[evlatests] Update on Strange R-L phase behavior

[Rick Perley]

Answers to Barry's suggestions:


On 3/30/22 11:54, Barry Clark via evlatests wrote:
> I think a couple of quick tests on the existing data are worthwhile, 
> despite a low probability of showing anything useful.
>
> 1.  Make sure the effect shows up on a single baseline, data 
> uncalibrated, from, say, baseline 1-9, just to positively exonerate 
> all the calibration software.  There's probably enough SNR to do this 
> on a single 2 MHz channel even.

It does.  I extracted a single channel (2MHZ) at C-band.  With no 
calibration whatever, for 3C286, we get the following phase differences 
(R - L) (in tme order):

Elevation    DeltaPhi

-----------------------------

     55            38

     60            38

     63            39

     67            38

     71            39

     75            39

     78            39

     82            39

     85            40

     87            41

     85            41

     81            40

     78            39

     74            39

     70            39

     67            38

     63            38

-------------------------------

     Note the magnitude of the effect is identical (3 degrees) to the 
plots I sent earlier today.

>
> 2.  Look at the self cross correlations to make sure that the effect, 
> for some mysterious reason, chose to vanish when Paul did his 
> definitive measurement.
>
     I don't think AIPS can access the 'self cross hand' phases. I'll 
have to consult with Eric.


> 3. Instead of calibrating each subband, transfer calibrations from one 
> subband to the others.  This may tell us if the effect is a phase 
> offset or a delay offset.
>
     Not sure about this one.  I've done the calibration in three 
different ways:  (1) Single delay solution over all SPWs, followed by 
calibration:  (2) individual delays, one for each SPW, followed by 
calibration,  and (3) No delay at all, with calibration using the 
central few channels.

All three show the same residuals.  I should think this is sufficient to 
show that the effect is not due to delay.  Agree?

Rick