[evlatests] S-band Issues
Rick Perley
rperley at nrao.edu
Thu Feb 16 17:17:26 EST 2012
An upcoming deep integration at S-band has encouraged me to take a
close look at S-band performance. It's not all perfect ...
It is clear from the rms noise on the calibrated data that some
antennas are much noisier than others -- high system temperature.
Utilizing the switched power tables, I get the following: ('A' = Tsys
in IF 'A', etc.) (I used an RFI-free band near the center of each 1
GHz-wide tuning. IFs A and C were centered at 2.5 GHz, IFs B and D are
centered at 3.5 GHz).
Antenna A B C D
-----------------------------------------------------------------------
6 31 28 28 27
7 32 32 33 34
8 54 45 66 46
9 29 17 17
15
10 18 16 16 15
12 16 15 29 26
13 21 25 22 27
14 26 25 27 26
15 28 28 27 27
16 24 25 19 24
17 22 24 28 24
18 29 26 28 27
21 28 28 27 27
22 23 25 22 25
23 28 28 29 30
24 56 38 57 40
25 29 31 29 28
26 23 26 24 27
27 33 35 50 62
28 34 32 38 33
-------------------------------------------------------------------------
Expected values are between 25 and 30 K. Anything below 20K is
certainly unrealistic, and can only be due to an incorrect table value
for the noise diode temperature,. I checked the 'CD' table, but it
seems to have reasonable values (at least, not equal to 1.0).
Tsys values greater than 40 are certainly real, in the sense of a
noisy receiver. Some are very much more than 40, and these need to be
checked.
I also checked the values for PDif, to see if they are reasonable
(in the sense of whether they show silly behavior). Two antennas do
show strong evidence of measurement troubles?
ea06 in both RCP and LCP (all IFs). The PDif values are rising and
falling, smoothly, by ~7% (pk - pk). These are not seen in the
visibility data, nor in the PSum values. As they are on all IFs, they
probably are due to something in the receiver. The two polarizations
have different patterns, and the characteristic is much worse on the LCP
side. This problem is seen *only* at S-band.
ea16 has the same problem as ea06, but only in the RCP side.
Otherwise, all PDif and PSum values look reasonable (other than what
RFI is clearly doing).
More information about the evlatests
mailing list