GK022's cover letter just went out (attached below). The (6) FITS files live at juw30: /juw30_2/data/rmc/GK022/GK022.IDI? New standard plots on public FTP site & on (still private) jive/correlator web page. Folder downstairs with hard copies of plots (& old style plots). Hi Preeti/Denise -- The correlation and FITS-DAT(s) preparation for GK022 has finished, and I've just finished reviewing the results and preparing various plots for you to look at. The correlation used: 15 stations 2 8MHz IFs (X-band, band-edges at 8407.49 to 8415.49, all USB) 4 polarization per IF (parallel & cross) 16 frequency points (32 lags) per IF/pol 4s integration time. You can check the station feedback via the form under the JIVE website: http://www.jive.nl/jivebin/jiveese. Individual station logs are on the VLBEER machine in Bologna. I'm sending DATs to each of you at: Indian Institute of Astrophysics; Koramangala; Bangalore 560 034; India Dept. of Physics; National University of Ireland; University College; Cork; Ireland EVN and Global VLBI Observations correlated at the EVN MkIV Data Processor at JIVE are now automatically calibrated via a pipeline process. In particular, the apriori amplitude calibration table (and the associated, final ANTAB file), plus various other tables and plots (derived only from scheduled calibrator scans) can be downloaded from: www.evlbi.org/pipeline/user_expts.html [Note: these CL tables should also include the effect of correcting station positions via CLCOR, as discussed at the end of this letter.] JIVE also invites PIs to visit Dwingeloo in order to analyse EVN data. Financial assistance is available to all users located within the EU (and associated states). Contact garrett@jive.nl for more details. The experiment comprised up to 38 thin-tape passes looking at a couple targets with calibrators and a few fringe-finders here & there. This is the first experiment I've distributed using the new standard plots. These plots are available via anonymous ftp at: ftp ftp.jive.nl cd pub/jive/correlator/GK022/ The names of the plot files are: gk022-weight.2007.ps.gz gk022-auto-1.SRC.ps.gz gk022-cross-1.SRC.ps.gz gk022-ampphase.SRC.ps.gz where SRC = 1308 or 2007. For further review of the correlation (the standard plots don't show every integration/SB/Pol/etc.), I made a set of 'old-style' plots. I don't explicitly discuss them here, nor have I put them on the ftp site -- but they do exist. Essentially, they are the same sort of plots as the standard plots, with the addition of amplitude-vs-lag plots, but cover more of the experiment (temporally, N_bsln, N_sb, N_pol, etc.) Individual station problems, etc.: There were a fair number of scans out of telescope observing limits; see the sched-derived gk022.sum file. Principally affected stations include Mk,Ov (early); Wb (latter part of 3c66b/3c78 portion); EVN (last 6 scans of 3c78); western VLBA (first few 3c270 scans -- Mk taking longer than the rest, of course); EVN (last couple scans before leaving the network); Hn,Sc (last ~6 scans of exp) On, Yb - only RCP Discussion of plots: (a) The weight plot shows the station-weights for the whole experiment (not just for 2007) for IF1. Sources & polarizations are color-coded (product of Nsrc*Npol > available colors, so color-key not terribly significant). The first scan of the various sources is labeled with the source name in the upper portion of each plot. Each 4s integration is plotted as a separate "+" (because of the vertical extension of the "+"-symbols, the weight plots could appear somewhat worse than they might otherwise. Similarly, the horizontal extent of the +'s gives the impression that the instances of servo-loss (the temporary fall of weights to 0 due to tape- tracking problems) occupy a large fraction of the whole experiment than they actually do. These weights are more properly the fraction of 'good' data per integration period. The autocorrelations & cross-correlations (baselines) are correlated independently, but the baseline weights are generally the average of the contributing stations' autocorrelation weights (as plotted), except in cases where one of the stations has zero weight because of servo errors, where the baseline weight also remains at zero. Here, the cross-polarization station weights are also plotted; they are similarly just the average of the L&R weights for the station -- for a single station, LR==RL. You'll notice that some stations have stretches with a channel at weight=0.5. This was the result of a bad track on DPU/SU 7&9. The affected channel in each case was IF1 LCP (the cross-pol becomes 0.75). There's not much affect on the relative strengths of the fringes in/outside of the time-ranges with the bad track (a benefit of the high fan-out factor of 4). We tried to minimize the effect of the dead tracks on a given station by moving them around during the