Tests of LO Blanking and Dopper tracking, Aug-Sep 2002

{F.Ghigo: Sept 13, 2002}
The LO system tracked an LSRK velocity, and the spectrometer used the LO blanking signal. There was only switching of the noise cal, no sig/ref switching and no frequency switching.

The object was to check whether the LO blanking of the spectrometer was working correctly and that the LO frequency and exposure times were correct.

Test Run, Aug 21, C-band
The bottom line, for the impatient:

Test Run, Sep 01, L-band

In the Sept 01 run, both a test tone and an astronomical signal were observed in the same band. Scans were done both with and without LO blanking enabled.
The bottom line:

Test Run, August 21, at C-band.

On August 21, we tracked a H110-alpha line (rest freq = 4874.157 MHz) source and did off-on scans, with the addition of a test tone at 4875 MHz. The spectrometer was used in mode 1N1-0A-12-3 (one quadrant, one sampler, 12.5 MHz bandwidth, 3-levels, 65536 channels, channel spacing 191 Hz). The scan that was examined was scan number 53, project 'Tfghigo_aug21', which was an off-source scan with the test tone at 4875 MHz injected at a power level of -70 dbm. The scan consisted of 500 one-second integrations. The LO frequency tolerance was 5 Hz, and the frequency updated about every 10 seconds.

For each integration, we determined the mean frequency of the test tone peak. The test tone was apparent in 4 successive channels. The weighted mean frequency was computed over a range of 15 channels centered on the peak, weighted by the amplitude.

Test Tone

This shows the test tone in integration number 1:

Frequency vs Time

In the next plot, the mean frequency of the test tone is plotted versus time in seconds from the beginning of the scan. We can see the 5 Hz changes in frequency happening about every 10 seconds, just as expected. There is no sign of strange excursions in the frequency as might happen if the blanking were not working.

Exposure time

The exposure time reported by the spectrometer for each integration is shown in the next plot. We notice a drop of about 0.02 seconds in the exposure for many of the integrations that correspond to times when the LO frequency is being changed. 0.02 seconds is about the expected blanking time. The problem is that not all LO frequency changes have a corresponding drop in exposure time. For example, between 0 and 100 seconds, we expect about 10 integrations to have been blanked, but only 6 of them show a drop in exposure time. So apparently either the blanking is not happening in many cases, or else the exposure time is not showing it.

Another anomaly is that the exposure times are all near 0.24, where we would have expected them to be close to 0.5 seconds. The specified integration time was 1 second, and there were two phases: cal-on and cal-off, so we would expect about half a second exposure time per phase. Yet the spectrometer is indicating about half the expected exposure.

Test Tone Power (Cal-on phase)

An even weirder effect shows up when we look at the integrated power of each test tone. As you can see from the next plot, the power changes by 20% or more, either up or down, at times when the LO frequency is changing. This effect is seen in the Cal-on phase, but not in the Cal-off phase. Maybe there is some interference between the LO blanking and Cal blanking signals.

Test Tone Power (Cal-off phase)

The Cal-Off phase shows no correlation between test tone power and LO frequency changes.

Test tones showing the change in power at certain times during the scan.

The first two plots show the test tone in integrations number 8 and 9. The LO frequency change happens during integration number 9 and we see that the Cal-On phase (Green line) drops for integration 9.

The next two plots show integrations number 18 and 19, where the test tone power in the cal-on phase increases significantly in integration number 19.

Test Run, September 01, at L-band.

On Sept 1, we tracked the 1665 MHz OH line, observing the source W3OH, and also injecting a test tone at 1662 MHz. The spectrometer was used in mode 2N2-4A-12-3 (two quadrants, two samplers, 12.5 MHz band, 3 levels, 65536 channels). It turned out that sampler 0 was accidentally disconnected internally, so there was data only from sampler 4, the RCP polarization.

The project was 'TLO1_fg_Sep01'. Scans of 10-minute length were run with an integration time of 3 seconds, yielding 200 integrations in a scan. The LO tracking tolerance was 1 Hz, and we noted that the LO updated about every 20 seconds during each scan.

Two scans were examined:

Three lines in the spectrum were examined: a) the test tone, b) the strongest OH1665 line component, and c) the weaker but narrower OH component at 1665.67 MHz. One would expect that during the scan, the frequency of the test tone will gradually change, and that the frequency of an OH line component will stay the same. This turns out to be the case.

For each of the 3 lines, the mean frequency was calculated by taking the weighted mean of the frequency for the channels across the line, and weighting by the amplitude of each channel. Also calculated were the width and the integrated counts of each line. This was done for each of the 200 integrations in the scan.

Plot of the whole 12.5 MHz band, for integration number 100 out of 200. The peak on the left is the test tone. The highest peak is the 1665.4 MHz OH line from W3OH, and the peak on the right is the 1667.4 MHz line.

Close-up of the 1665.4 MHz OH line The small peak on the right ("1665.67MHz") was examined, as well as the main peak.

Plots of test tone frequency
The next two plots show the tone frequency as function of time, first for scan 30 (LO blanking enabled), next for scan 32 (LO blanking disabled). There seems to be no significant difference between these plots. They both have an rms of about 7 Hz about a linear trend. The frequency is changing smoothly througout the scan, as one would expect.


Exposure Time, scan 30.
The exposure time is plotted for scan 30. We see that the exposure time drops, corresponding to the blanking, but this does not happen for every change in the LO. The LO frequency changed about every 20 seconds, but there was not a corresponding drop in the exposure time for all LO frequency changes. No drops are seen near the beginning of the scan and near the end.

Exposure Time, scan 32.
In scan 32, the LO blanking was disabled. Here we see the exposure time constant throughout the scan.

Test Tone Integrated Counts
Scan30: The power in the test tone showed a significant jump whenever the LO frequency changed. This jump is seen only in phase 1 (probably the cal-on phase). Note that in the corresponding plot for Scan 32, this effect is missing. The LO frequency tracking is happening in both scans, but we see the jumps in power only in the scan where the LO blanking is enabled. Thus the blanking signal must be causing the power jumps.

OH 1665 main line
The mean frequency of the OH1665 line is shown in the next two plots. The frequency is constant, with an rms of about 11 Hz. There is no apparent difference between the cases with and without LO blanking.

OH 1665 power
The integrated counts show the same effect for the OH line as for the test tone. Phase 1 has spikes that apparently happen in the integrations when blanking is happening. This is seen in Scan 30 where LO blanking is enabled, and not in scan 32, where blanking is disabled.

OH 1665.67 side line
This line is narrower than the main OH1665 line. These two plots show the mean frequency of the line for each integration of the scan. Scan 30 (first plot) was done with LO blanking enabled, and Scan 32 with blanking disabled. The scatter in both plots is about 7 Hz, and to this accuracy the frequency of the line is stable throughout the scan.

These two plots show integrated counts of the the 1665.66 line. Again, one phase shows spikes in the scan when LO blanking is enabled, and no spikes when disabled.