July 13, 2012

Details

  • Location: NTC-200
  • Attendees: Morgan, Todd, Saini

Minutes

1) Question about whether the attenuators are being set correctly during measurement.

It seems that the attenuators might not be getting set correctly for the cross-pol 180 degree scan (aka the "third scan".)   However that cross-pol data is not used for any calculation.    MM will investigate next time beam scans are being done.

2) Why does the beam pattern analysis software run the beameff calculation program two times?

It runs once to calculate fits and efficiencies for the pol0 and pol1 co/cross scans.
It runs a second time to calculate for the 180-degree "third scan"

No action required, though it would be desireable to have all the logic related to the third scan embedded within the beameff C program rather than partly in C and partly in PHP.   MM will investigate refactoring this code.  No immediate changes needed.

3)  How to properly use Richard Hills' spreadsheet?

We did not get to this today.   We feel it is not as critical as it once was because we have Todd's independent analysis program to fall back on instead of the spreadsheet.   Still it would be good to get the SS working as well.

Actions:
TH will share his analysis code with us and will provide a quick-and-dirty user manual -- perhaps just a terminal session with comments.

4)  Should we be applying OEWG pattern correction to our data during FF processing?

Todd looked through old meeting minutes and emails.   Found facts:   NSI probe offset file is binary and proprietary.  They quoted $10K to translate pattern files into the target format.  (This was back when we were truing to use standard gain horns rather than OEWG.)   Decision was made at that time to try OEWG and try out the probe correction.   It seems that trying out the probe correction was never done.   The correction was expected to be < 4% in voltage (and so even less in power.)

Action:   MM, KS, JE to try processing a data set with and without probe correction for band 3 and band 9 or 10.    See how much difference it makes.
MM to decide how to implement selecting the correct probe model in SW.   It is not saved as part of the NSI template file and so must be selected by the operator or by NSI script.

5)  Problems with analysis, all bands in chamber 2:

We discovered that for all but band 10, if the analysis sequence is run twice, the 2nd result is in spec and matches Todd's independent results well.   So we have a software bug where somehow the state of files on disk or in the database is not being set up correctly the first time analysis is performed.     Action:   MM will capture state from file system and database before and after analysis is run once and then twice.

6)  Problems with band 10 in both chambers:

The phase-fitting algorithm is not quite right for any band, but particularly falls apart for band 10.    We tried a number of temporary fixes to the code and decided to try the following, which will take some careful work by MM on the code:

Start with an initial guess for the phase center of {0, 0, 0}.   Fit the phase with a mask limiting the search area to 1 degree radius, rather than the full subreflector radius of 3.6 degrees.    Then fit the phase again with a mask covering the whole subreflector.    (How we do it now starts with an initial guess other than {0,0,0} and for which the x and y terms are angles when they should be lengths.  Then only does the fit once.)

Additional action:   MM will share code with TH for review while improvements are made to make sure that the calculations and accompanying comments are correct.

I think that covers everything we looked at.   The software, as ever, needs some care and cleaning.  I feel a lot more confident doing so than I did before today.

-MM 
-- ToddHunter - 2012-07-23
Topic revision: r1 - 2012-07-23, ToddHunter
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback