A2471: looking at bursts from dwarfs

jan10

Analyzing the data
Debugging changes in stokes I

Intro:

A2471 used the mock spectrometers to monitor bursts from dwarfs.The emission should be 100 % circularly polarized. The data taking setup was:

use the mock spectrometer with 172 MHz bw/band, 8192 channels/band, 7 bands, and full stokes. The data was sampled at .1 second intervals.
600 second on source were followed by a 10 sec calon,caloff (still on source).

The data processing was:

Use masstokes to input and calibrate the on scan and cals.

Bandpass correction was done with the cal off (which is actually still on source).

The data is then converted to Jy using the gain curves.
The rms along each channel is computed using stokes I.

A mask is created excluding channels with large rms.
The power is then computed for stokes i,q,u,v for each of the 7 bands using the rms mask.

Plotting the data:
Plots were made for each source on each day.

total power plots

IQUV, vs data sample and frequency band.

The horizontal axis is data sample (1 second). I have left out the gaps between each 10 minute dataset. The vertical axis is milliJy.
There are 7 frequency bands with 4 per page.
The data is plotted I:page1,2 Q:page3,4,U:page5,6, V:page7,8.
Red vertical lines are drawn at the start of each 10 minute dataset.
A second order polynomial has been removed from each 10 min set of data.

This will make in hard to see bursts that last for > a few minutes.

I data

The median value for I has been removed.
The variation is the za dependence of the gain (the SEFD is higher at large za).

Processing the pulsar data:

On 30dec09 data was taken on pulsar B2110+27 at cband.
600 secs of data was taken at 20.0047619 millisecs sampling
the total power for each of the 7 172 MHz bands was computed and then folded at the pulsar period.

Notes:
I'm using positive and negative burst rather the left and right circular. We need to use the helix in the bowl to see which one comes out where in the data processing.

30dec09

IQUV total power data

J2101+17 (.ps) (.pdf) (only 10 minutes)

Fold 600 secs of pulsar B2110+27 at psr period (.ps) (.pdf)

sampled data at 0.020047619 secs
used 7 172MHz bands.
No rfi removal attempted.

06jan10(AST)

IQUV total power data

plot showing blowup of stokes V burst (.ps) (.pdf)

07jan10(AST)

IQUV total power data

HD102195 (.ps) (.pdf)
HD106252 (.ps) (.pdf)

stokes I variation (baseline?) Q=0

HD114762 (.ps) (.pdf) (only 10 minutes)

08jan10(AST)

IQUV total power data

09jan10(AST)

IQUV total power data

30nov10 (AST)

IQUV total power data

J10475385 (.ps) (.pdf)

I blew up the vertical scale a little since the baselines were pretty flat.

01dec10 (AST)

IQUV total power data

J10475385 (.ps) (.pdf)

Checking out some of the signals in the plots:

pnt 2250 I

band 1 positive spike. this is the 4250-4350 rfi.
band 2-8 negative spike. Looks like the 4250-4350 pushed the system into compression so all other bands went negative.
Any signals in other stokes params should be ignored during this period

pnt 3600 I

band 1 positive, bands 2-8 negative. Again this was the 4250-4350 rfi compressing the system
Any signals in other stokes params should be ignored during this period

pnts 5300 to 6300 I

Remember that the first plot has a 2nd order polynomial removed from the 10 minute 10 series.
band 1: There is a little 4250-4350 rfi in the first band so it should be suspect. The rfi is not strong enough to compress the system.
band 2-8 I: the variation is in all bands
Where could it come from:

checking the pointing, the encoders output gave a sky variation of about 1asec so we were looking at the same spot.
The dynamic spectra showed that the power was pretty uniform across each band.
Stokes Q has nothing (except band 1 where there was interference)

An electronic gain variation in the system could have caused the variation in I.
You would expect an electronic gain variation to be different in polA and polB since they are different electronics. This would have caused power in stokes Q which is not present. So i think you can rule out electronic gain variations.

Stokes u,v have no change.. So the change is completely unpolarized.
The tiedowns pulling or releasing the platform could have changed the optical gain. The distomats showed the average height of the platform to be stable during this time (this would have also required the source to have some flux to start with).
I checked the dewar temperature for cband (measured every 30 seconds). It showed no variation during this period. We could also look at the 2 cal on,offs take during this period and see if they give the same value for the cal deflection.

