Oscillations from dewar monitoring.
Beam 2a,2b of alfa has looked noisier that the
other beams for about a year. These two beams had been switched to an
open loop bias card back some time in dec09. On 15feb11 it was found
that the total power variation of these beams had a periodicity of
about 5.22 seconds. This was later traced to the monitoring of the bias
voltages and currents that occurs approximately every 5 seconds.
To get a better idea of the problem and when it
occurred, data was taken:
- 08,14feb11,16feb11 with
spectrometers at 10hz. The telescope
was sitting still
- 08feb11: 172 Mhz bw centered at 1470 Mhz
- 15feb11: 172 Mhz bw centered at 1450 with 100 Mhz rf filter
installed to get rid of rfi
- 16feb11:172 Mhb bw centered at 1450 Mhz.
- open loop card now in beam 1
- also take data with dewar monitoring turned off.
- a2010 alfalfa
data from the last year was used to get an
idea when the problem started.
Mock data at 10hz.
The data was taken with the telescope sitting still.
10 hz sampling was used with a 172 MHz bandwidth. Only
the upper band (that did not include the radars) was used.
- 10hz mock data
cfr=1470 MHz (.ps) (.pdf)
- Page 1: total power vs time for 900 seconds.
- top polA, bottom, polB. Each colored trace is a different
- Page 2: blowup of total power vs time (100 seconds).
- The vertical deflection of the periodicity gets up to 1% of
- The duration of the periodicity is 5.22 secs. The duration
it is high is about 1 second.
- Page 3: spectra of the total power. top polA, bottom polB
- You can see the .1915 hz periodicity in polA and polB (green
- To make sure that the problem wasn't related to a radar in
band the setup used was:
- 100 MHz rfi filter (1390-1490) installed. This filtered out
the radars (after the dewar).
- 10 hz sampling, 172 MHz bw centered at 1450 MHz.
- Pages 1-3 have the same details as the 08feb11 data.
- You can see the oscillations clearly in beam2a,b
- for some reason beam3a had a large variation in gain (trouble
with the filter switch??). It later stabilized.
- spectra vs time for 70 seconds of data
- You can see the 5.2 second periodicity in the dynamic spectra.
- The oscillation does not show much frequency dependence.
open loop bias card has been moved to beam 1
with 1 second sampling.
A2010 data (alfalfa) was looked at to see
when the problem occurred.
- Using the wapps, a 100 MHz bandwidth centered at 1380 MHz is
sampled for 600 seconds at 1 hz sampling.
- The telescope is stationary.
- Page 1 has 4 frames each showing a 600 second total power data
- The beams are color coded.
- PolA and polB are over plotted.
- the vertical scale is degrees K (offsets have been added for
- Page2 has the spectra of the total power from page 1.
- the Nyquist frequency is .5 hz.
- Vertical dotted lines have been placed at multiples of .1915 hz
- A 5.22 sec periodicity was seen in beam 2 of alfa.
- If was not present on 16dec09. It was present on 11jan10. So it
appeared somewhere between these two dates.
- the amplitude of the oscillation is up to 1% of Tsys (typically
.25 to .5 % of Tsys).
- Turning the dewar bias monitoring off caused the oscillation to
- The dewar monitoring reads all of the info from the bias box,
waits 4 seconds, and then repeats the procedure.
- it must take about 1 second to read the information. This is
when the gain increases on the measured total power.
- When beam2 open loop bias card was moved to beam 1, the 5.2
second oscillation moved with the open loop card to beam 1.
- Changing the open loop bias card back to close loop will decrease