[evlatests] Some thoughts on Right/Left phase differences

[Barry Clark]

This measurement will tell us primarily about module-to-module
variation in several not very suspicious modules - front end
amplifiers, T305s, D30Xs, D351s, the LO doublers and tripplers
in the front ends, and various splitters and wires that deliver
what is supposed to be the same LO to two different modules.

Vivek's clever technique of looking at a stong source in a narrow
band continuum mode is certainly the best way of looking for
short timescale variations.  He used a maser at Q band; at other
bands there are probably sufficiently strong sources that you
could use real continuum sources at 6 MHz bandwidth (the delay
switching phase jump at that bandwidth is 0.7 degree).  More
conventional observations are fine if you average to a few tens
of seconds before you start looking at things.

These measurements provide a check on the consistency of the
modules above, but may have important consequences themselves
if they are large enough.  They correspond to a swing of the
polarization vector, and this may prove a limiting factor in high
dynamic range mapping.

The differences Vivek sees at Q band are surprisingly large
(perhaps 5d peak to peak) and at a surprising one minute timescale,
but are not as alarming as they might be, since they only amount
to about 0.3 ps.

The differences he sees at K band are of comparable magnitude, in
degrees, and therefore at a rather more alarming level in picoseconds.
The timescale in his example is longer than at Q band, but still shorter
than one might expect ab initio.

We should look at all bands (and if the few degree wandering around
continues to 20cm or even 6cm it is pretty disastrous on its own),
looking at several timescales (seconds to hours) and a range of 
elevations.  We need a good characterization, probably just RMS over
all baselines with various averaging times, to see where we are in
systematics.  Vivek's examples of bad behavior, while instructive,
leave me wondering whether we have a systematic problem or a few bad
behavers.