tape-changes: tape1 tape2 tape3 dpu/su 7 Mk Hn Kp (for tape2&3, IF2 LCP similar) dpu/su 9 Ov La Nl You'll also notice that all VLBA-format stations (VLBAs & Eb) have their weights drop to zero exactly at midnight at the day-boundary between the first & second days. Their weights stay at 0 for the rest of the scan that crosses midnight, but recover in the following scan. Thus these stations lose the last 5m36s of the 3c78 scan that started at 23:49:36. This is a recently discovered problem that the station units have in decoding the time in VLBA-format recordings at the day-crossing boundary; it's able to fix itself at the next-following tape-stop (in this case, at the 00:07:35 3c84 scan). An iteration of trying to fix this has occured, but didn't completely resolve the problem. (b) The auto plots show the station autocorrelation-amplitude spectra (bandpasses), with each IF plotted separately. These are done for 1 minute of data from 2 different scans (2007+777 & 1308+326 [vlba only]). The weight cut-off was 0.7, as annotated on the plot header. The color-code is (mostly) Red=LCP, Green=RCP, light blue=cross (for a station, RL==LR, so only one cross-pol is evident). All plots (including the different IFs for the same station) scale independently in the y-axis (not necessarily going down to 0); labels come only on the left-hand column. On & Yb recorded RCP only (thus their cross-pol is on the order as strong as the single-pol). The term "mostly" was applied above to the color-coding because if an IF is missing a channel entirely during the duration of the plot (i.e., due to being below the weight cut-off), then the color scheme might get thrown off. In this specific case, times when IF1 LCP is below weight=0.7, either due to a bad track (e.g., Mk in the 2007 scan, Nl,Kp in the 1308 scan) or recovery from a servo problem (Eb in the 2007 scan), then the remaining RCP is plotted in red (instead of the missing LCP). (c) The cross plots show amplitude and phase for baselines to Eb (for the 2007 scan) or Hn (for the 1308 scan). They have the same lay-out characteristics as the auto-plots, with the cross-pols now distinguishable: dark blue = LR, light blue = RL. (The amplitude labels still need some work.) You can see the effect of On&Yb having only RCP: their RR & RL are essentially the same (since for them L=R), and their LL & LR are as well (for the same reason). The detections in the 1308 scan are more clear in a scalar average over the whole 1308 scan (rather than just 1min). La didn't head peak for the pass with 2007 in it. It seems to have the highest residual delay, about ~60ns, remaining fairly constant throughout the experiment. (d) The ampphase plots show amplitude & phase as a function of time in the vicinity of the above scans, again on baselines to Eb. It plots only LL for both IFs. In the 2007 scan, the only sources appearing are 3C66B (red,blue) and 2007 (green, light blue). There is another color-shift in the Eb-Mk plot because some Eb-Mk IF0/LL didn't meet the weight cut-off. In the 1308 scan, there are more sources: 1308 (light blue, light green) 1156 (orange, dark blue) 3C264 (red, pink) 3C270 (green, yellow -- may look like really light green) Because only LL is shown, On & Yb show no detections. The phase slopes with time are generally pretty flat for calibrators. Since the positions of the targets are given to around the nearest " in the schedule, there are higher phase-rates for these sources (which get worse at longer baselines): working out the first-order expansion of delay-rate with a position offset of 0.5" gives ~0.6[x]b ps/s, where b=baseline in thousands of km, and [x] = the combination of cos/sin of Dec_src, Dec_bsln, HA (i.e., <1). This would give a phase-wrap time of ~200/([x]b) seconds. In the old-style phs(t) plots, you can follow the phase on the parallel-hands on almost all baselines to Eb (where both stations at El>0...), but, for example, 3c66b on Eb-Kp or -Br wraps almost once a minute. Of course, the orientation of the baseline matters (via [x]), so the actual delay rates are functions of time, and of course atmospherics effects will come into play, more so at lower elevations. FLAGGING -- Weights, DR=0 problems I flagged the data prior to making the FITS file(s) at a weight cut-off of 0.4. This resulted in a loss of 8.4% of the correlated baseline data. I also flagged times corresponding to the delay-rate=0 problems (which cause spurious amp(t) spikes as the phase-cal tones correlate with each other). Since the duration of the interval to be flagged is inversely proportional to frequency, this X-band experiment lost very little data to this effect (only 373 individual 4s integrations over the whole experiment). ------------------------------------o------------------------------------- Bob Campbell Correlator support scientist, EVN MkIV Correlator at JIVE