SEFD vs freq for new lbw receiver
On 17apr03 1 minute on/off position switching
was done on 3C138. The frequency stepped through the range 1100 to 1700
in 100 Mhz junks (4*25Mhz bands). This was repeated a second time going
from 1110 to 1710 Mhz in 100 Mhz junks. The System Equivalent Flux Density
(SEFD) was computed using (polA + polB)*.5 (since 3C138 is 7.5% polarized
at 1400 Mhz). This uses the flux density of the source (taken from
chris salter's fits). It does not use the cal values.
The first set of data covered za 10.7 to 6.5 degrees.
The second set covered 5.7 to 2.0 degrees za. The plots shows the
SEFD vs frequency for the two frequency passes.
Any pointing errors will increase the sefd (since i did not search for
the peak). With the current model they are probably small. The two passed
repeated pretty well.
Top plot. This is the sefd vs frequency at the full resolution of the data
(25 Khz). Red is the first pass through the frequency and blue is the second.
Center plot. Each band pass was cumfiltered
(across frequency) and then the median was computed from the filtered data.
Red is pass 1 and blue is pass 2. The black line is a 3rd order polynomial
fit to the data (excluding the one outlier). The squares are the SEFD from
the old receiver in dec02 on 3C138 with za < 11 deg.
Bottom plot: This is Tsys for (polA+polB)/2. for the off position spectrum.
The green line is a 3rd order fit to the data.
The SEFD is 2.5 Jy/Tsys down to about 1450 Mhz. It
then begins rising until it reaches 4+ at 1100 Mhz. It is better than the
old lbw receiver down to about 1180 Mhz. The lower plot shows that the
system temperature rise is the major cause in the SEFD increase (and not
a loss of gain). The Tsys values use the measured cal values, so you can
look at it the other way and say that the cal values are probably pretty
good since they cause the Tsys to follow the SEFD increase (this then assumes
that the gain is constant).
The sefd fit should work pretty good for data below
15 degrees za (no spill over and no gain loss).
The coefficients from the sefd fit (with freq in Ghz) are:
sefd(fGhz)= 45.800962 -77.522389*f + 46.350673*f^2 - 9.2439278*f^3