[evlatests] EVLA test report

[David Sevilla]

This is a document to report findings of a test performed on EVLA Antenna
13, IFs A and B. The tests were performed while tuned to standard X-band.
An spectral analysis was performed on data collected from
the deformatters to identify possible sources of internal RFI as well as
leakage of electronic noise into the output data. For a time period of 5
minutes for IF A and 10 minutes for IF B, data samples were taken and
processed, for a total of 30 and 60 average spectra respectively.  In each
10 second period, the system was able to obtain obtain approximately 140
scans of 32k samples from IF A and about 125 scans of 32k samples from IF
B.  Statistics, including the average power spectrum, were computed.

For this test the input attenuators in the T304 modules were set very high
to block the incoming astronomical data, and the output attenuators were
set to 0.  This maximizes sensitivity to RFI generated inside the T304.

There were a total 16k channels in 1024 MHz for a resolution of about 62.5
kHz per channel. The main focus of this text is to describe the
differences between IF A and IF B, since extra shielding was added
recently to the T304 downconverter in IF A. It was found that IF B had
significantly more signals with stronger relative values than IF A. Also,
the ~8800 MHz line reported by Rick disappeared from IF A, but still
appears in IF B, which means that the extra shielding added to IF A
restricted the mentioned line. In IF B, such line had a brightness of
about 2000 times more than the background.

Earlier we reported seeing an RFI line in the exact center of the 1024 MHz
sampled spectrum at both C and X band -- a feature probably associated
with either the T304 or the sampler as these are the only analog devices
that are tuning agnostic.  With reduced astronomical "contamination" more
sensitive measurements of this line and related lines were made. Following
is a comparison table of the findings in IFs A and B for every multiple of
64 MHz within the band (with 512 MHz being the previously detected line).
All those lines were found only in a single channel (i.e. narrow band).
The table shows the channel number, the frequency from the edge of the
1024 MHz band, and the brightness value relative to the specific
background value, specified as a multiple of background (which is
reduced by a factor of about 80 compared to normal due to the attenuator
settings).

Channel	Freq. [Mhz]	Power IF A      Power IF B
1024	64		background 	background
2048	128		8 X		2 X
3072	192		background 	background
4096	256		4 X		4 X
5120	320		background 	background
6144	384		60 X		5 X
7168	448		background 	background
8192	512		6 X		22 X
9216	576		2 X		2 X
10240	640		12 X		2 X
11264	704		background 	background
12288	768		3 X		2 X
13312	832		2 X		2 X
14336	896		2 X		2 X
15360	960		background 	background

Most of these are even harmonics of 64 MHz or harmonics of 128 MHz.

Following are some frequencies/channels where other lines were found in IF
B in X-band observation, none of them were found in IF A

Channel	Relative 	Power IF B 	number of channels
 	Freq. in X band
4036	9089.91 MHz	2000 X 		15   "Rick's ~8800 line"
10332	8696.41 MHz	130 X		5
12228	8577.91	MHz	30 X		3

-David Sevilla and Walter