[evlatests] 12.5 MHz noise tests for different numbers of antennas

[Mark Claussen]

I used some dynamic time before software today to try to test the 
12.5 MHz extra noise problem, changing the number of antennas, as 
Barry suggested, so that a possible timing problem somewhere (VLA
correlator controller ?) might be pinpointed.

I had an observing file at C-band which repeated essentially what
Rick has done in the past:  Observe in mode 1A at 50 MHz, giving
a channel width of 3.125 MHz, in mode PA at 25 MHz, giving a channel
width of 3.125 MHz, and again in mode PA at 12.5 MHz, which gives
a channel width of 0.781 MHz (four times narrower than the 50 or
25 MHz modes).  Observations were made of 3C147 as a calibrator and
then for 3 min toward a field selected by Rick, which at least does
not have any strong sources in it.  The integration times were 1.67s.
The file was run four times:  once with 26 antennas, once with 20,
once with 10, and once with 5 antennas.   I calibrated the data and 
then looked at the histograms of the noise.  The AIPS program UVHGM, 
as Rick has pointed out, will, if asked, make a Gaussian fit to the 
histogram, and report the center and the rms noise.

Below is a table I made of the rms noises reported by UVHGM.  Note
that as Rick has reported several times, in the 12.5 MHz table, for
26 antennas, the noise in parallel polarization (RR, LL) is elevated,
in this case, by about 30% over the cross-hand polarization (RL, LR).
The cross-hand polarization is close to the right noise level, when
compared to the 50 or 25 MHz, although one might argue that even
they are elevated by 15% or so.  In addition (again, as Rick has
reported) the histograms of the parallel-hand polarizations differ 
significantly from a Gaussian distribution (see the plot 26antennas.pdf)

When I reduced the number of antennas to 20, the discrepancy between 
the cross and parallel-hand noise distributions went away.  When 
comparing the 12.5 MHz noises to the 50 MHz mode 1A noise, these agree 
to about 7%.  However, one might argue that the effect still exists, 
based on the histogram plots, as there is still some deviation from the 
Gaussian fit at low levels (see the plot 20antennas.pdf, and note that 
the order of RR, etc. is different in this plot than the previous one
--- see the header information for the proper polarization).  The 
generally lower noise levels in the PA 25 MHz channels may just be 
an effect of the bandpass shape (not as many spectral channels in the 
25 MHz observations as the other two) since I didn't correct for the 
bandpass.  

The effect for 10 and 5 antennas is no longer discernible.

I also tried averaging channels together in the 12.5 MHz to see if the 
noise was reduced as we expect it to in such a case.  Even for the 26 
antenna case, the noise averaged down as expected for averaging by two
and four channels.


Mode 1A  50 MHz, channel bandwidth 3.125 MHz

No. of antennas          RMS Noise for channel 7
 
26                              0.144
20                              0.142
10                              0.141
 5                              0.142


Mode PA 25 MHz, channel bandwidth  3.125 MHz
No. of Antennas        RMS Noise for channel 4
                       RR       LL       RL       LR
26                   0.139    0.133    0.138    0.135
20                   0.134    0.134    0.136    0.134
10                   0.142    0.137    0.145    0.143
 5                   0.134    0.133    0.134    0.130


Mode PA 12.5 MHz channel bandwidth 0.781 MHz
No. of Antennas        RMS Noise for channel 7
                       RR       LL       RL       LR
26                   0.411   0.395    0.303    0.311
20                   0.298   0.301    0.297    0.280
10                   0.302   0.293    0.301    0.305
 5                   0.304   0.288    0.274    0.287 


Mark
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 26antennas.pdf
Type: application/pdf
Size: 16406 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20080708/aa831ed3/attachment.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 20antennas.pdf
Type: application/pdf
Size: 15896 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20080708/aa831ed3/attachment-0001.pdf>