1.2 hz oscillation in lbw gain (something loose in the dewar)

may 05

Links to plots:
16apr05: jumps and oscillation in the a2006 total power data (.ps) (.pdf)
17apr05: tapping the dewar causes the gain to change (.ps) (.pdf)
10apr05: the periodicity's in the time series for 53 data files taken in early apr05 (.ps) (.pdf)

Jumps, 1.2 hz oscillation seen 16,17may05.

Project A2013 looked at a flare star with lbw. The setup was:

linear polarization's
100 Mhz bands centered at 1170,1370,1470,1570 Mhz with 128 lags
10 millisecond dump time, full stokes.

When looking at dynamic spectra and total power versus time, the power level was unstable. There was a 1.2 hz periodicity as well as discreet jumps. When the polB power level jumped, the polA level would always jump in the opposite direction. The plots show the jumps and oscillation in the a2006 total power data (.ps) (.pdf):

Fig 1 top: PolA total power versus time.
Fig 1 2nd: PolB total power versus time.

The 3 colors are the 1170,1370, and 1470 Mhz bands. The power jumps are at 70 seconds and again at 270 seconds. The dashed lines are blown up in the 3rd and 4th plots. The polB jumps up to 8% of Tsys. Pol A is a little less.

Fig 1 3rd: PolA blowup in time.
Fig 1 4th: PolB blowup in time.

In the blown up data, you can see the 1.2 Hz oscillation. This is coming from the dewar crosshead.

Fig 2 top: Periodicity's in the polA time series.
Fig 2 Bot: Periodicity's in the polB time series.

The power time series was transformed and then the magnitude was taken. The 3 colors are the 1170 (black), 1370 (red), and 1470 (green) 100 Mhz bands. The spectral resolution is 1.6 millihz. The strongest period is 1.2 hz spikes. It is strongest in polB where it gets to .28% of Tsys. (the red comb is the 12/2 sec rotation period of the faa radar).

The dewar crosshead has a 1.2 hz period. On 17may05 we did some tests:

Turning off the crosshead caused the 1.2 hz ripple to go away.
Tapping the lbw dewar caused jumps in the gain. Tapping the sbw dewar made no difference.
The postamp chassis are mounted with the dewar. To see if the variation was in the postamp chassis we did:

The polA input to the postamp chassis was switched to a 30db amp.
The dewar was tapped again. PolB jumped but polA was rock solid. This says the problem is in the dewar.

continuum data was taken (5 Mhz centered at 1420 Mhz with a 1ms time constant) for 120 seconds. During this period we tapped the lband wide dewar. The plots show how tapping the dewar causes the gain to change (.ps) (.pdf):

Fig 1 top: This shows the total power versus time. Black is polA while red is polB. We started tapping the dewar around 62 seconds and continued tapping for about 10 seconds. The power level for polB jumped up by 5% while the power level for polA decreased by about 2%. The power levels did not return to their previous values until 30 seconds after the tapping stopped.
Fig 1 bottom: This is the magnitude(fft(tp)) for the 120 seconds. It shows the 1.2 hz periodicity of the crosshead.
Fig 2 Top: PolA mag(fft(tp)) taken in 20 second sections.
Fig 2 bottom: PolB mag(fft(tp)) taken in 20 second sections.

These plots show how the periodicity varied during the 120 seconds. Each strip comes from 20 seconds of data. The 1.2 Hz oscillation is present in polB from the start. The 4th strip (blue) started 60 seconds from the start when the tapping commenced. The 1.2 hz oscillation increased because of the tapping.

The 1.2 hz oscillation was present but weak in early april05

Project A2013 looked at a flare star on 08apr05 thru 10apr05. 4 100 Mhz bands were taken centered at 1170,1370,1470, and 1570 Mhz with 128 lags and a 1 millisecond dump time. I looked to see if the 1.2 hz oscillation is present in this data by:

Compute the total power versus time for the 1470 Mhz band using the first 300 seconds of each file. Do this for 53 data files spanning the 3 days.
Compute the magnitude(fft(timeseries)) to see the periodicity's in the data.

The plot shows the periodicity's in the time series for 53 data files taken in early apr05 (.ps) (.pdf) :

Top: PolA
Bottom: PolB

Two of the strips have power at the 1.2 Hz crosshead oscillation frequency. It is stronger in polB than in polA (as is the may data). The amplitude of the oscillation is a factor of 10 less than the 16may05 data. The occurrence of the oscillation is also much less in early april05.

Summary:

There is something loose in the lbw dewar that causes the gain (or attenuation) to vary.
PolA and polB gain change 180 degrees out of phase with each other.
Tapping the dewar causes the gain to jump. It is then more susceptible to the crosshead vibration. The tapping must jar something loose and then the crosshead continues to shake it.
Normal motion/vibration of the dewar by telescope motion is probably causing the gain jumps seen in a2006 of up to 8% of Tsys.
The 1.2 hz oscillation was present in early april05 but it was 10db weaker and occurred much less frequently.

processing:x101/050517/osclbw_17may05.pro, usr/a2006/chkosc.pro

 home_~phil