System temperature versus za.
I've gone through the telescope data sets taken
with the correlator (april thru july 2000) and extracted all of the cal
on/off pairs (about 15000).The system temperature in the calOff portion
of the cycle was computed using the tabulated cal values.
The plots can be found at:
Computing Tsys using the cal onoff pairs.
calOn/Off pairs typically track a position on the sky
for 10 seconds with the cal on and then continue tracking the same position
(on the sky) for another 10 seconds with the cal off. If the observer
was position switching then the sequence is: positionOn, positionOff, calOn,calOff
where the calOn/calOff pair tracks the off source position.
The system temperature is computed using the total power
information (0 lag). If [K] is kelvins and [corTP] are
correlator total power units (linear in power but arbitrary) then:
delta[corTP]= calOn[corTP] - calOff[corTP] gives the
size of the cal in [corTP] units
calTPtoK = calSize[K]/(delta[corTP]
is the size of the cal in K / size of the cal in [corTP] so
will convert [corTP] to kelvins
Tsys[K] = calOff[corTP]
* calTPtoK is the system temperature.
Tsys is computed for the calOff scan so
that the cal is not included. When computing (calOn - calOff) we
assume that the difference is only do to the cal. For small cals and high
zenith angle, there is a small change from tsys(za) dependance since
we continue to track the same position in the sky for the calon and caloff.
Fitting the system temperature za dependance:
The beam should start to spill over the dish at 15 deg
za and start to leave the ground screen at 17 degrees. Looking at the data
there is a linear increase out to 15 degrees za and then it turns up. I
fit a linear term to za 0-20 and then a 2nd and 3rd order term for za >
14 degrees. To remove outliers, I did the fit, removed all points greater
than N sigma (this is labeled on the plot) and then refit the data.
About the plots:
There are two plots for each data set. One has the data labeled with +
polA, * polB . The second plot has both Pols labeled with a
period so you can see the fit to the data.
The 3rd order fit is: Tsys[za] = c0 + c1*za + c2*(za-14)^2 + c3*(za-14)^3
where za is in degrees and the last two terms are only included for
za > 14. The polA coefficients are in column 1 and those for polB
are in column 2.
The sigmas for the final iteration of the fit is in degrees kelvins.
the number of points started with and the number of points used in the
what sigma value was used to iterate the fit.
The bandwidth * integration time is of order 10Mhz*10 seconds so 1/sqrt(b*tau)
or .01 %. For system temperatures of 30-40 K this is 3milliKelvins.
The fit rms of order .5 kelvins for lband is most likely from confusion
lband narrow 1400-1425 Mhz
1400 through 1425 Mhz were kept giving 2157 samples. There is only 1 tabulated
cal value near 1400 Mhz. The outliers for pol A + occurred when the receiver
was warmed up and then cooled down during the first part of july. polA
came back 4 degrees K higher than before the cycle. The solid line in the
second plot is the fit. The dashed line is the fit to data taken in jul98
normalized to match this data at 10 degrees za. The edges are different
but the slope 5 to 15 degrees is the same.
lband wide 1300-1425 Mhz then 1650-1670 Mhz
lband wide was broken into two sets 1300-1425 and then 1650-1670 Mhz (OH).
The scatter in the OH may be from the galaxy in the off position raising
the system temperature. Both of these bands show the linear increase in
tsys 0 to 15 degrees. Also note that Tsys(1665) is about 2 degK < Tsys(1400)
for lbw. The lbw OMT provides linear polarization. There is a hybrid after
the dewar (switch selectable) that converts this to circular polarization.
This data was taken in circular mode. The tabulated cals are for the linear
mode. The cal values used here are the average of the polA,polB linear
sband narrow shows a flat tsys(za) out to 15 degrees za and then curves
The cband data comes from a single observer who used 4.1,5,5.9 Ghz. Each
frequency has a separate curve. This could be the receiver or the cal (we
need to check the sefd plots for the calibrators the user had). Each individual
curve looks pretty flat out to 15 degrees za.
Things to think about:
The Tsys(za) curve is not flat for 0 to 15 degrees for lbw,lbn. It is flag
for sbn and cband. These receivers differ in the horns illumination of
the tertiary. If this is a tertiary spillover problem then we still need
to find a reason why the spillover is a function of za. It might be a good
idea to revisit the reasons why we wanted to overilluminate the tertiary
with these horns.
The Tsys(za) curve turns up at 15 degrees rather than 17 degrees. Is this
from the taper of the beam illumination on the dish, diffraction, or is
the ground screen seeing some of the ground?
Tsys vs za polynomials:
The polynomial fits to Tsys vs zenith angle are listed
below. The ^2 and ^3 order terms are only used for za > 14 degrees. The
za is in degrees.
lbn 1400-1425 Mhz
polA Tsys(za)=29.474 + .13599*za + .215752*(za-14)^2 - .020024*(za-14)^3
polB Tsys(za)=27.315 + .12809*za + .207750*(za-14)^2 - .019014*(za-14)^3
lbw 1300-1425 Mhz circular polarizations
polA Tsys(za)=36.550 + .12215*za + .237660*(za-14)^2 - .015145*(za-14)^3
polB Tsys(za)=35.739 + .14101*za + .197705*(za-14)^2 - .007744*(za-14)^3
lbw 1650-1670 Mhz
polA Tsys(za)=34.556 + .10924*za + .215577*(za-14)^2 - .017823*(za-14)^3
polB Tsys(za)=34.382 + .09409*za + .241162*(za-14)^2 - .020710*(za-14)^3
polA Tsys(za)=26.000 - .00016*za + .078998*(za-14)^2 + .005007*(za-14)^3
polB Tsys(za)=25.127 - .00738*za + .061695*(za-14)^2 + .009357*(za-14)^3
processing: x101/tsys/Readme for a description
plotting : x101/tsys/plotall.plt