beam 4a has not been working since shortly after alfa was
reinstalled in aug11. It has strong oscillations.
On 06oct11 (in the afternoon) the drain current was lowered
from 17 to 11 milliamps.
The oscillations stopped and 4a started working.
With the drain current = 11mamps, the Tsys was about 90 K.
A 12 db amp was placed after the dewar to increase the power
level. Tsys decreased to about 55 K.
A 16 db amp was also tried but this gave about a 10 kelvin
higher Tsys.
Tsys measurement:
On 10oct11 dana used the expensive spectrum analyzer
to verify that the standard Tsys measurements on bm4a were
correct. He did this with the 12db amp inserted and then with the 16db
amp. Tsys was taken with the mock spectrometers and then the spectrum
analyzer (they were not taken simultaneously). The setups were:
Tsys alfa setup:
Compute Tsys using 172 MHz bw centered at 1450 MHz.
Use the Mock spectrometer.
Winking cal at 2 hz for 3 seconds.
sample at 20 hz. Use 3 out of every 5 samples for computation
(to exclude cal transitions).
Compute Tsys over entire 172 MHz (using cals defined over
entire range).
Tsys spectrum analyzer setup:
center alfa at 1375 MHz.
Use 3MHz rbw centered at 1388 MHz, zero span.
Toggle cal with 1 second toggle. for about 15 seconds.
He did each measurement twice.
The measurement results:
The table below shows the tsys values measured. The
last column shows the gains that the mock spectrometer used for beam 4a
and beam 4b (a normal beam).
setup
TsysBm4a
[K]
spec Analyzer
TsysBm4a
[K]
mock Spectrometer
mockGain [db] bm4a,bm4b
12db amp
72.1
71.5
52.9
18/11
16db amp
59.0
60.7
62.3
11/11
The 16 db measurements made with the spectrum analyzer and the
mocks agree. The small variation may be from Tsys variation with
frequency.
For the 16db measurement, the gains used by the mocks was the
same as beam 4b (a normal beam). So the 16db amp gives the power level
normal for alfa. The 12 db amp measurement had 7db more gain in beam 4a
of the mocks.
Top: average spectra for calOn,calOff for both sets of
measurements
Black,Red: 12db
caloff,calon average spectra
Green,Blue:
16db caloff, calOn average spectra.
The spectral shape changes between the 12db and 16db
measurement.
This could be the different attenuations used in the mock
mixer chassis (-18,-11) but it is probably that the 12db,16 db amps
have different bandpass shapes.
middle: Plot the total power vs time for the 12db and 16db
measurements.
For these plots, the total power was computed as the mean of
each spectrum.
For the online Tsys measurements the program removes rfi (by
fitting to the cal dif) before doing the sum).
Black, red show the 12db and 16db mock
measurements.
The dotted lines show the computed average values.
Bottom: Total power Vs time in CalDeflection units (calOn-calOff).
Black, red show the 12db and 16db mock
measurements.
The lower line of each measurement is Tsys(caloff) in CalUnits
The upper line of each measurement is 1 above the Tsys line
(since the units are caldeflection).
It is assumed that the calDeflection in both cases does not
change in time (or else the whole thing is a waste of time !).
The ratio of the two measurements Tsys16/Tsys12 give a
ratio of 1.176.
The ratio of the reported Tsys values (from tsys alfa using the
same data is:)
62.3/52.9 =1.178
The difference1.176,1.178 is probably do to the rfi removal
in the tsys alfa computation).
Summary:
Non simultaneous Tsys measurements can vary because the sky
contribution is changing. A .1 jy source will change Tsys by about
1degK.
The Tsys measured using the spectrum analyzer and the mocks are
the same for the 16db amp
The Tsys measured using the spectrum analyzer and the mocks
differ by about 10K with the 12db amps.
The mocks measure Tsys12= 52.9 and Tsys16=62.3K.
You would expect the 16db amp to have a "better Tsys" than
the 12db amp since the power is still low for the 12db amp.
Reasons why this might not be true are:
The noise figure of the installed 16db amp is worse than
the noise figure of the installed 12db amp.
The 16db amp is causing compression somewhere in the signal
path.
The spectrum analyzer measures Tsys12=72K while Tsys16=60K.
Tsys16 matches that of the mocks.
You would expect Tsys16 to be better than Tsys12.
Reasons why Tsys12 from the spectrum analyzer might be high
The signal was taken from the front panel output and the
circuitry may have different noise figures than the path measured using
the mock.
I checked the schematics, and it looks like the front panel
output has the same amps/pads as the signals that go to the mock
spectrometer.
on 11oct12 more measurements were done with the spectrum
analyzer, mock spectrometers, and square log detectors.
We found that the bm4a levels going into the alfa filter bank
were too low. The filters were adding about 10K to Tsys.
This doesn't explain the differences, but the power levels we
were using were too low..