[evlatests] On-Line Averaging is working.

[Rick Perley]

    Tests were run today to see if the on-line averaging is working again. 

    Short answer:  All seems fine.  We can move ahead with regular 
observing.

    Details:

    There are four databases, all centered at 2627 MHz. 

    1) 128 channels of 500 kHz width each, averaging = 1 second.
    2) 64 channels of 1000 kHz each, averaging = 1 second.
    3) 128 channels of 500 kHz width, averaging = 5 second.
    4) 64 channels of 1000 kHz each, averaging = 5 second. 

    A quick calibration was done on each, using 3C286 as reference.  The 
targets were J1800+7828 (dot calibrator), and a noise field 10 minutes 
of time away. 

    The rms noise (as determined using histogram widths, in the 
cross-hand channels), with Hanning smoothing applied:
    1) 263 mJy
    2) 185 mJy
    3) 127 mJy
    4) 87 mJy.

    The result from (1) predicts 118 mJy for (3).   The observed is 127 
mJy -- a little high.
    The result from (2) predicts 83 mJy for (4).    The observed is 87 
mJy, also a bit high. 
   
    The small discrepancies between predicted and actual are likely due 
to single antenna effects -- I made no attempt to remove edge effects 
(this will preferentially affect the 5 second data, I should think), or 
bad antennas (other than ea23LCP, which is very noisy). 

    I conclude that on-line averaging is working as expected.