Status of EVN Amplitude Calibration --------------------------------------------------- The following table shows the median absolute amplitude error for the EVN stations in the first two sessions of 2009 (March & June). These results were derived from the pipeline amplitude self-calibration results of all EVN experiments which have been pipelined to date: 5 Nov 2009. The number in brackets after each entry is the number of experiments that were used to determine the median error for that entry. Session 1/2009 -------------------------------------------------- Station 18 cm 6 cm 5 cm -------------------------------------------------- Cm 0.150(5)* 0.212(6)* 0.078(10) Ef 0.018(5) 0.049(7) 0.038(10) Jb 0.148(5)* 0.119(6)* 0.049(9) Mc 0.077(5) 0.061(7) 0.039(10) Nt 0.079(5) 0.136(7)* 0.073(10) On 0.038(5) 0.078(6) 0.054(10) Sh 0.064(2) 0.053(4) Tr 0.068(5) 0.031(7) 0.036(10) Ur 0.078(3) 0.067(1) Wb 0.023(5) 0.021(7) 0.073(10) Mh Ar 0.031(1) Ro 0.026(1) Ys 0.055(3) Hh -------------------------------------------------- Session 2/2009 -------------------------------------------------- Station 18 cm 6 cm 5 cm -------------------------------------------------- Cm * * Ef 0.033(10) 0.046(8) Jb 0.109(10)* 0.123(8)* Mc 0.091(10) 0.040(6) Nt 0.079(10) 0.066(8) On 0.078(10) 0.066(6) Sh 0.073(7) 0.243(6)* Tr 0.129(10)* 0.070(7) Ur 0.089(6) 0.115(8)* Wb 0.051(10) 0.042(8) Mh Ar 0.104(1) 0.078(2) Ro Ys 0.043(6) Hh -------------------------------------------------- The blank entries indicate insufficient data. The numbers above are the median absolute error in the antenna gain amplitude (as calculated from pipeline amplitude self calibration). A value above 0.1 indicates a significant error which should be investigated. In addition to the absolute errors summarised here, the EVN pipeline provides details on every experiment processed at JIVE including the sign and time variability of the errors. In each experiment, the self-calibration results of all the available bright sources were used to get the reliable results. *Cm/Kn: The 16 MHz signal was split into two subbands in the 16MHz/subband experiments. So the proper amplitude calibration factors were not available since these channels were not flagged out. *Jb: Correction factors varied from subband to subband. Larger error in LL 3&4 (BBC 3) at 18 and 6 cm in the Gbps experiments. In the two sessions, the rxg files were updated on 27 July 2009. The calibration experiments should be performed carefully during each session. A stabilised cal diode box will be installed on MK2. *Sh: Low correlation amplitude and clearly S-shape phase bandpass in IF 5-8 (BBC 5-8) in the 6cm Gbps experiments of Session 2/2009. *Ur: Low correlation amplitude in LL 7&8 (BBC 7) in the 6 cm Gbps experiments of Session 2/2009. *Tr: Large error in IF 7&8 (1691 & 1706 MHz) in session 2/2009, which was probably caused by RFI in the 18cm experiments. *Nt: Calibration factors varied from suband to suband at 6 cm in Session 1/2009. The reason was that the Tcal of the RCP was set too low. EVN station should insist in performing calibration experiments according to the EVN schedule and report if Tcal show significant variation and rxg files are updated on the feedback web page of CL experiments. Points of note: Session 1/2009 -- The station gains (On, Jb2, & Cm) at 5 cm were improved using the amplitude calibrators of N09M1. The total flux density measured by EF during the EVN sessions were also provided on the web (http://www.mpifr-bonn.mpg.de/div/effelsberg/EVN_cal/). The WB data in the IDI-Fits format were also available for the PI to calibrate the data. The FS procedure CALTSYS can be optimised by reversing the order of bbcagc and caloff. The program antabfs.pl was modified to deal with the FS log files of MK5B stations (e.g. Yebes). Session 2/2009 -- Nt corrected the rxg files at 6 cm; the rad files provided by CM & Kn were used to create the antabfs files at JIVE. The EVN status table and calculator were updated to include the information of three Russian QUASAR