[evlatests] Perplexing L-band Problems

[Rick Perley]

    I have carefully reviewed the 1 hour of data taken last evening on 
our northern point-source. 

    As noted earlier, there were three phase jumps noted, all within the 
first minute of the run.  None of these appear to be of the 180 degree 
variety.  The final 59 minutes of the run had no phase jumps. 

    Numerous amplitude 'steps' were seen -- some of the 1, 2, or 3 dB 
variety.  Some were not. 

    Sub-bands 1 and 3 were placed in regions of extreme interference 
(centered at 1052 and 1564 MHz), and only a cursory inspection of these 
were made.  The characteristics noted below appear to apply to these two 
sub-bands as well. 

    Lots of troubles were seen on the two 'clean' subbands (centered at 
1180 and 1950 MHz). 

    1) Two baselines had unique and steady phase slopes associated with 
them, on all IFs, over all times:

       18 x 23:  1.5 turns (exactly) across the 128 MHz sub-band width. 
         1 x 24:  1.0 turn (exactly) across the 128 MHz sub-band width. 

          It is important to note that these devastating phase slopes 
*** cannot be corrected by FRING or BPASS*** !!!!!  They are unique to 
the baselines noted, and are easily found after standard delay and 
bandpass calibration has been executed. 

    2)  After removing the two baselines noted above, I split out a 
'pseudo-IF' by coherently summing across the channels (this to improve 
SNR and decrease computing time) for sub-bands 2 and 4.  The channel 
ranges were selected to avoid the visible RFI. 

    3) Sub-band 4 (1884 - 2012 MHz) gave us clean data -- no 
oscillations in amplitude and phase.  I made an image, which gave 
10,000:1 DR, but no background sources were visible.  This is quite 
impossible at L-band in C-config., but the explanation is probably at 
hand.  The database arrived with no baseline coordinates (all u, v, w = 
0), so UVFIX was run.   However, Eric tells me that UVFIX will not 
compute correct baselines, due to a change in the definition of the 
direction of the axes (combined with some other still unresolved 
issues).   If the resulting baselines are indeed semi-randomized, then 
no coherent image will be obtained (except at the field center). 
    
    4) Sub-band 2 (1116 -- 1244 MHz) is a total mess.  Wild visibility 
oscillations are seen on almost all the short spacings.  Particularly 
bad are the ten spacings involving antennas 2, 9, 18, 25, and 28 -- all 
on the north and east arms.  No west-arm antenna gave significant 
oscillations (nor did 3 x 9, adjacent antennas on the east arm).  The 
periods of these oscillations are near 6 seconds, and as always, there 
is no obvious connection with geometry.  The source HA was +8.5, 
elevation = 23 degrees, azimuth = 330.   The amplitudes of these 
sinusoids are HUGE:  phase oscillations of up to 100 degrees, and 
amplitude oscillations up to twice the flux density of the 2.5 Jy 
source.  None of the long spacings gave any visible oscillations.  (Long 
= 5.5 klambda, half the total extend of this configuration).  The 
spectrum is clean -- if these resonances are due to RFI, it HAS to be 
broadband.