LRO MiniRF Polarization calibration session Aug 30, 2009

project code: AGBT09A_070_04

• Plan: do spider scans on 3C286 from transit -2hrs to +2hrs to calibrate the polarization properties of the feed.

• Runs 17:15 to 22:15 UT (13:15 to 18:15 EDT)
• Thats 10:31 to 15:32 LST
• 3C286 RA=13:31

• spider_Aug.config: Proposed configuration for spider scans
• spider_Aug09.sb: Planned scheduling block for running spider scans

• miniRF_LROspider.py: Config file actually used for spider setup
• spider_Aug09_used.sb: Scheduling block actually used for spider scans

• Amanda's report and plots of Mueller matrix solutions

• Discussion of TCALS

• Bryan_reduction.txt: Bryan's discussion of the calibration method (Sept 21)

---

Observing Notes
Scans 2-5 peak up on 3C286  LPC=0.084',0.203',  TA=21.6K
Scans 6-7 ONOFF on VirgoA, balancing on source SPinput levs about -6db, attens=25,34.
           offsource, SPinput levs about -14,-16 db
Scans 8-9 another ONOFF on VirgoA, but this time set "standard" attens:
              SP attens to 40 and 44 (input levs onsource -38 and -31)
              IFRack 16 and 16db
              ConverterRack 12 and 14.
Scans 10-13 return to 3C286, peak up again. LPCs=-0.012, 0.288'
      14-15 OnOff on 3C286, balance on source.

17:57 UT (11:14LST) starting spiders - delay due to - problems with SP balancing, somehow wrong frequency in SP
      revised configuration to base on C1Config with changes for spiders.

18:20 UT (11:37 LST) - spiders finally started ok.
          tuning to 2380; IFrack attens = 0,0; conv rack attens 12 and 14
          SP bandwidth 20MHz; SP input levs -6 to -8 db
  - repeating spiders ad inf

20:20 UT (13:33 LST) SP hung in stopping. Dave had to reboot it.  JUst at the critical transit moment.
   Restart spiders.  SP now balanced near -11 instead of -6.
   - another spider sequence.
   - stop and try to re-balance for -6 -- ok (SP attens are 26 and 34)
more spiders, starting scan 137

22:14UT (15:32 LST) all done.  

• AGBT09A_070_04.summary: Scan summary for session 04
• AGBT09A_070_04_log.txt: Astrid Logs for session 04

---

Just for amusement, here is a plot of I=0.5*(XX+YY) and XX-YY versus parallactic angle.
I presume the minimum in XX-YY near 30 deg has something to do with the position angle of the linear pol for 3C286 being about 33 deg.
These points are derived by taking the spectrum at the middle integration of each cross-scan, i.e., presumably the one positioned on the source. Each spectrum is reduced to a single number by averaging over the spectrum and scaling to Kelvins.

• XX+YY and XX-YY vs parallactic angle:
  

---

Processing and derivation of Mueller matrix.

Amanda Kepley has processed these data to solve for the Mueller matrix. Her e-mail reporting on this:

• amanda1mail: Amanda's report on solving for the Mueller matrix
• amanda2mail: Mail with Carl's response

All 4 of her plots as postscript files are attached below.

• Plot of Mueller matrix solutions: 1-D with new cals.
  

---

Discussion of TCALS

Amanda noted in her analysis that in order to equalize the gains in the X and Y channels it was necessary to change the Tcals from what was in the TCal database. The TCals can be checked by observations of astronomical calibrators. Taking a spectrum of a calibrator, assuming you know its flux density, the telescope's aperture efficiency, and the atmospheric opacity, one can derive the TCals.

We processed 2 datasets:

1. the two OnOff spectra on the source 3C147 taken during the LRO session of August 23.
2. spectra of 3C295 during an SCAL run September 5.

These were analyzed using Ron's SCAL idl routines and my scripts that use them. (see /home/gbtidl/release/gbtidl/contrib/scal*, and /users/fghigo/ptcs/src/scal* and doscals_1beam)

The following are the results for the high cals in linear polarization X and Y.

	Table 1
	TCal(X) Kelvin	TCal(Y) Kelvin	source
1.	17.743	9.154	from Tcal database, data of Aug 1, 2008, value at 2380 MHz
2.	18.494	8.405	needed to equalize X and Y gains in polarization solution
3.	18.166	9.028	derived from spectra of 3C147, Aug 23, 09 observations, average over 1.2MHz
4.	20.552	8.573	derived from 3C295 spectra taken Sept 5, 09, average over 20 MHz

Flux densities of sources at 2380 MHz, based on VLA calibration data (ref: http://www.vla.nrao.edu/astro/calib/manual/baars.html)

Source	Flux Density (JY) at 2380 MHz
3C295	13.798
3C147	14.707

The values in lines 3 and 4 of the Table 1 (above) are derived on the assumption that the aperture efficiency is 71% and the zenith opacity is zero. Typical zenith opacities at S-band are usually of the order of 0.01 or less, so would make less than a 1% change in the derived TCals. Fiddling with the efficiency, the opacity, or the calibrator flux density can make a scale change in the derived TCals, but both X and Y TCals would change by the same factor. For example, if we take the values in line 3 of the table and increase the X value (18.166) to match Amanda's value in line 2 (18.494), i.e., a factor of 1.0806, then we have to increase the Y value (9.028) by the same factor, resulting in 9.191 which disagrees with the value in line 2 by about 10%. Using the data in line 4 leads to a worse disagreement.

The numbers in line 3 come from observations with the Spectral Processor with a 1.25 MHz bandwidth, i.e, the same setup used for the miniRF observations done Aug 22 and 23, but not exactly the same as the polarization cal observations of Aug 30 which were done with a bandwidth of 20 MHz. The results shown in line 4 of the table are based on spectra taken with the ACS with 800 MHz bandwidths.

Although one would like to think that it should not matter which spectrometer or which bandwidth one uses to derive TCals, none of these results agree with each other to better than 5-10%. So what is happening? The high cals, 10-20K, can increase the system temp by 50-100%, a problem especially for the ACS spectrometer in 800 MHz (3-level) mode which has a narrow dynamic range. Also the satellite RFI around 2330 MHz is very strong, and despite the notch filter may be affecting the spectra. It does not seem that we can derive a reliable improved value for the high cals.

In any case I would guess that the values Amanda derived for the polarization cal are likely to be the best values to use for analyzing polarizations.

This plot shows the new TCals vs frequency, derived from Aug 23 data on 3C147, using the SP: TCals from these solutions are listed in line 3 of Table 1. Solid lines are the derived TCals vs frequency; the pluses are from the TCal database, as listed in line 1 of Table 1. Diamonds are the points from line 2, i.e, TCals required to equalize gains in the pol solution.

This plot shows the new TCals vs frequency near 2380 MHz derived from the data of Sept 5, '09, using the ACS: Solid lines are the derived TCals; the values at 2380 are tabulated in line 4 of Table 1. The plusses show values from the TCal database, listed in line 1 of Table 1. The Diamonds are the points from line 2 of Table 1.

---

---

-- FrankGhigo - 27 Aug 2009