TOG Meeting, 2005 July 2, Onsala. Performance and Reliability of the EVN ====================================== Session 1/2005 -------------- Session 1/2005 went well in general, the NMEs produced good results. There were three short sessions in K, S/X and C-bands, and a long L-band session. A major problem was the weather, because it was often snowing and sometimes there were high winds. A pre-session ftp test was organised (independent of the block schedule) that showed an Effelsberg problem early. This saved the main K-band ftp test (and the session itself) because there is no chance to check for ftp fringes without Ef at 1.3cm. During the session there were three other ftp tests, which in principle produced feedbacks before the experiments ended - for those stations that sent the data early. The Socorro correlator complained about poor playback from EVN tapes. Cm - trouble with the microwave link at the end of the L-band session (due to the snow); the K-band rate spectrum was very broad in the NME and also in EI006A, reason unknown Ef - stopped N05K1 because of too short switching cicles (that were otherwise allowed by SCHED) - may need revisiting the slewing model of the antenna; had "snow stops" in several experiments; snow, snow, snow... Jb - 4 channels dead in the geodetic experiment EC019B - bad patching; stopped observations due to high winds in GD018 and EK020A; because of a Lovell Telescope drive problem, the MkII was used in EZ013 and GD017B On - intended to use pigwire setup in N05C1 (to secure the following tape-only experiment), but did not - most of N05C1 was lost, did not record to disk; receiver problem in the L-band session, used uncooled HEMTs in N05L1 and ES053A Mc - no ftp-fringes in the K-band yet (while disk data OK); snowing sometimes; K-band NME showed high polarization leakage - may be responsible for the lack of ftp fringes, combined with lower sensitivity because of bad weather???; bad disk caused problems in N05L1, and a more serious Mk5 problem caused losses in GP042 and GD018 (feedback said "FS-Mk5 communication problem") Nt - no fringes in the geo experiments EC019B (N05K1 was OK earlier) and EC024; there were pointing problems at high elevations in some experiment; ES049A and ES049B showed a problem with channel L2, related to BBC#4 U sideband; Tr - some snow; BBC#3 was unlocked during the whole session (due to a problem in the electronics unrelated to the BBC itself); in the last L-band experiment (EH010) the pipeline showed an L-band sensitivity problem: only the brightest fringe-finders were detected (VLBI friend was informed about this, reason unknown); the L-band NME did not show the sensitivity problem, and the other user experiments are not pipelined yet Ur - some snow; bad VC#2 replaced immediately after ftp fringe test in N05C1; high polarization leakage seen in N05L1, and also there was a receiver power supply failure near the end of that experiment; RCP Tsys was too high in GD018, reason unknown; Sh - no fringes in EC019B (N05K1 was OK earlier); sometimes "trouble with antenna control" - missed N05C1 completely; occasional minor Mk5 problems; Chinese telescopes now routinely participate in ftp fringe tests, even the most difficult K-band one was a success in Urumqi and Shanghai! Hh - some rain and RFI reported; the only major problem was the drive interface failure in EH010, which resulted in some data losses Wb - high Tsys due to snow; some feedbacks still missing Mh - N05K1 was all OK, but there were no fringes in the geo experiment EC019B Ro - had some tape position errors, and in EI006B also tape changing problems (because there was not enough time left in the schedule); strong RFI reported in the L-band Ar - seemed to be OK; missing L-band feedbacks Wz - no feedback for the only experiment, EC024 Session 2/2005 -------------- As long as ftp fringes concerned the session went well. However Noto dropped out from the session very soon, because the azimuth engine broke. Torun used Rubidium maser which is not stable enough for C-band. In Effelsberg the MkIV formatter broke and three experiments were recorded on tape. The NMEs have not been correlated yet. Although Mk5 recording is superior to that of tape, it seems a number of times recording stops ('Mk5 hangs'), and there is often trouble with the 'FS-Mk5 communication'. This can be best solved by rebooting the Mk5, which is time consuming. Occasionally there are bad disk packs. Cm - Trouble with the elevation engine in some experiments. It completely broke in ES054B, but the