[evlatests] L300 Comb RFI.

[vdhawan at aoc.nrao.edu]

	RFI from the 128 MHz comb in the EVLA LO system.
		   V. Dhawan, D. Mertely.


There is concern over the level of leakage from the comb generator
(L300) which drives the EVLA LO system.

A previous test (October 14, evlatests email, R. Perley) was
inconclusive: EVLA antennas were probably not working. The first 2
EVLA antennas 14 & 16 do not have L300s. The function is handled by
the L302, I'm not sure of the details - clarifications welcome.

This time it worked a bit better, but another follow-up is needed.

Summary:

1. A spare, UNshielded L300 was installed in ant 14. Antenna 14
otherwise normal (after startup glitches).

The L300 requires 65 dB shielding at L-band, 50dB at K-band. See
table, column S_add.

2. The surprize was the detection of comb lines on antenna 16 which
was in normal operation (again after startup glitches). 16's comb
_should_ have been shielded, but was just 10dB below antenna 14 at
1408, even worse at K-band.

Someone please remind me of the expected shielding (for L300/302's 
and similar modules) going from open box to closed, installed in 
rack, with the rack door closed.

3. Another calibration of leakage from vertex room to EVLA receiver
was done, showing again about 70dB at 1397 MHz. Also note-7 below.

4. Another run is required to look for the comb in normal operation 
- we missed some scans. I propose to do it (this week?) at several
bands/settings with enough time to verify fringes etc. to look for 
the comb lines in normal use.

--------------------------------------------------------------------

The details:

A. Spare L300 module in ant 14, powered on, but its signals not used
in ant 14 - output should have been terminated in 50-Ohm loads, (was
it Mert?). L300 with lid off, not in the shielded rack.

B. Observation in narrow-band spectral-line mode, 200KHz BW, 1.5KHz
resolution, RR,LL,LR,RL correlations, centered on a harmonic in each
RF band: 1408.0, 4864.0, 8448.0, 22272.0 MHz, avoiding known VLA
internal and external RFI frequencies. These observations at 85deg.
declination, no celestial source expected.

Also observed with same center but a bit more BW 780KHz, on a 3Jy
continuum source nearby, at each band, to verify that EVLA antennas
were fringing. In this case only 1 X-band scan survived the startup
glitches. In retrospect I could have sprinkled the test with many such
scans, and at the narrowest BW.

The observations were scheduled to repeat 3 times, with (a) L300
OFF. (b) L300 ON. (c) L300 OFF, signal generator on emitting -68dBm in
vertex room. In the event, the first 2 were lost as EA14 and 16 were
misbehaving. (c) was converted to an ON scan, another scan was added
with sig-gen at 1397MHz.  The OFF scan will be made up at the next
opportunity.

C. Results so far: The table shows the detected or upper limit in
autocorrelation spectra of the EVLA receivers by frequency.

---------------------------------------------------------------------

F         14R  14L   16R    16L         Comment        HT    S_add
MHz       dBm  dBm   dBm    dBm

1397.0   -139  -136  <-165  <-165  -68dBm cal signal. -200   61 64

1408.0   -139  -139  -149   -155                      -200   61 61

4864.0   -162  dead  <-165  <-165  14L dead;14R OK.   -196   34?
                                   16 R,L prob. OK.

8448.0   -164  dead  <-165  <-165  14L weak. rest OK. -195   31?

22272.   -143  -140  -145   -145                      -190   47 50

---------------------------------------------------------------------

Table Notes: 

N1. Calibration as before by comparing the line strength to the mean
noise power k.T_sys.BW = -152dBm L, C, X-bands; -148dBm at K-band.
T_sys 30K at L,C,X; 60K at Kband. BW=1.5KHz here.

N2. Mertely has measured the L300 comb emission levels in the 'RFI
chamber' - I need to get those numbers to compare with cols.2-5.

N3. HT = harmful threshold in dBm = 0.1*(k*T_sys*BW) at 1km/s BW.
    Assumed T_sys=30,30,30,60K @ L,C,X,K band. BW=5,16,28,74 KHz,
    and 8 hour integration time.

N4. From the measured level (cols. 2,3,4,5) compared to the HT, I
conclude that the L300 and similar modules need shielding of 65dB at
L-band frequencies, 50dB at K-band, to meet the harmful threshold
requirement in each band. That is, going from open box to sealed box
in rack with door closed.

N5. As stated above, the levels on ant.16 are a shock: it is
supposedly in the shielded rack.  Someone please remind me what the
specs are for the bin/rack shielding.

N6. The porosity of the EVLA L-band horn to radiation in the vertex
room is of special concern because of its construction technique (also
used at S-band). The current measurement at 1397MHz indicates a loss
between vertex room and EVLA L-band receiver of 68-71 dB; not too
different from the 2 measurements on VLA antennas 6 & 22 of 71-73dB
(quoted in the recent VOIP phone tests.) See next item.

N7. The leakage into the EVLA horn with its mouth covered (absorber
plus metal sheet over it) was measured by Dan Mertely and found to be
only ~40-50dB down at 1.4GHz. THIS NEEDS TO BE RECONCILED WITH THE
PREVIOUS STATEMENT - I will sort this out with Mertely.

----------------------------------------------------------------------