[evlatests] L300 comb generator RFI search

[Rick Perley]

    Some concerns have been raised that the L300 128 MHz comb generator may
be radiating beyond acceptable limits.  At Steve Durand's request, a quick
observation was made this morning to see if obvious signals could be 
detected. 

    It should be noted first that neither antennas 14 nor 16 have an 
L300 in them
at this time.  The functionality of the L300 is provided in these 
antennas, at
this time, by the L302s, which are expected to 'leak' in a similar manner. 

    As the 128 MHz combs will be pure CW, the best way to detect them is
with a correlator in its minimum channelwidth mode.  For the VLA correlator,
this provides 390 Hz resolution, which is obtained with a single IF, 200 kHz
passband, and 512 spectral channels.    I used channel 'A'. 

    As this was an exploratory experiment, I decided to tune to a specific
multiple in L, C, X, and K bands, rather than integrate deeply at any one. 
I chose harmonic numbers 11, 38, 66, and 175, which are located at
1408, 4864, 8448, and 22400 MHz.  Observations of a calibrator -- to
establish if the antennas are actually working -- and of the north pole --
to reduce phase winding to zero, allowing integration for more
sensitivity -- were made.  The north pole integrations were 5 minutes 
long. 

    These observations were made after Barry had completed his bandpass
shape modifications, and before Jim swapped the T304 modules between
channels A and B. 

    The results are disappointing, in that only antenna 14A gave detectable
fringes (on the calibrator), and then only at C-band.   

    For this single band, single IF, single antenna, the autocorrelation
spectrum shows no RFI in channel 256 (where 4864 MHz emission would
be located) to the level of the noise.  This is the case for both the 
calibrator
and the north pole observations, except for the last record of the north 
pole
observation, when strong emission was noted.  However, I believe this
single point can be discounted, as the entire VLA shows 'crazy data' at 
this,
and only this, point in time.

    If we discount this single point, a simple analysis sets the level 
of the
emission.   We don't know the antenna's effective collecting area to the
source of emission, so we make the standard assumption that it is 0 dBi.
In that case,  the spectral power flux density necessary to give a signal
equal to the  rms noise will be  -200  dB W/m^2/Hz.  (This  assumes
a system temperature of  35K,  and employs the BW = 390 Hz, and
an integration time of 300 seconds).   The corresponding power flux
density of this upper limit is -174 dB W/m^2. 
    The limit for emissions set out in EVLA Memo #46 at C-band is
-188 dB W/m^2.  Thus, the limit set by this short observation is
14 dB above the EVLA limit, (and about 6 dB above the ITU levels). 
(Not bad, actually, given the stringent levels we are trying to achieve). 

    Given the considerable doubts about whether the EVLA antennas
were actually working, I recommend that we  re-do this experment,
when we confirm things are working better.