[daip] bad data this last weekend?

[Eric Greisen]

robert becker writes:
 > hi, I had two observing runs this weekend for AH939.
 > When I run fillm I get lots of error messages such as
 > 
 > FILLM1: MCANT: Bad total delay for antenna 23 at   0/18:57:10.0
 > FILLM1: MCANT: Bad total delay for antenna 23 at   0/18:57:20.0
 > FILLM1: MCANT: Bad total delay for antenna 23 at   0/18:57:20.0
 > FILLM1: MCANT: Bad total delay for antenna 23 at   0/18:57:20.0
 > FILLM1: MCANT: Bad total delay for antenna 23 at   0/18:57:30.0
 > FILLM1: MCANT: Bad total delay for antenna 23 at   0/18:57:30.0

     In the current setup, it is possible for an antenna at the end of
the east (?) arm at very low elevation to require a delay beyond the
current capabilities of the system.  This is a temporary trouble until
the EVLA is fully implemented.  The data affected should have been
flagged by FILLM.
 > 
 > or
 > 
 > FILLM1: Number bad nominal sensitivities: ant 19 IF 1 =      8038
 > FILLM1: Number bad nominal sensitivities: ant 19 IF 2 =      8038
 > FILLM1: Number bad nominal sensitivities: ant 19 IF 3 =      8038
 > FILLM1: Number bad nominal sensitivities: ant 19 IF 4 =      8038
 > FILLM1: Number bad nominal sensitivities: ant 21 IF 1 =      8038
 > FILLM1: Number bad nominal sensitivities: ant 21 IF 2 =      8038
 > FILLM1: Number bad nominal sensitivities: ant 21 IF 3 =      8038
 > FILLM1: Number bad nominal sensitivities: ant 21 IF 4 =      8038

This is a summary at the end to show how many integrations had a NaN
recorded instead of a measured nominal sensitivity.  These data too
should have been deleted.

 > 
 > or
 > FILLM1: MCCHED:  0/11:22:33.3  5 x 23  BAD GAIN EXPONENT    -118
 > FILLM1: MCCHED:  0/11:22:33.3 20 x 23  BAD GAIN EXPONENT    -118
 > FILLM1: MCCHED:  0/11:22:33.3 15 x 23  BAD GAIN EXPONENT    -118
 > FILLM1: MCCHED:  0/11:22:33.3 28 x 23  BAD GAIN EXPONENT    -118
 > FILLM1: MCCHED:  0/11:22:33.3 10 x 23  BAD GAIN EXPONENT    -117
 > FILLM1: MCCHED:  0/11:22:33.3 14 x 23  BAD GAIN EXPONENT    -118

Note that these all involve antenna 23 - probably at the times that
the delay was out of range.  This error tends to occur when an antenna
is actually not functioning, but the other ModComp flags have not yet
caught up to that fact.  FILLM is defending itself against 2**(-118)
underflows, asserting that this "gain" cannot be correct.

Eric Greisen