| Version 5 (modified by , 15 years ago) (diff) |
|---|
TestCaseData/2009-10-09_bandpass_interpolation
Description
Bjorn Emonts did some tests of how the new code that allows time-variable bandpass tables (ticket:90) improves things.
Method: I observed a 2 Jy calibrator roughly every 20 minutes at 3cm.
I divided the data in two parts (even and odd scans) and used one data-set to calibrate the other. Basically, I thus got a bandpass calibration roughly every 40 min and observed the 'target' in between the calibrator scans. The continuum in the resulting image is subtracted, so you only see the residuals.
The 'entire' data-set used only 1 bandpass solution interval to calibrate the data (i.e. the old way). The spectral dynamic range that I get is only 1:100 (quite poor).
The 'interpolate' data-set interpolated between the bandpass calibrator scans. It is clear that the spectral dynamic range improved significantly!
The plot is too large to attach. Added to the operations subversion tree.
Data Set(s)
- 2009-04-12_2357.CX171 (ifsel=2). Frequency range is 7.976-10.025 GHz.
Tickets
- ticket:90 - Implement time variable bandpass solutions
