[evlatests] R-L delays -- HUGE!

[Ken Sowinski]

On Fri, 31 Jul 2009, Rick Perley wrote:

>    But the biggest surprise comes from the delays in the R-L
> correlators -- the time delay between the two polarizations.  After
> removing the 'parallel-hand' delays, I find the RL and LR correlators to
> have a 305 nsec delay error -- of opposite sign in RL to LR (as it
> should be), and the same for all baselines (as it should be).

In a hallway conversation Rick said it is more like 130 ns.
Still big.  It arises in the optical demux used for ea24,
which by convention is the reference antenna for delays,
where it imposes a delay difference of about 120 ns between
IFs A and C.  There is a similar, but smaller difference
between B and D.  The optical demuxes used for Widar have
a different IF-based delay signature which must be accounted
for, but is not, in the control software.