Anonymous | Login | 19-09-2024 01:24 UTC |
All Projects | CNI | DZB | equalizer | IVC/ADC | LFFE | MFFE | Puma | Puma 2 | RFI | RT | submission | system time | Tadu | Time/LO | TMS | VLBI |
Main | My View | View Issues | Change Log | Docs |
Viewing Issue Simple Details [ Jump to Notes ] | [ Issue History ] [ Print ] | ||||||||
ID | Category | Severity | Reproducibility | Date Submitted | Last Update | ||||
0000329 | [DZB] 0.unknown/unspecified | major | N/A | 27-01-11 18:36 | 23-02-11 18:57 | ||||
Reporter | Gyula Jozsa | View Status | public | ||||||
Assigned To | |||||||||
Priority | high | Resolution | open | ||||||
Status | resolved | ||||||||
Summary | 0000329: Baseline dependent gain errors (closure errors) | ||||||||
Description |
Report by T. Oosterloo (Ger and first Hans v.S. had already been looking into this): Hans van S has recently discovered that since a few months there are very large closure errors in WSRT data. Although there never is a reason to question Hans's judgement, I checked some recent data. attached are 2 plots from a line obs done on 2-Jan-11. This field has a strong (1Jy) pointlike source. One image shows the image after several standard selfcal interations (including amplitude in the last few). Rings are still visible which do not disappear, whatever i try in the selfcal. the rings are at the level of 0.1% or so (dyn range Note: 0000300). The other image is after an additional baseline dependent (but time independent) correction (of up to 4%!!!!!). Rings are gone, image is noise limited (dyn range > 3000) Ger added: It indeed appears that the closure errors are fairly stable, and are therefore correctable. However I checked the closure errors on the pair of calibrators of yesterdays Abell 773 cluster observation (3C48 = 11100453 and 3C286 = 11100456) and it is clear that their closure errors still differ at the 0.5% level. This means that a transfer of closure errors will not be producing ideal data. This level of 0.5% is still much higher that what WSRT used to deliver (<0.02% typically in amplitude and no phase closure errors). My last WSRT 8x20, 64x4 WSRT continuum observation was done in late September (3C286 11005333) and was Ok in the sense that the closure errors were at the 0.1% level, a level determined by noise in the 15m observation. So if your email meant to imply that we have to accept this I strongly disagree. With possibly 2 more years to go with the IVC/DZB and a range of 92cm legacy proposal being prepared, I find this very worrying. |
||||||||
Additional Information |
Comment by Josh: Toms report implies baseline dependent gain errors which are, however, constant in time. The amplitude gain is wrong by up to 4% PER BASELINE. The fact that this is a constant calibration error per baseline may be a hint where this is coming from (backend, software?). Tom describes how to solve the problem after it has occurred. However, the method implies the need for a nice calibration source in the observed field that may not be available all the time on top of some experience. On top of that, Ger reports that the time independence is not valid at a 0.5%-level, which was what George found, experimenting with the long-standing problem with (mostly) RT3 (see below) Baseline errors occur since a long time, but not at many baselines, only on singular ones (reported in ISSUE 220). However, both phenomena may be related, since it seems that also the baseline amplitude error reported in issue 220 has vanished in Toms experiment (meaning that the error had been constant in time). |
||||||||
nr subbands | 1 | ||||||||
nr polarisations | 2 | ||||||||
nr channels | 1024 | ||||||||
bandwidth | 10 MHz | ||||||||
multi-subarray | no | ||||||||
recirculation | no | ||||||||
Attached Files |
with.pdf [^] (315,531 bytes) 27-01-11 18:36 without.pdf [^] (305,330 bytes) 27-01-11 18:37 |
||||||||
|
Relationships | ||||||
|