Data from May 16, 2006

Noise performance of the first four channels of UXO DAQ system, measured April 22, 2005. Inputs are shorted, attenuator is disabled (also called "high gain mode"). Power (+12V at 206 ma, -12V at 720 ma) comes from a HP E3631A bench power supply.

gzipped raw data file, 262144 data points. Time between points is 3.97μs, full scale is +/-32768. The AD7676 CNVST pins are pulsed 1080 ns after the last SCLK pulse.

Noise measured from a pair of coils at the Richmond Field Station on April 28, 2005. Red trace is with them adding (1.23 mV rms), green is subtracting (0.306 mV rms), blue is a shorted amplifier channel (3.25 μV rms).

Two gzipped raw data files, added and subtracted configurations, 262144 data points each. Time between points is 3.97μs, one bit is 1.907μV. In each file, column 1 is the coil pair, column 2 is a shorted amplifier channel. I can provide the original data sets on request, they contain over 5 million points each. Here's another view of the same data, same color assignment, but this time on linear scales:

2005-05-16: Data from a first test of synchronous transmit/receive operation. This plot comes from one second of acquisition, averaging 7/9 of the 540 Hz (total for all three coils) transmit pulses. Receivers are in pure high-gain mode, one bit is 1.907μV at the coil.

The excitation pattern is 9-9-9, exactly 463 samples per excitation, 4.00 μs per sample. A complete repeat pattern of 54 pulses takes a nominal 100.008 ms, subject to unmeasured crystal oscillator offset (unmeasured, probably not more than 50 ppm). I made no attempt to frequency lock this to the AC line. We tried to make the excitation sequence in the X-Y-Z order, for whatever definition of X, Y, and Z you used before. After taking try1 and try2, we flipped the sign of the transmit triggers for X and Y.

Raw data: after unzipping, each has 250020 lines of four columns each. The time axis is implied, 4 μs per line. The four columns are the raw output of the four ADC channels:

Column 1 comes from coil set 5
Column 2 comes from coil set 6
Column 3 comes from coil set 7
Column 4 comes from the transmitter current probe

File try1.dat.gz: columns 1&2 set to high gain, 3&4 set to low gain
File try2.dat.gz: all channels in low gain mode
File try11.dat.gz: all channels in high gain mode, transmitter disabled

The plot above comes from try1. The transmit power supply read 0.82 amps 13V average. Jim estimates we were running 8 Amps peak on the transmitters. We should verify that by looking at column 4.

2005-06-03: After some board rework, mostly revolving around the voltage reference handling, I measure much improved noise performance. For the eight channels on boards 1 and 3 (used for receive coils in the next run), in low gain mode (equivalent to 270 Ohm termination in high gain mode):

Total rms noise on each channel is about 1.3 μV, and the contribution between 10 Hz and 10 kHz is very close to the OPA637's advertised 4.5 nV/sqrt(Hz).

Board 2 isn't in quite as good shape; channels 1 and 2 are similar to the above, but channels 3 and 4 are noisier. Jean-Francois and I will try to figure out why -- later.

2005-06-03: First full 16-bit data set from some UXOs: a 40 MByte .tar.gz file.

Channel and time calibration as it was on 2005-05-16. Each data file has nine columns, the first is excitation current, followed by receive coils 1 through 8. File names are

t1_low.dat through t6_low.dat, all channels in low gain mode, UXOs 1 through 6
t1_mix.dat through t6_mix.dat, all receive channels in high gain mode, UXOs 1 through 6
z1_low.dat, z2_low.dat, all channels in low gain mode, no UXO present
z1_mix.dat, z2_mix.dat, all receive channels in high gain mode, no UXO present

All data sets are fully synchronous and time aligned with the 9-9-9 excitation pattern, 540 pulses of 463 samples each. What I don't know is the order and sign of the transmitter excitations (I plan to cure this defect in future data sets with the next revision of firmware and software). Between Ocean and Torquil, I bet we can reconstruct the information here, looking both at the excitation current and receiver responses.

I don't list here which UXOs correspond to data file index 1 through 6. I randomized the order from the original collection sequence. They include two objects, two depths, and two orientations.

Sample data file with embedded GPS info. Oops, I got a GGA message where a GST message belongs.

Sample data from the bldg. 71 parking lot.

June 4, 2005
Larry Doolittle