[evla-sw-discuss] [Fwd: Re: EVLA WIDAR: Delay Model - format and content]

[Barry Clark]

John:
There may be compatibility problems with using an earth-center reference
for the VLA.  I put the reference point on the ground, at the point where
the current VLA phases up, because the devices in the old VLA antennas
have very restrictive limitations on what they can handle in third 
derivatives.  (Also because doing it for EVLA antennas requires a lot
more calculation, and I felt we had enough problems without having to 
worry quite yet about CPU loading.)  We may want to change this when
we start doing real-time on baselines over a few hundred km.  And we
would certainly do so for processing played back VLBI data in a subarray.
But, as I said, I don't believe there has been any serious design work 
on doing either of these.

Brent:
Completely unimportant, but as I understand it, in the current devices 
there is a 1:8 multiplexer which comes up in a random phase, so the 
time from central to when the first sample of a fiducial frame was taken 
is indeterminate by a few ns.  My belief is that errors caused by this
are far below other uncertainties in the system.  It is just possible
that some systematics in our curren delay accounting (based on what we
thought the delay through the waveguide might have been) might need to 
be changed.  But that affects only the system in which we solve for 
baselines, rotating it by a microsecond or two of time.

John:
Should the phasing boards output data as if sampled at the earth center,
or at the middle of the array?  (I can see arguments both ways, but none
very strong.)