apr08 pol B noisy
On 22apr08 the stability of lbw was tested using the
high correlated cal. The setup/processing was:
Plots were then made showing the total
power (.ps) (.pdf)Page1
Fig:1,2: total power with cal on for polA (1st) and polB (2nd)
plots. The vertical scale is in correlator units. The two lines per
plot are the two different sbc measured (with the same input signal).
There are gain differences between the different plots.
- Telescope sitting at az=255, za=13
- center freq=1430 MHz, 1100-1800 filter
- interim correlator with 9 levels,25 MHz, and 2 boards with the
same data fed to each.
- take 270 1 second samples. This gave 135 (calon-caloff)
- Toggle the cal once a second using the hardware 1 second toggle.
- Using corblauto fit a 6th order harmonic and 1st order
polynomial to (corAvgOn/corAvgOff). This removes the bandpass.
- Throw out all points with fit residuals above 3 sigma. Use
the remaining points to compute the total power.
- For each record compute the total power over the computed mask.
- Page1 Fig:3,4: total power with cal off for PolA (3rd) and polB
- Page2 : cal Deflection in Kelvins. top pola, bottom polB. I've
used the old cal values but they should be pretty close. The two lines
on each plot are from the 2 sbc (that had identical signals input).
- Pol B is less stable than polA. The measure cal Deflections were:
(1 sigma errors)
- polA 8.7 +/- .02 K
- polB 11.2 +/- .1 K
- The expected error is: deltT=1/sqrt(22e6*1) * 32K = .007K. There
is baseline variation from gain changes cause by sky/temperature. So
.02 K is not unreasonable. But .1 K is pretty large.
- Looking at the first page, the power in Tsys +cal Off is also
noisier in polB then in polA so the problem is not the cal or cal