[evlatests] widar test data, jun 08

[Vivek Dhawan]

About 1hr on 0217+73 on Jun 08, single subband of 128MHz
(The intention was to write out only 1 subband; 4 subbands
were written but bands 0,3,4 were 0.)

I processed this and a dataset from Jun 3rd, same source,
corresponding subband. The sequence was SETJY=4.0Jy; FRING;
BPASS; CALIB; SPLIT; IMAGR (plus plots vs. t and f).

1. The delays have changed by ~60ns for 3 stations, betweeen
3rd and 8th. But no significant differences were found in
the bandpass shapes other than the delay.

There are still 2 groups of antennas as before 5 in each set,
180deg jumps between sets, none within a set.

2. The dynamic range in the image, averaged across 120MHz was
  ~4000:1 on each day, the later one about 10% worse but this
could be differences in flagging out edge data. This is at
least a factor 20 above the noise (if I got the sum right).
I don't see any abvious signs of structure in the images,
the residuals floor looks a little gridlike perhaps, rather
than pure noise.

3. So, I can confirm that the raw data show nothing wrong, but the
images have higher rms, just as Rick saw. Writing 1 subband and
not 4 out of the CBA makes no difference.

4. The weights vs. time look different on the 2 days, so I tried
imaging with weights off - not much difference. (Incoming wts
were 1.0 for all data before calibration)


What next?