[evlatests] T304 Attenuators Gone Wild? -- Maybe not ...

[Rick Perley]

    I sent around a summary of apparent changes in power levels in three 
antennas (2, 9, and 24), which conceivably could be due to some very 
strange behavior by the T304's step attenuators. 

    Further review of the available data show that the situation is more 
complex than reported -- and likely involves the correlator. 

    The new information comes from a review of the autocorrelation 
spectra ('total power', as determined on the station boards), and the 
input cross-correlations.  (The previous report was derived from 
antenna-based calibration solutions). 

    Below are the revised observations/conclusions:

    A) Only antenna 24 shows the previously reported behavior in the 
autocorrelation spectra.  Antennas 2 and 9 are fine.   In particular, 
the autocorrelation spectra show both the '1-Hz' oscillations, and the 
`1-dB' power changes associated with scan changes. 

    B) All baselines between antenna 24 and the others show unusual 
behavior.  No other baselines have unusual behavior of the types 
described earlier.  ***However*** -- the rapid oscillations (1 Hz) *** 
are not the same* on all baselines.  In particular, the baselines 24 x 2 
and 24 x 9 show rapid oscillations that are completely different in 
character than those between 24 and all other antennas.  (In other 
words, this rapid oscillation behavior does not close). 

    I append four plots to illustrate the behavior.  The first two are 
plots of autocorrelation power for all 12 antennas.  Note antenna 24 (on 
the second plot).  Note that the rapid oscillations are small in the 
beginning, and grow continuously until they are replaced by the `1-dB' 
changes associated with individual scans.  

    The final two plots show all the 11 baselines of the other antennas 
to antenna 24.  Note that 9 of these appear identical to each other, and 
to the autocorrelation plots.  Note that the other two (the baselines to 
antennas 2 and 9) appear the same qualitatively, but in fact are very 
different quantatively -- the 1-Hz oscillations start off as large, but 
end up as very small at the point of transition to the other behavior.

   
-------------- next part --------------
A non-text attachment was scrubbed...
Name: XCROSS
Type: application/postscript
Size: 2279263 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20091112/b83ab9d7/attachment.ai>