[evlatests] Modcomp-Free Test
Rick Perley
rperley at nrao.edu
Mon May 7 17:59:32 EDT 2007
A test of the modcomp-free system was attempted over lunchtime.
In order to compare Modcomp-free with Modcomp-driven, I have
modified by test procedures to run a brief all-band, all-system check
with the modcomps in control, then repeat the test, exactly 50 minutes
later, with the modcomp-free system. We can then more or less compare
'apples with apples'.
Because the Modcomp-free system cannot change mode unless there are
at least two consecutive scans in that mode, the sequence of both tests
is the following:
A) Continuum, with 0.418 seconds averaging: L P C X K and Q
bands. All at 50 MHz BW, except for P-band at 3.13 MHz.
B) Line, mode 4, with 3.3 seconds averageing, same sequence as
above. 12.5 MHz BW, except for P-band at 3.13 MHz.
Reference pointing was done, at X-band, between the X band and K
band observations.
Results:
1) The Modcomp-driven data look fine. (Not perfect -- the usual
little problems are there!)
For some reason, the operator believed that to change to the
Modcomp-free system, the prior observation had to be aborted while in
the middle of a scan. So the last scheduled observation (Q-band, line)
didn't return any data.
The conversion to a Modcomp-free system took place. Then ...
2) No sensible data whatsoever was returned from the Modcomp-free
observations!
Ken Sowinski printed out the log, which shows quite plainly that the
scheduled observations took place, and on time. The operator reported
that everything worked well, so far as he could tell.
Yet the data which were filled are complete nonsense:
For the Continuum observations:
a) Data filled at L-band are at the wrong frequency, and wrong
time. No fringes.
b) Data filled at P-band are at the wrong frequency, wrong
bandwidth, and wrong time. No fringes.
c) Data filled at C and X-bands are at the right frequency and BW,
but wrong time. No fringes.
d) No data at all was filled at K-band.
e) The data filled at Q-band gave beautiful fringes, but at the
wrong time -- these are clearly data taken at another freqency!!! There
was also data at the right time and frequency -- but these data have no
fringes.
For the Line observations:
No data at all was recorded at L, P, C, X, or K bands.
Data filled at Q-band have the same characteristics as the continuum
data: Strong fringes at the wrong time (and certainly at the wrong
frequency), data at the right time and frequency have no fringes.
Is there any other diagnostic I can report on?
More information about the evlatests
mailing list