This data was taken around 7 am so we need to worry a little about the sun coming in the sidelobes.

J10475385 Stokes I flare (.ps) (.pdf)

The stokes I data for scans 8,9,10 (sample points 4669 to 6670) were used to display the stokes I flare.
Page 1: data (black) and fit (red) plotted vs time.

The y axis is in Jy. This is mainly from the system temperature.
The x axis is time (resolution .9 seconds).
The fit (red) is a 2nd order fit using the edges of the data were there is no flaring.

The fit was done in za rather than time.

the 4608 fit didn't work as well as the others.

Page 2: data - fit

top. The Stokes I flare is plotted vs time.

plots are offset by .05 Jy for display.

bottom: the I flare is overplotted.
The flare peaked at about .15 Jy.

processing: a2471/jan10/doit101201_flareI.pro

02dec10 (AST)

IQUV total power data

J10475385 (.ps) (.pdf)

The glitches:

the stokes I glitches are from polB only. polA was stable
A power failure the previous nite had caused the dewar to partially warm up. It was cooled back down by about 1am. This observing started at 5:30 am.
The vacuum pump was connected during this observing run. It was disconnected about 8:30 am.
at 16:00 on 2dec10 i took some data and saw no glitches in polB

polA,QUV total power data.

For these plots i replaced stokesI with 2*polA so we could see if there was any flaring (at least in 1 pol)

none seen.

J10475385 2*polA (.ps) (.pdf)

03dec10 (AST)

IQUV total power data

J10475385 (.ps) (.pdf)

04dec10 (AST)

IQUV total power data

J10475385 (.ps) (.pdf)

05dec10 (AST)

IQUV total power data

J10475385 (.ps) (.pdf)
Plot showing blowup of stokes V burst (.ps) (.pdf)

Page 1: Stokes V vs time for the 2nd and 3rd scans (sample points 667-2001)

Black is the data, red is the
The fit used a 2nd order polynomial vs za. It excluded points around the burst.
The gap at 667 seconds is the 10 sec calon, caloff.

Page 2: Stokes V (Data - fit) vs time.

The x Axis is seconds since Jd 2455535.899120 with .9 second resolution.
The dashed purple lines are near 4325 MHz peaks @ 706.3 and 756.0 seconds from Jd start.
You can see the right most peak move earlier in time as you go to higher frequency.

processing:a2471/jan10/doit101205_pulse.pro

06dec10 (AST)

IQUV total power data

J10475385 (.ps) (.pdf)

07dec10 (AST)

IQUV total power data

J10475385 (.ps) (.pdf)
Stokes I has some activity in the 6th file (around sample 3335). Not obvious if this is real or not.

17dec10 (AST)

IQUV total power data

HD50554 (.ps) (.pdf)

Stokes I burst, q=0

HR8799 (.ps) (.pdf)

stokes I burst,Q small

J00325937 (.ps) (.pdf)
J01514155 (.ps) (.pdf)
J03284265 (.ps) (.pdf)

Stokes I burst, q=0

J10475385 (.ps) (.pdf)

Stokes I burst, q=0

epsTau (.ps) (.pdf)

18dec10 (AST)

IQUV total power data

Stokes I burst, q=0

J0039191 (.ps) (.pdf)
J01514155 (.ps) (.pdf)

20dec10 (AST)

IQUV total power data

02jan11 (AST)

IQUV total power data

55Cnc (.ps) (.pdf)

stokes I burst, q=0

HD114762 (.ps) (.pdf)

stokes I burst, Q =0 1 scan, q!=0 2nd.

J07271824 (.ps) (.pdf)
J10475385 (.ps) (.pdf)

stokes I burst, Q=0

03jan11 (AST)

IQUV total power data

J07271824 (.ps) (.pdf)
J09373487 (.ps) (.pdf)
J10475385 (.ps) (.pdf)

stokes I burst, Q=0

J12382851 (.ps) (.pdf)

stokes I burst, Q=0

11jan11 (AST)

IQUV total power data

J10475385 (.ps) (.pdf)

20jan11 (AST)

IQUV total power data

J10475385 (.ps) (.pdf)

07feb11 (AST)

IQUV total power data

J10475385 (.ps) (.pdf)

08feb11 (AST)

IQUV total power data

J10475385 (.ps) (.pdf)

15mar11 (AST)

IQUV total power data

J15232263 (.ps) (.pdf)
J16322911 (.ps) (.pdf)

check

J1711457 (.ps) (.pdf)

16mar11 (AST)

IQUV total power data.. total power has lots of variation..

