Lbw cal values measured 23feb06
28feb06
Links to PLOTS:
hcorcal:
Fits
to the Average CalDeflection/Tsys (.ps) (.pdf):
Hcorcal
in kelvins (.ps) (.pdf).
diagnostics:
CalDeflection/Tsys
for the individual passes (.ps) (.pdf).
Fits
to the CalDeflection/Tsys for the individual passes (.ps) (.pdf).
othercals
The
average calValues in kelvins and the fits (.ps) (.pdf):
Over
plotting all of the cals (in deg K) (.ps) (.pdf).
diagnostics:
CalDeflectionCalX/calDeflectionHcorcal
for the 8 passes (.ps) (.pdf)
Fits
to the CalDeflX/calHcorcalDefl for the 8 passes (.ps) (.pdf).
Over
plotting the new and old cal values (.ps) (.pdf).
Links to SECTIONS:
Why the cals were remeasured.
Measuring the high correlated cal using
blank sky and absorber
Measuring the other cals on blank sky
relative to the high correlated cal.
Why the cals were remeasured.
New diodes were installed on lbw on 23nov05. On
04dec05 polB cal value jumped by about 20 %. The sky/absorber and
other cal calibration was redone on 23feb06. The cals were installed
in 28feb06 and backdataed to be valid from 06dec05.
Measuring the high correlated
cal using sky and absorber: (top)
The high correlated cal value (diode 1 going to
polA and polB) for lband wide was measured 23feb06 on the telescope using
the sky/absorber
technique.
The absorber was done about 3 pm. The hcorcal sky was done at 6pm. The
observations used 3 second calon followed by 3 second cal off. For the
sky observations, blank sky was tracked. In both cases (sky and absorber)
the faa radar blanker was used as well as the lbw filter bank.
The temperatures used in the computation were:
Tabsorber |
302 K |
Tsky |
5 K |
Treceiver |
from test shack feb03 |
Tscattered |
15 K |
The band 1120 to 1720 Mhz was covered 9 times on
absorber and 27 times on sky. The ratio (CalOn-CalOff)/CallOff was then
computed for the data. Each spectra of 600 Mhz (6144 points) was then fit
to an 8th order harmonic and 1st order polynomial (the order was chosen
to include the ripples in the caldefl/Tsys spectra). The fit was iterated
throwing out points whose residuals were greater than 3 sigma . Whenever
a point was excluded, 5 points adjacent to the fit were also excluded .
After examing the fits, 3 passes in polB (while
on absorber) were excluded because of jumps in the 1320 to 1420 Mhz band.
When on sky, 3 strips in pol A and 7 strips in polB (out of the 27) were
exlcuded. This was mainly from jumps in 1 single 25 Mhz band near 1280
Mhz.A robust average of the passes was then computed (iterating and
throwing out outliers). The average spectra was then fit with the same
function. See reducing
the cal data for more info on the reduction.
The results of the reduction are:
-
Fits
to the Average CalDeflection/Tsys (.ps) (.pdf):
This shows the average Tcal/Tsys data with the fits over plotted in red.
The top two plots are on the absorber (polA,polB) while the bottom
two plots are on the sky. The number of passes through the frequency
band used in each average were: polA abs:9, polB abs:6, polA sky: 24, polB
sky:20. The units are Tsys (about 30K for sky and 300 K for absorber).
The fitRms is computed for the fraction of the spectra used in fitting.
The rms and fraction of spectrum used are printed on each plot. The
radiometer equation should give:
rms=sqrt(2ratio)./sqrt(25e6bw/256chan*3secs*9loops*2hanning)=.0006
The absorber fits match this. The sky fits are about 5 time larger. This
is because the fits are not fitting the 1 Mhz standing wave from the dish.
This is ok since that ripple should not be in the cal value anyway. You
can also see that the radar 1240/1260 still gets in through even on the
absorber.
-
The Hcorcal
in kelvins (.ps) (.pdf).
-
The first two plots show the cal fits in kelvins measured from the Sky,
absorber, and the sky, absorber ratio (Y factor). The top plot is polA,
the middle plot is polB. The dashed line is the receiver temperature used
for calSky. The calAbs and calY agree while the calSky is not so close.
The calsky can be shifted up or down by changing the amount of Tscattered.
The Trcvr curve will also affect the Calsky the most.
-
The bottom plot is the cal In kelvins from the Y factor. The * are spaced
every 10 Mhz. PolA is black and polB is red. These are the values that
will be used for the cal.
-
The major ripple has a period of about 80 Mhz. It is visible in both polarization's
(although the level/phase changes a little). It is there on the absorber
and the sky so it is not coming from in front of the horn. If it is a reflection
in a cable, then it would be about 1.3 meters long (assuming index of refraction=1/.68).
Both of these signals are coming from diode 1. The cal measurements using
diode 2 also have a ripple.
Diagnostics:
The first set of plots show the calOn-caloff/caloff
for each pass through the data. The second set over plots the fits to each
pass to see how stable the system is.
-
CalDeflection/Tsys
for the passes (9 on abs, 27 on sky) (.ps) (.pdf).
