[daip] APCAL suggestions and bug report.

[Craig Walker]

First the suggestions:

1.  Make the Sec Z plot scale limit apply to the opacity plot in
addition to the Tsys plot.  Now, it just goes to some high value,
sometimes VERY high.

2.  Make it more clear what the units of "OPACITY" in the plots
is.  I'm still not sure.  A nominal values seems to be near 1
which might mean that it is a gain value.  But I suppose it might
also be percent.

3.  Somehow, I repeatedly failed to notice the inputs note that 
TREC and TAU0 are used as the starting point for the fits and failed
to specify them, then got frustrated when nothing was fit.  I don't
know whether to ask for more obvious notes in the inputs.  But 
perhaps, if OPAC or GRID is specified, but the TREC and TAU0 are 0,
the program should complain.

Now the bug report:

I was getting very strange behavior.  "OPACITY" below 1 was derived for
a block of time later in the experiment (about 2 hours) and the results
were clearly spurious.  Plots of the OPACITY showed sort of a 2 state
situation.  The SN gains also had a region that looked really wrong.
This happened with both GRID and OPAC.  But with no opacity correction,
all was ok.  This morning I got suspicious about array sizes so I
grabbed a copy of APCAL.FOR and modified one number - the value of the
PARAMETER MXVAL.  I increased it from 4096 to 24096.  With that one
change, the program works fine.  My TY table has 17595 rows, which I
suspect is the critical number.  There are so many because this is a 43
GHz phase referencing experiment with 15 second scans - nearly 1900 of
them!

So two things should be done, IMHO.  The size of MXVAL should be 
increased significantly.  And overflow of the arrays that use MXVAL
should be trapped, as it obviously was not.

Cheers, 

Craig

PS.  Also, don't forget my last message concerning use of the WX table
by APCAL.