[evlatests] Antenna 8 polarization

[Rick Perley]

    The 6cm receiver on antenna 8 was correctly re-oriented yesterday 
(having been found to have been mounted 90 degrees away from the 
standard position), and a quick observation of an unpolarized source 
made at noon today. 
    The cross-polarization between antenna 8 and all the other wide-band 
antennas is now around 4%, rather than the 15% that was seen on the 
weekend data. 
   
    But we can do a little more with these data.  Presuming that all the 
observed cross-polarization is due to the receiver (and that none of it 
is due to the antenna), rotating the feed by 90 degrees flips the sign 
of the cross-polarized contribution from that antenna.  So, if on a 
single baseline observed an unpolarized source, the two antennas both 
have their feeds oriented normally, the RL correlator data, after 
calibration for parallel-hand gains, gives an output given by:

RL = (Dr1 + Dl2*)I

where I is the total intensity. 
If antenna 2's receiver is rotated by 90 degrees, the relation becomes

RL = (Dr1 - Dl2*)I

The sum and differences give us the actual D terms amplitude and phase. 

I tried this, by hand, on a single baseline, 28 x 8.  I get:

Dl28 = 5.8% at pa = 113.
Dr8 = 8.6% at pa = -52. 

Note that these are fairly high, and that their phases are nearly in 
opposition (meaning, physically, that the R and L antenna ellipticities 
are nearly orthogonal).  This explains why, when in the correct 
orientation the polarization is seen to be low (the vector sum nearly 
cancels the contributions), while when one feed is rotated by 90 degrees 
(which rotates the D-term by 180 degrees), the observed polarization is 
very high (the vector amplitudes very nearly add tip to tail). 

This procedure can be done, using the data in hand, to derive the 
absolute cross polarizations for every antenna in the array.  If we want 
to do this for every band, we'll need one antenna's feeds to be 
similarly rotated, for a little while ...