-
The first page shows on absorber for the 9 passes through the receiver
band. The top plot is polA while the bottom plot is polB. The spectra have
been offset for plotting purposes. The units are Tsys (on absorber it is
about 300K). For polB the band between 1320 and 1420 Mhz had some
jumps (in strips 0,3,5). These strips were not used in the computations.
The jumps are over the entire 100 Mhz band. This used 4 correlator boards
and 4 digitizers so it is probably not in the correlator. The 1320 to 1420
band is the only 100 Mhz that uses the 1230 to 1470 Mhz filter so this
may be part of the problem .
-
The second page shows on the sky for the 27 passes through
the receiver band. The top plot is polA while the bottom plot is polB.
The spectra have been offset for plotting purposes. The units are Tsys
(about 30K on the sky). The dashed lines in the bottom plot show the 100
Mhz boundaries (that were taken separately). The 25 Mhz band centered
as 1280 Mhz is not evenly spaced. This band jumped in both polA and polB
(it may have something to do with the strong radar signals in the adjacent
bands. There is also a jump in the 100 Mhz band polA for 1420-1520. This
strips were excluded from the averages.
-
Fits
to the CalDeflection/Tsys for the 9/27 passes (.ps) (.pdf).
This over plots the fits to each pass (6144 points covering the 600
Mhz.). You can see the jumps in the data for polB on absorber. The absorber
fits vary by more than the Sky fits since deltaTsys is a larger fraction
of the cal when you are on absorber. The 1220 to 1320 band varies for polA
and polB do to the aerostat radar at 1240/1260.
processing: x101/llb/cals/feb06/hcorcal/lbwinp.pro,lbwfit.pro,lbwcmp.pro,lbwplot.pro
Measuring the other
cals using sky and the high correlated cal (top)
The high correlated cal was measured (see above) using
sky and absorber as the hot and cold load. The other cals were then measured
relative to the high correlated cal on 23feb06 starting around 18:30. Blank
sky was tracked, the filter bank was used, the radar blanker was on,
and the following cal sequence was run:
-
hcorcal(on,off)
-
hcal(on,off),hxcal(on,off),h90cal(on,off)
-
hcorcal(on,off)
-
lcorcal(on,off),lcal(on,off),lxcal(on,off),l90cal(on,off)
-
hcorcal(on,off)
100 Mhz at a time was measured (4 by 25Mhz) going from
1120 to 1720 Mhz. The cal was cycled on/off for 3 secs at each step.
The entire frequency range was repeated 8 times.
The ratio (calOnX-calOffX)/caloffX was computed
(X is the other cals) and then it was divided by (calOnHcor-calOffHcor)/calOffHcor).
A spectrum for the entire pass was then constructed of the other cals relative
to the hcorcal. The spectra for the 8 passes were averaged. The average
spectra was multiplied by the hcorCal value in kelvins (this removed the
hcorCal shape). The resulting spectra was fit with an 8th order harmonic
and 1st order polynomial. For more info see
computing the cal value.
The results of the reduction are:
-
The
average calValues in kelvins and the fits (.ps) (.pdf):
The 8 passes have been averaged together and then multiplied by the hcorcal
fit (in kelvins). There are 14 plots. 7 cals each with polA and polB. The
red lines are the fits to the data. The fit rms's are better than 1% over
the full band. The only problems appear around the 1250-1290 where the
radars occur.
-
Over
plotting all of the cals (in deg K) (.ps) (.pdf).
The top plot is the high cals and the bottom plot is the low cals. The
solid lines are polA while the dashed lines are polB. There are two sets
of lines that follow each other. That is because the same diode always
feeds two types of cals (e.g. diode1 goes to polA for hcorcal and for huncorcal).
Diagnostics:
-
CalDeflectionCalX/calDeflectionHcorcal
for the 8 passes (.ps) (.pdf).
There is 1 page for each calType (7 pages). The top plot is polA and the
bottom plot is polB. The 8 passes through the freq range are over plotted
with an offset. The units for the y axis are TcalHcorcal since each of
the cal deflections have been divided by the hcorcal deflection.
-
Fits
to the CalDeflX/calHcorcalDefl for the 11 passes (.ps) (.pdf).
This over plots the fits to each pass (6144 points covering the 600
Mhz.). These fits include all strips (not all strips were used for the
averages).
-
Over
plotting the new and old cal values (.ps) (.pdf).
The solid lines are the new cals. The dash lines are the old cal values.
The plots are:
-
Top HiCalsPolA: black Diode1 -> polA, Red diode2->polA.
-
2nd HiCalsPolB: black Diode1->polB, Red diode2->polB
-
3rd LoCalsPolA: black Diode1 -> polA, Red diode2->polA.
-
4th LoCalsPolB: black Diode1->polB, Red diode2->polB
The old cal values were for the same diodes before polB calvalue jumped.
You can see in the plots the offset of polB from the new to old cals. There
is also a change in the high frequency values for polA.
processing: x101/lb/cals/feb06/othercals/lbwinp.pro,lbwcmp.pro,lbwfit.pro,lbwplot.pro
home_~phil