Because this input has been processed by the DAS with the same digital filters as the output data stream, comparison of the input and output permits a direct estimate of the pure sensor transfer function. 

Due to a design limitation, the Q330HR will not generate a random binary series longer than 4.5 hours (16383 sec) nor with a frequency lower than 0.49 Hz.  The series shown in Figure 1 and Figure 2 was generated with the following parameters: Signal attnenuation -12 dB (+/- 2.5 V), frequency .4902 Hz, duration = 4 hours.

The response of the Z, N and E components of the seismometer are shown in the top three traces, the input to the coils, on the bottom trace.  Note that the polarity of the Z signal is different from the N and E components.  This is always the case for the STS1 sensors.


SAN 2012 (IDA MK8) DAS

Like the Q330HR, the MK8 records the random binary input signal, in this case on the BHZ-10 and LHZ-10 channels, so a direct estimate of the sensor transfer function may be made.  Two examples of waveforms from random binary tests are shown here.   In the first example (Figure 3 and Figure 4), intended to capture the low frequency response of the sensor, a twelve-hour long signal with a clock period of 10 seconds is routed into the cal coils of the STS1s.  The response of the Z, N and E components of the seismometer are shown in the top three traces, the input to the coils, on the bottom trace.  In the second example (Figure 5 and Figure 6), a higher frequency random binary signal ( clock period 0.1 sec, duration 30 mins) is used to deduce the short period response.


REFTEK 9716 and 9718 (IDA MK7) DAS

Unlike the systems listed above, the IDA MK7A, MK7B and MK7-ISP systems do not record the input to the coil.  Waveforms from a random binary test recorded by a MK7 DAS are shown in Figure 7 and Figure 8.  These waveforms were recorded on a KS-54000 sensor.  In such cases, the polarity of the Z trace should be the same as for the N and E traces.

The clock period used for the MK7A and MK7B DASs is typically 10 seconds and for the MK7-ISP, 20 seconds.  An example of the input signal for each is contained in files input_mk7_cl10 and input_mk7_cl20, respectively.




The following files will be useful in dealing with data loggers preceding the MK8:

input_m7_cl10  - input signal from the MK7 DAS with a clock period of 10 seconds.  File is written in 4-byte floating  point numbers.

input_m7_cl20  - input signal from the MK7 DAS with a clock period of 20 seconds.  File is written in 4-byte floating  point numbers.

ranseeds  - random numbers used as seed for the input signal generator.  File is written in 2-byte integers.

randstp.c  - C-language program to read file ranseeds and generate a pseudo-random binary time series.




Berger, J., D.C. Agnew, R.L. Parker and W.E. Farrell, Seismic system calibration: 2. Cross-spectral calibration using random binary signals, BSSA, 69, 271-288, 1979.

Davis, P. and J. Berger, Calibration of the Global Seismographic Network using tides, SRL, 78, 454-459, 2007.

Davis, P., M. Ishii and G. Masters, An assessment of the accuracy of GSN sensor response information, SRL, 76, 678-683, 2005.

Fels, J.-F. and J. Berger, Parametric analysis and calibration of the STS-1 seismometer of the IRIS/IDA seismographic network, BSSA, 84, 1580-1592, 1994.