[evlatests] P-band switched power leakage into L/S (wide-band version)

[Paul Demorest]

hi all,

A while back I sent some results from a test that compared S-band 
switched power measurements done with the P-band cals switching vs off 
(see evlatests email on 2016/06/17).  That previous test was done using 
a single 128-MHz-wide spectral window centered near 3 GHz.  I've since 
repeated the same test using the full receiver BW at both L and S bands.

Results for a few antennas are attached.  The top panel of the plots 
shows the ratio of the P-band cal signal to the L or S cal, measured in 
two different ways: 1) the usual switched power table values; 2) WIDAR 
autocorrelations, which provide better frequency resolution.  These two 
methods seem to give consistent results.  The bottom panel shows the raw 
"on-off" autocorrelations with either the P-band cals on (red) or with 
the in-band (L,S) cals on (blue).

The main takeaway from this is that this P-band cal leakage has lots of 
structure on ~10-100 MHz frequency scales.  My previous ordering of 
which antennas were "good" or "bad" mostly reflects whether or not they 
happen to have a big spike near 3 GHz.  For example ea03 (the worst one 
last time) would have looked great if I'd done the original test near 
2.5 GHz instead.  When viewed across the full 1--4 GHz, all the antennas 
have similar features somewhere in the band.

This may not be a big problem for current observing practices, where 
flux calibrators are observed frequently and the switched power is 
mainly used as a kind of intermediate scaling.  I think this effect 
would be an issue if we were to move towards a scheme where the switched 
power is used to set flux scales.  Also while the average power level 
increase across >100 MHz bands is not too big, over small freq ranges 
this may be large enough to noticeably affect Tsys.

In general this seems like a messy thing to be happening and I think we 
should either filter the P-band cal signal appropriately, or else just 
turn off the P-band cals during non-P projects.  I realize VLITE is 
always observing P-band, but it's not clear to me whether or not they 
actually make use of the switched power signal; would be good to get a 
definitive answer to that.

Cheers,
Paul
-------------- next part --------------
A non-text attachment was scrubbed...
Name: P_to_LS_leakage.ea03.png
Type: image/png
Size: 91869 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20160812/846d625e/attachment-0003.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: P_to_LS_leakage.ea13.png
Type: image/png
Size: 88621 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20160812/846d625e/attachment-0004.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: P_to_LS_leakage.ea18.png
Type: image/png
Size: 78450 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20160812/846d625e/attachment-0005.png>