September 19, 2014 Spike tests with OGP Calibration
The Offset Gain and Phase (OGP) calibration of the four ADCs interleaved to create the VEGAS output has been suspected as impacting the spikes seen in VEGAS. To see the effects of this calibration, we tested the spectra using Bank A of VEGAS without OGP calibration, and then repeated the test with the OGP calibrated. We inserted the IF noise source into the center of the passband to see the effects of signal strength on the spikes (the spikes disappear with signal). It was found that the OGP changes the spike but does not remove them. The tests with calibrated OGPs appeared to decrease some spikes, and increase others.
The data can be found in TGBT14A_912_33, in particular scans 6 and 7. The astrid part of the script can be found in TGBT14A_912/ogp_tests, though it should be noted that while the system was activating, the signal for polarization 0 was switched to the noise source in the IF rack and the attenuation adjusted to 21dB such that the output voltage at the IF rack was near 1V . VEGAS was power cycled to clear the OGP values stored in shared memory, and the Bank A manager was restarted to properly connect to the system. We then ran a 10 minute scan as described without OGP calibration (scan 6). After the scan, the system was allowed to configure the OGPs and a second 10 minute scan was run (scan 7).
The spikes are faint but apparent in the data from scan 6
, especially on the high frequency side of the image. The high-pass digital filter (see CICADAGreenBankSpectrometer2014Aug27_2
for more information) makes the spikes readily apparent
. After OGP calibration, spikes still appear
, but they now are more apparent in the lower frequencies of the band, which is also supported by the filtered data
When plotting the filtered data on top of each other, the changes from before and after calibration
become much more interesting to discuss. Note that the white data are from scan 6 (no OGP) and the red from scan 7 (OGP). It appears that the spikes are all still there, but with different magnitudes. The calibrated OGP data clearly helps with the very large spikes at higher frequencies, but increases the spike amplitude in the lower frequency part of the passband. It should also be noted that the calibration did not remove the high frequency spikes, just lowered them significantly (the white was plotted on top of the red, covering those spikes).
In all, it appears the OGP calibration does affect the spikes, but it does not remove them; unless, of course, the calibration used is wrong and a better set of calibration values could remove them.