[evlatests] P-Band sensitivity

[rperley at nrao.edu]

The observations made last month to investigate the low frequency flux
scale included brief observations of blank fields.  In preparation for the
upcoming low frequency talk in Pune, I've made 'blank field' maps from
these observations.  From these, I can determine the noise and from this
the array sensitivity.  There seems to be a problem.

>From Paul's spreadsheets, the system temperature is measured to be about
100K.  If the efficiency of the antenna is ~ 0.4, then the expected SEFD
is:

SEFD ~ 5.62*Tsys/effic ~ 1500 Jy.

However, the noise measurements give a very different answer.  Using a
single 250 kHz channel, in an RFI-free subband (#8), on a blank field --
and after removing the obviously noisy antennas, an image made with 15570
visibilities, in single ('RR') polarization, and 3 seconds integration
provided an rms noise of ~40 mJy.  This corresponds to an SEFD of well
over 6000 Jy -- four times too high.

Just in case the extra noise came from various sidelobes, I made a really
large image of the entire beam, using almost the entire subband width, and
both polarizations.  This gave a noise of 6 mJy/beam, corresponding to an
SEFD of nearly 9000 Jy.

I also made noise maps for three clear channels in each of the 16
subbands, to see how the sensitivity varies with frequency.  As expected,
the noise is very high at the lowest frequencies (10 times higher at 220
MHz), and dropped steadily as the frequency rises, reaching the 'best'
value in subband 7.  Surprisingly (to me, at least), the noise then stayed
at about this level all the way out to near subband #15 (440 MHz) before
rising a bit in the last two, highest frequency subbands.

I had expected a clear maximum sensitivity in the middle, and decreasing
sensitivity at higher frequencies, due to increasing defocusing.  But this
appears not to be the case.  I didn't think the decreasing galactic
synchrotron contribution to be enough to offset the defocussing, but
perhaps I'm wrong.

On the other hand, the apparent fact that the best sensitivity is four
times worse than expected indicates either I am doing something very wrong
in the calculations, or there's something very wrong with the system
temperature and/or aperture efficiency.