[evlatests] Delay errors in MODCOMP-free data

[Vivek Dhawan]

May 27th I reported that modcomp-free data showed several
10's of ns delay errors, changing over 2 hours as the source
transited - I thought it looked like a time error.

To confirm this, I did a test (in line mode) tracking a pair of 
calibrators to the east and west alternately (0956+252 & 1925+211).
With modcomps in control, the fringes are slow and delays are
near zero. The same file repeated an hour later without modcomps
shows delay errors upto 100ns, large differences between E and W
arms and between the 2 sources. The errors roughly increase with
offset from array center. The fringe rates are also higher
with modcomp-free data, a turn every few minutes, but not
terrible.

The plot shows delay vs time for 3 antennas. From left to R,
the first 2 scans are under modcomp rule, the next 2 without.
The sources are 0956, 1925, 0956, 1925. (The modcomp-free data
are lower SNR so the scatter is higher).

Vivek.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: J.PS
Type: application/postscript
Size: 15702 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20070601/eddb4b38/attachment.ai>