Open bennahugo opened 4 years ago
Here is delaycal+gaincal at 30s intervals (B transferred from another calibrator): after solving the residuals at 4s intervals with CASA K+G solutions the following -30dB (abs peak positive/abs peak negative) map is obtained
If you are feeling brave, you can give the issue-326-jsk branch a shot @bennahugo . It is still a WIP but should help in the case of bad delay errors. Produces initial guesses by taking the FFT of the data along the frequency axis.
I've done a calibration test against CASA 4.7 (note 4.7 not 5 - 5's delaycal is also broken) at native resolution of 4s. The cubical gains diverge whereas the 4.7 solutions substantially improves the image. My options are pretty much box standard. The chi^2 diverges in a lot of tiles, however there is ample SNR to solve for the delays: