[evlatests] A Most Bizarre Event

[Rick Perley]

    The short L-band observation taken last night (around 7PM) was 
affected by an event of unknown origin which greatly modified the 
switched power table for one scan. 

    There are four scans, two of the calibrator and two of a nearby 
(offset 1 degree or so) blank field. 

    The first three scans are completely normal.  All switched power and 
visibility values appear normal. 
    The last scan (duration ~ 2 minutes) was disrupted by some sort of 
event which greatly modified all switched power values, and the 
visibilities. 

    The major characteristics are:

    1) All antenna were affected equally. 
    2) All subbands were affected, but not in the same way. 
    3) Effects were seen in both PDif and PSum, but they are not the same. 
    4) The effect caused the monitored power to change *smoothly* during 
the 2 minute scan.  The preceding scan (a few seconds earlier and only 1 
degree away) is fine. 
    5) The effect is *much* stronger in RCP.  For some subbands, there 
is no effect at all in LCP. 

    Most remarkable is how the PDif and PSum were affected differently, 
for different subbands.   In all cases, the PDif dropped, typically by a 
factor of four.  For one subband (4), PDif is actually negative!  But 
PSum dropped for 12 subbands, and rose for the other four.  The four 
subbands with an apparent power increase are:  4, 9, 10, and 15.  All 
antennas apparently behave in the same way. 
    It's hard to imagine how the switched power can drop by a factor of 
4 while the summed power can rise by about the same factor. 

    I've looked at the cross-power spectra, and there is certainly 
strong evidence of an unusual event, seen in some subbands, occuring 
during this time.  Some spectra are simply blank (presumably, all 
integer zero), and others are clearly messed up.  But other subbands 
look fine during this time -- but evidently are not, as judged by the 
system power monitors. 

    The IAT time for this event was from 1:38 through 1:42 (or, 6:38 to 
6:42, PM, local time).  It would be useful for someone to look at other 
monitors to see if something very odd was occurring then.