[evlatests] Antenna 14 Y-Factor Tests at C-band

[Rick Perley]

       Bob Hayward, Jim Jackson, and I, with the help of Ken Sowinski 
and Mike Revnell,
have determined sensitivity of Antenna 14, at C-band, at various points 
through the signal
chain. 
    For the benefit of the non-engineers on this list, the Y-factor is 
the ratio of the power
between a hot and a cold load.    The hot load was a thermal absorber at 
290K placed over
the feed.  The cold load was the cold sky, as a liquid nitrogen bath for 
the 2-foot aperture of
the C-band feed is excessively large.  For the numbers quoted below, we 
take the sky emission
plus spillover temperature to be 10K.  From these temperatures, and the 
measured Yfactor,
system and receiver temperatures can be determined. 
    We ran this experiment with the 'slow' switched cal on (about 1.4K 
with 5 seconds on
and 5 seconds off).  The apparent value of this cal can be derived from 
the Yfactor.
Conversely, the system temperature can be independently derived from a 
knowledge of the
cal.  (This is mentioned here because of curious discrepencies, noted 
below). 

    We made these measurements at four locations:

    1) At the RF, following the receiver.  The center 
frequency/bandwidth was 4850/600 MHz.

    2) At the output of the LSC converter (T302).  The IF frequency was 
11000 MHz, corresponding
to a sky frequency of 4744 MHz, with a BW = 100 MHz.

    3) At the output of the downconverter (T304).  The IF frequency here 
was near 1720 MHz, and
a 150 MHz wide filter was used.  The center sky frequency was 4744 MHz. 

    4) At the output of the VLA T5 module.  The BW was set to 50 MHz, 
but the power meter is
apparently not sensitive to the bottom 10 MHz, so the effective BW was 
40 MHz, with a center
frequency within the same band measured at the antenna. 

    All power measurements were made with the same equipment we have 
used in the past year,
and at levels between -30 and -42 dBm, where careful calibration has 
assured us the power meter
is very linear.  Because of the overhead of moving the Milhouse laptop 
system, we have not
digitally recorded these data, electing rather to log them manually. 

    For the three antenna measurements, the procedure was simply to 
record the 'hot' and 'cold'
powers, with and without the cal. 
    For the T5 measurements, we had to use a different methodology, as 
Mike assured us the
requantizer could not handle the ~11 dB power range linearly.  Hence we 
inserted attenuation to the
T304 to keep the T5 output power close to -40 dBm, our preferred 
operating range.  This also required
turning down the T5 module analog gain to near its bottom setting.   It 
is felt that this will not bias
the results.  Finally, the AGC was turned off. 
    It was quickly found that the nominal attenuation values of T304 
attenuators are not correct.  This
required a separated calibration of the T304 attenuators, which we did 
with the power meter.  Variances
as large as .76 dB were found, and accounted for. 
    We did two sets of T5 output power measurements.  The first set used 
Mike's new recommended
sample depths, of about 2.8 bits rms.  The second set used the previous 
recommended levels of 9 bits
rms, for which Mike made suitable adjustments internally to prevent 
overflow of the 4-bit requantizers.
For each of these two sets, we adjusted the T304 attenuator in 1 dB 
increments (nominal) to judge
the effect on Y-factor of various bit depths. 

    The weather was outstanding throughout.  Clear, calm, dry. 

    Results

    The results from the three (fully analog) antenna measurements are 
given below.

Test      Y-factor (dB)      Tsys      Tcal
-------------------------------------------------------
    1            10.977            24.30      1.38
    2            10.996            24.18      1.47
    3            10.998            24.17      1.34
-------------------------------------------------------

    The differences in Y-factor, Tsys, and Tcal certainly reflect the 
errors in the
power measurement/logging system.  These are small enough for us to 
confidently
conclude there is no degradation in system sensitivity within the analog 
system up
to the samplers. 

    The results from Test 4 (T5 output) require elaboration. 

    4a    rms level of ~9 to the sampler. 
          For this test, a single 'cold sky' measurement was made.  The 
1-2 GHz T304
attenuator was set to 0 dB to get the rms = 9 condition.  We then ran 3 
'hot load'
measurements, using 9, 10, and 11 dB attenuation in the T304 to keep the 
rms level
near 9, and the T5 output power near -40 dBm.   In the table below, the 
Y-factor
has been adjusted to the correct differential attenuation between the 0 
dB and the
9, 10, and 11 dB settings. 

    Test      Y-factor            Tsys      Tcal
-------------------------------------------------------
    4a-9      11.176            23.12      1.41
    4a-10    11.127            23.40      1.73
    4a-11    11.037            23.94      1.50
-------------------------------------------------------

    Note the Y-factors are a bit higher than the (wider bandwidth) 
values found
at the antenna (or, Tsys is lower), and the cal values are similar.  (As 
these are
differential measures, the errors are probably a couple of tenths of a K). 
    We conclude there is no deterioration of system performance with this
setup. 

    4b      rms level of ~2.8 from the sampler.
             For this test, we did a single 'hot load' measurement, with 
20 dB
attenuation in the T304 to get the signal down to the recommended level
of 2.45.  We then did four cold sky measurements, with the T304 attenuator
set to 8, 9, 10, and 11 dB.  The calibrated differences have been used
in the following table. 

    Test      rms         Y-factor      Tsys      Tcal
--------------------------------------------------------------
    4a-8      3.25       11.316         22.3      0.90
    4a-9      2.9         10.900         24.8      0.90
    4a-10    2.5         10.494         27.4      0.89
    4a-11    2.3         10.022         30.9      0.88
----------------------------------------------------------------

    There are a couple of things to note here: 
    - The Y-factor drops steadily with decreasing cold sky input power. 
The simple interpretation is that the digitizer is adding its own noise,
which is appreciable at levels less than and rms of 3.  You will note
the same trend in the high rms data (Test 4a), but the dependency is
much weaker. 
    - The derived Tcals are at least 50% lower than the correct values
as determined at the front end.  Because of the limited number of bits,
the power levels as seen on the power meter were much noisier, so
higher error is expected, especially for a derivitive value such as the
Tcal.  However, I would not expect all four independent values to
be off the same way (their close agreement MUST be fortuitous!),
and wonder if we have a non-linearity here.  If we calibrate these data
by assuming a Tcal = 1.40, then the true system temperatures are
raised by about 50%, to values near 40K. 

    At this point, I think we need some experts' explanation of these
data, particularly the low-rms results.