[evlatests] Bizarre behavior (X-Band)

[Rick Perley]

    Ten hours of data were taken on the night of June 12 of the powerful 
source 3C405, along with a nearby phase/amp calibrator (2.5 degrees 
away), and the flux/polarization calibrator 3C286.  All data were taken 
in OSRO mode 1 (2 IFpairs, full polarization, 128 MHz subband width, 64 
channels per spectrum, giving 2 MHz spectral resolution).  Observations 
were made at four bands, L, S, C, and X, cycling round and round with a 
3.5 minute duration for each observation.   'Set and Remember' was 
applied to all IFs. 

    Many unusual, and undesirable phenomena were noted.  This report 
deals only with X-band.  The data were loaded without any flagging so I 
could get a better sense of system performance. 
    About 5% of the loaded visibilities were flagged by AIPS as integer 
zeros.  This fraction was the same for all bands.  Nearly all the 
flagged data were within a few seconds of a band/source change.  
Occasionally a few antennas gave integer-zero data for up to minute.   
No more than 1% of the zero data were in the middle of a scan. 

    Antenna 25, IF 'D' gave no fringes throughout. 
    Antennas 1, 8, 9, 12, 13, and 14 all 'failed to tune' (which means 
did no fringe, giving only noise, for a single scan) on occasion.  This 
failure is not common:  of the 53 scans, and 26 antennas, there was a 
total of 12 'non-fringing' events -- only 0.4% of the total.  Antenna 13 
failed four times -- each time on both IF pairs.  Antennas 12 and 14 
failed once on both IF pairs.  Antennas 1, 8, and 14 each failed once on 
a single IF pair.  Antenna 9 failed 3 times in IFpair AC.  (All such 
failures involve both polarizations, suggesting an LO failure). 

    Amplitude stability for some antennas was very poor.   The best that 
can be said is that a few antennas look pretty good.  Most of the 
antenna show various forms of instability which cannot be easily 
summarized.  A few antennas showed clear stepped gains -- three of them 
(5, 17, and 12) are exactly 1 dB in power, suggesting that 'set and 
remember' does not work in the manner intended. 

    Phase stability is non-existent on some antennas, showing a partial 
to total failure to return to phase following observations at other 
bands.  Particularly notable are:  ea05 (identical changes on all four 
IFs, all of them matched by amplitude changes as well!), 8, (on IF AC 
only), and ea11 (on IF BD only).  All of these show multiple phase 
states.   Antennas 10 and 14 also show multiple phase states, but on 
both IF pairs, and differently between the IF pairs. 

    Bandpass stability is good on a few antennas, and quite poor on some 
others.  Two of the antennas showing 1 dB amplitude changes show notable 
bandpass ripple when the 1dB amplitude change occurs -- the ripple has a 
wavelength of 250 MHz, indicating a mismatch pair separated by about 50 
cm.