stations, and the Kunming 40m radio telescope. The "tpicd" monitoring messages were not outputted at Tr and Mc due to missing the "datavalid=" command. The problem has been solved in session 3/2009. The 5cm experiments need re-pipeline as Wb used an improper DPFU. K band -- Only one user experiment EM059D. The amplitude calibration at 22 GHz looked ugly. A median error >10% is quite "normal". A group consisting of the experienced persons should be formed to further investigate the problem. Cal Experiments -- EVN station should insist in performing calibration experiments according to the EVN schedule and report whether their Tcal show significant variation and rxg files are updated on the feedback web page of CL experiments. Sampler statistics -- Bad sampler statistics are often seen in some BBCs and at quite a few stations in the EVN user experiments. It could cause a sensitivity loss up to 25%. Technically, sampler state statistics can be monitored via systests samplestat procedure and script, or MK4 decoder samples commands, to check for potential problems in analog or digital domain. Thus, optimal sampler statistics is not a difficicult thing at the EVN stations now. Sick BBCs -- These BBCs usually had fringes with much lower (<0.8x) correlation amplitude and require higher amplitude scale factor in the post pipeline data reduction. For example, Ur had the problem in VC3 because of a bad pin connection; Nt was found in BBC01 where 76% samplers were located at status "01" in N09M2 and the auto-correlation plots had a huge spike at 1st frequency points. These BBCs can be identified by checking the correlation amplitude of fringes in the ftp-fringe test. S/X band -- Only one user experiment (EH023B) was scheduled in session 2/2009. Ef show large amplitude error at elevation <20 degree because of the surrounding hills. The other participated stations had good amplitude calibration. RFI -------------------------------------------------------- RFI seems to be the major source of errors in particular side subbands at L band. Due to the impact of RFI, the Tsys data were much noisy. Calibration at this frequency is variable with occasional experiments having quite large errors. The following is a summary in the Gbps experiment EP064D based on a few 1-min extracts, and looking at animations of the autocorrelation passband (after the 2-bit van Vleck correction) from 15:00-16:00 (an arbitrarily selected hour): The observation have a bandwidth 128 MHz (1586 -- 1714 MHz, 16 MHz/IF, 32 freq.pt/IF). Everyone has very variable interference in the upper half of IF2 (a frequency range which has been known to be susceptible to Iridium intermods (up to ~1628 MHz, a little bit into IF3). Jb -- IF1 frq.pt 14-18 variable Wb -- Prior to the 2-bit van Vleck, IF2-3 were horrendously affected, but in a sense that the entire pass-band would be moving up & down This was also seen in the 512Mbps L-band e-VLBI 21-22 apr, for which IF3 here was the same sky frequency as IF1-2. On -- IF1 middle portion variable probably also issues in IF6 near the top end, IF7 ~3/4 up. Mc -- Intentional filtering at station kills IF1 entirely, lower part of IF2 and upper part of IF7 (see BASELINE SPECTRA plots below). IF2 also has a stable strong spike at upper end IF4,5 fairly stable spikes near middle of each; some variability lower down in IF 5. IF6 show variable spikes at very low end and near top; smaller stable feature around freq.pt 11. IF7,8 have huge variable spikes near middle (seemingly symmetric about LO in between these two "IF"s) -- causes ringing through each passband. Tr -- IF1: fairly stable spike (~20% variation) around frq.pt 12. IF5,6: very bursty strong spikes in LCP (frq.pt 5,18 (IF5); 15,28 (IF6)). IF7: very variable very strong feature over 6-8 freq.pts (3-4MHz); enough to affect the level of the pass band after the 2-bit van Vleck correction. Ar -- IF1 variable spike in RCP near middle IF7 very variable spike around frq.pt 22 Jun YANG for the Science Operation and Support Group (JIVE)