problem was fixed before the next experiment, N05L3. ES053B was a failure due to an error in the MERLIN control software: telescope stuck on one source. Had Mk5-FS communication problem in N05L4 - failure. Ef - Formatter trouble in ES054A, VC#6 and #8 lost. Formatter broke in ES054B - continued on tape (including three other experiments) till EVN spare arrived from Medicina. About half an hour data lost in EK020C due to Mk5-FS communication problem. Lost 20 minutes in EI008 due to FS crash. In EG034 data were lost from 02:16 UT till the end - again for a Mk5 problem. Wb - Accidentally overwrote EB030 logfile. Lost recording for about half an hour after bank switch in ES054A. Real time array control system got stuck for more than an hour in N05L4 - calibration data lost. Lost vacuum several times in GB049C. In EP049C also had trouble with recording, but now the Mk5 recording stuck (twice - more than an hour of data lost). Jb - ES053B was a failure due to an error in the MERLIN control software: telescope stuck on one source. Lost 10 minutes in N05C2 due to a bad disk pack - no other problems. On - Lost about 30 minutes in EB030 because Mk5 hanged. VC#2 was unlocked in ES053B. In EI008 scan_check could not deal with three scans. Mc - Was late from N05L3, recorded only from scan#40. Lost 2 hours in EJ007B due to an FS problem. In GB049C had to start late due to various technical problems. Nt - Observed successfully the first experiment only (N05L2). The first day of EK020B had trouble with the Mk5. Afterwards a failure of the encoder of an azimuth engine stopped the antenna. Could not fix the problem during the session. Tr - Used Rubidium frequency standard in the whole session, because the H-maser was sent for refurbishment to the manufacturer. Note that although Tr produced fringes, the C-band data will be useless due to the very short coherence time (we had the same problem with Noto in session 3/2004). BBC#3 unlocked all the time (rack-related problem, not the BBC itself). Ur - Observed according to version 1 of the block schedule, instead of version 3 - missed EK020C for this reason. Sh - No major problems reported. Hh - Lost more than an hour in EB030 because Mk5 failed to record after bank switch. Ar - FS lost contact to Mk5 in EK020C. About half of the experiment was lost. C-band ftp-tes showed fringes. GT006 went OK. Ro - Participated in ES054A. Apart from RFI and some antenna limitations, there were no problems. Ftp tests --------- We are now routinely using 4s ftp files for the fringe tests. This speeds up all steps of the data processing (ftp to JIVE, Mk5-K5 conversion, correlation). Except for Urumqi and Shanghai (and for those who do not ftp their data!), we are able to produce a result within two hours. This will further improve only when all the stations have good connections, and the files are not only produced, but also automatically transfered to JIVE by the FS. Note that automatic file transfer could be made operational by the next session for most of the stations. It cannot be done in Shanghai, since the data have to be taken from the telescope to the institute first. In Urumqi the problem is that they do not have a valid IP address, and the Astron firewall do not let them login. Westerbork still produces smaller than expected ftp files. This is not a major problem as long as Effelsberg or the Lovell telescope is available as reference antenna, but would be good to fix. Since there was a pressure from the stations to have pre-session fringe-tests, we suggest to introduce a regular Tuesday ftp fringe test (sessions start on Thursday). TOG action item on 1 Gbps software correlation: installed new version of the NICT software correlator that is able to handle data recorded at 1 Gbps. Have not tested yet. K-band performance of the EVN ----------------------------- Andreas brunthaler processed the K-band NME and wrote a report on K-band phase- referencing. It seems phase-referencing works well on the shorter baselines. The longest baselines need further self-calibration. Applying the polar motion model did not lead to significant improvements. The data have been insufficient for determining station-based zenith delay errors - this proved to be very effective in current VLBI astrometric projects. TOG action item on K-band amplitude calibration: investigated whether we could use the NME gain solutions for the user projects - no. The gains are not stable, they often change significantly within a project. Zsolt Paragi for the Science Operations and Support Group (JIVE)