J144601 (.ps) (.pdf)

check

J1711457 (.ps) (.pdf)

check

J16322911 (.ps) (.pdf)

date (AST start)	IQUV total power data for sources
110718	51Peg (.ps) (.pdf) J1841086 (.ps) (.pdf) J21011544 (.ps) (.pdf)
110719	51Peg (.ps) (.pdf) HD178911B (.ps) (.pdf) J1841086 (.ps) (.pdf) J21011544 (.ps) (.pdf)
110906	51Peg (.ps) (.pdf) HD189733 (.ps) (.pdf) HD209458 (.ps) (.pdf)
111010	J01514155 (.ps) (.pdf) J03454316 (.ps) (.pdf)
111020	J0326137 (.ps) (.pdf) J03454316 (.ps) (.pdf)
111021	J03284265 (.ps) (.pdf) J03454316 (.ps) (.pdf)
111022	J0326137 (.ps) (.pdf) J03454316 (.ps) (.pdf)
111027	J0039191 (.ps) (.pdf) J01514155 (.ps) (.pdf) J03284265 (.ps) (.pdf)
111105	J0036159 (.ps) (.pdf)
111119	J02074284 (.ps) (.pdf) J0326137 (.ps) (.pdf) J07003664 (.ps) (.pdf) J07271824 (.ps) (.pdf)
120207	J07464256 (.ps) (.pdf) J09373487 (.ps) (.pdf) J11463449 (.ps) (.pdf) J14243909 (.ps) (.pdf)
120209	J07003664 (.ps) (.pdf) J08251958 (.ps) (.pdf) J0850359 (.ps) (.pdf) J09121469 (.ps) (.pdf)
120306	J07464256 (.ps) (.pdf)
120309	J16241436 (.ps) (.pdf) J16322911 (.ps) (.pdf)
120310	J1711457 (.ps) (.pdf) TVLM-513 (.ps) (.pdf)
120311	J1711457 (.ps) (.pdf) TVLM-513 (.ps) (.pdf)
120329	J16322911 (.ps) (.pdf) TVLM-513 (.ps) (.pdf)
120330	J1711457 (.ps) (.pdf) TVLM-513 (.ps) (.pdf)
120331	J16322911 (.ps) (.pdf) TVLM-513 (.ps) (.pdf)
120405	J07464256 (.ps) (.pdf)
120406	J07464256 (.ps) (.pdf)
120407	J07464256 (.ps) (.pdf)
120429	J07464256 (.ps) (.pdf)

Notes:

Blowup of Stokes V burst:

The noise is larger on the 4325 band than the other bands. This may be from the Tsy being larger, or the cal value might be a little off.
Arrival time at 4325 for peak: 2455202.855182 JD.

This was measured with a .9 second resolution.

Stokes I burst. i've labeled some of the runs that had bursts in I over all freq bands. Usually q=0. This is not necessarily coming from the source. It could be receiver based, or sidelobes.

processing: usr/a2471/jan10/doitall.pro

Debugging total power variation. (top)

Some days, stokes I has shown variations in all 7 bands. To examine this, the data for J1711457 from 15mar11 and 16mar11 was examined:
Total power for J1711457 from 15-16mar11 (.ps) (.pdf)

Processing:

Each of the 7 frequency bands was normalized to the median value for the day and then 1. was subtracted.
Each frequency band is plotted with a different color.
vertical dashed lines show the start of each scan.

Page 1 : 16mar11 Stokes I, all bands

top: Over plot stokes I vs time (minutes) from the start of run.
middle: Stokes I vs za
bottom: Stokes I vs az

Page 2: 15mar11 stokes I, all bands

Same as 16mar11

Page 3: Over plot average of 7 bands vs azimuth for 15,16th mar11.

Summary:

The total power is varying by up to 6% of Tsys (3.2Jy*.06=192mJy).
Time periods for the changes vary from 1 to 6 minutes.
The changes track the same in the 7 freq bands.
The changes to not repeat day to day when plotted vs azimuth.
Variation were also seen with the telescope sitting still and absorber covering the horn

To do:
track blank sky using the winking cal. This will tell whether the changes are from Gain or Noise.

processing: usr/a2471/jan10/debug_15_16mar11.pro

home_~phil