[evlatests] L-Band EVLA and VLA Sensitivity

[Rick Perley]

I reviewed a little more closely the L-band tuning/sensitivity tests 
from last Friday evening. 

For this report, I have analyzed only the IF `D' data (IF #2, LCP) which 
sample the spectrum between 1212.5 and 1987.5 MHz, at steps of 25 MHz. 

There were eight working EVLA, and 19 working VLA antennas which gave 
good stable fringes.  No phase peculiarities of any kind (jumps, spins, 
whatever) were seen.  

Not all the eight EVLA antennas tuned at each frequency.  Antenna 13 
failed to tune at five of the 33 frequencies, antenna 18 at four of 
these.  Antennas 14, 16 and 17 tuned up at all 33 frequencies.  Antennas 
24 and 26 were late in joining the array due to work being done on 
them.  These two failed to tune at the highest three frequencies (1937, 
1962, and 1987 MHz) for reasons unknown -- they tuned correctly in the 
other IF (at 1925, 1950 and 1975 MHz) at the same time. 

RFI had a significant effect at only three of the 33 tunings:  1537 MHz 
(which was completely wiped out), 1562, and 1687.  For these latter two, 
only ~4 of the 15 spectral channels (about 3 MHz) were unusable, but it 
is clear from the generated sensitivities that the input power from the 
RFI has significantly lowered the effective sensitivity, even in the 
unaffected channels.  A much smaller decrease is seen at 1312 MHz. 

At no other frequency was any interference, or any effect attributable 
to interference, observed! 

Observations of 3C147 were used to calibrate the data -- the spectrum of 
this source is well known, so the sensitivities given below are 
correctly adjusted for the source flux density.   The elevation of the 
observation was 56 degrees.  Three of the 15 channels, selected from the 
middle of each band, were utilized for the sensitivity calculation. 

I used the square root of the `AIPS Weight' as the sensitivity metric -- 
this is proportional to Ae/Tsys. 

A) VLA Sensitivity. 

The mean sensitivity (average of 19 antennas) is flat between 1350 and 
1712 MHz.  The sensitivity declines by ~15% between 1350 and 1237 MHz, 
below which it drops (dramatically!) to zero.  (There are 'islands' of 
VLA sensitivity below 1212 MHz, but they are narrow).  The VLA 
sensitivity plummets to zero above 1737 due to a low-pass filter 
inserted long ago, apparently to block an RFI signal which is apparently 
not now present. 

The spread of sensitivities amongst the VLA antennas is about 25% about 
this mean.  It is reasonable to attribute this to the variation in Tsys 
between the receivers. 

B)  EVLA Sensitivity. 

Two of the 8 antennas had a behavior clearly different than the rest:  
Antenna 13 has a continuously declining sensitivity from low (1212 MHz) 
to high (1987 MHz) frequencies.  This is known to be due to poor high 
frequency gain, and will (eventually) be corrected.  Antenna 24 gave 
rather poor (half) the sensitivity of the others bewteen 1600 and 1900 
MHz -- the cause is unknown. 

For the six others, the following conclusions apply, which I believe are 
valid for all other IFs:

- EVLA sensitivity is better than the VLA mean below 1400 MHz.   By 1237 
MHz, the mean EVLA sensitivity is 33% better than the VLA's.   (The four 
best EVLA antennas are nearly 50% better than the mean VLA, and better 
than even the best VLA antenna, between 1200 and 1450 MHz). 
- Between 1500 and 1700 MHz, the EVLA is less sensitive than the VLA, 
but only by a small factor -- about 10 to 15%. 
- Above 1700 MHz, the EVLA stands alone, (of course), with a sensitivity 
which rises by about 15% from 1700 to 1950 MHz.  There is a modest 
fall-off of perhaps 10% between 1950 and 2000 MHz.