GBT Commissioning Meeting for 14 February 2003 AGENDA 1. Az Track Status -- Bob A. 2. Spectral Baseline work -- Rick 3. K-band and other commissioning -- Ron 4. Project scheduling -- John 5. Observing schedule -- Carl 6. AIPS++/Dish status -- Joe M. 7. Software status -- Nicole 8. Spectrometer status -- Rich 9. Any other business REPORTS and DISCUSSION 1. Az Track Status o Lee King is wrapping up calculations on stress improvements for the trial modifications. He has requested that we add a one inch weld to the sides of the baseplates prior to attachment of the side blocks. The preliminary numbers generated will be an improvement over what we have now. We have informed the potential contractors for the welding demonstration, and asked for their recommendations on techniques. Bob Anderson has had numerous discussions with all 3 companies, and they are making progress toward their submittals. This latest addition will push their submittals into next week. o David Parker and Tim Weadon are in the process of reducing the track bolt test data. o Gadsden Tool has not received the wear plate material from the mill yet. They expect to ship to us the week of the 24th . o Harry Morton has performed a survey so that we can try drilling the bolt holes in the new plates without needing to keep an existing plate off of the track to use as a template. He will reperform the survey to validate the method. o Dennis Egan is in Germany to observe the Effelsberg track repair. -- RA Discussion: Fred Schwab asked if a post-weld heat treatment was required. Bob responded that it was, and is a complication, because the heat must be kept below the temperature that will damage the grout and concrete. Heat conduction will be measured in the mockup test that the welding firm will do. Bob also noted that the cracked plates are all on the south and west sides of the track. There is no immediate explanation of this, although theories are certainly abounding. 2. Spectral Baseline work Roger and Galen have produced a thorough analysis of reflections at the ends the cables used to connect modules in the GBT IF System. The conclusion is that quite small reflections and the delay in these cables, which are several meters long, introduce a significant frequency dependence in the transfer coefficient. Changes in the lengths of these cables of less than one electrical degree around 1 GHz is enough to cause the drifts in the imposed passband ripple that we have seen in our spectrometer measurements. The cable lengths and ripple periods tentatively agree well, but we need to take a full inventory of cable lengths and match these to the various ripple periods measured. Temperature variations in the equipment room are typically on the order of a few tenths of a degree C, which is about as good as we can expect. This is enough to change the cable lengths more than can be tolerated. After determining which cables are involved, the plan is to replace them with phase-stable coax, which should provide a factor of ten improvement. Data on the K-band (18-26 GHz) receiver from several observing periods have produced conflicting results on receiver gain and baseline stability. We will continue to run diagnostics on this system, but we will leave the receiver on the telescope until a reproducible problem can be established. At X-band (8-10 GHz) a test to look for resonances in the feed structure by covering joints with conductive tape showed no change in the system temperature frequency structure. The Ku-band (12-15.4 GHz) receiver is in the GBT equipment room undergoing cold load tests. Several sharp features in the noise spectrum of this receiver have been identified, and Roger and Mike are in the process of eliminating various possible causes of these apparent resonances. This likely will require work inside the dewar. Restrictions of the azimuth travel of the telescope have made the investigation of baseline calibration procedures too difficult and inefficient. Hence, we will concentrate on the IF system problems, finish reduction and documentation of the most recent measurement runs, and begin writing a summary report on all findings in the baseline investigation over these last three months. We'd like to release the remaining GBT time currently scheduled for baseline tests and request time for IF and receiver tests as needed for the immediate future. In mid March we'd like to make presentations in Green Bank and Charlottesville on this work. A TUNA talk is scheduled for March 11, and we'll schedule something similar in Green Bank around that time. -- JRF Discussion. Rick noted that the group has about 35 individual test reports that they will summarize in the overall report on the study. 3. K-band and other commissioning Since the weather has kept us from observing, most of our plans for this week did not get executed. Most activities revolved around observer support, proposal checkout, training, and staying warm and out of the wind. The proposals we tried to check out, and our conclusions about each, are: 2B-002: Looks like a project that can get scheduled. The observers are doing an absorption experiment against sources with Ta < 0.5 K. They will look at about two dozen objects for 1 hour each in a 200 MHz bandwidth at different upper-K-band frequencies and will use the NOD beam-switch observing mode. The observing setup is possible. I integrated for 3 hours on a 0.5 K source and found very acceptable bandpasses. I need to send the spectra to the PI's before they can tell us whether the current telescope performance is suitable for their experiment. 2B-004: I worked with Harvey on checking this proposal out and we found it to have a few technical flaws. First, the PI's are observing a point source and desire to observe with both the upper and lower-K-band receivers simultaneously. Since this is not possible, we'd have to double their time of the telescope to accommodate them. They also want to observe 8 transitions simultaneously in two polarizations in a 50 MHz bandwidth. Although the Spectrometer has 16 inputs and could accommodate the observations, the wiring between the Analog Filter Rack and the Spectrometer for the K-band system will only allow up the simultaneous use of only 8 Spectrometer samplers. We'd have to again double their telescope time to accommodate the observing. 3A-019: The winds shut Harvey and me down before we could try out any observing. This proposal want to observe two transitions with both polarizations ion a 50 MHz band. All we know is that we can get the appropriate signals into the appropriate Spectrometer samplers. 2C061 & 2C062: Frank's attempts at checking out these proposals were also thwarted by the weather. All we know is that the system can be wired up into the appropriate Spectrometer samplers. We have yet to start the checkouts on: 2A63, 2B03, 2B04, and 2C12. The plans for the next two weeks is more proposal checkouts by Frank and me and Spectrometer checkouts by Karen, plus observing support for all. -- RJM Discussion. Richard raised the issue of whether the Q-band receiver could be installed within the next week or two, so that efficiency measurements could be done. These can probably be done with the present azimuth restrictions, and might be a good use of time now. John replied that this would have a direct impact on the spectral baseline work, since it involves the same people, but that he would look into it. 4. Project scheduling Project planning: We have been busy this week working on the 6 month plan. We are nearly to the point of making decisions on the list presented last week. Much information remains to be considered before we make concrete (at least plaster of paris) decisions. Priority projects: Az track work: Consisted of bolt re-torquing and data analysis on the bolt tests. Baselines: Ku band receiver work in the lab continued, along with IF testing. Astronomical tests are also still underway. PTCS: PTCS telescope time was winded out. It will have to be rescheduled. -- JF Discussion. Phil reiterated that the proposed top 4 priorities in the 6-month forward look were the az track, ease-of-use improvements, spectral baseline program, and PTCS. 5. Observing schedule This week's summary: Last week ======= Observing for GBT02A-049 (pulsar), GBT02A-008 (low freq backup project), GBT02C-033 (high frequency observing) More severe azimuth motion restrictions implemented Some observing not possible and other programs very inefficient February ====== Remaining Astronomy ~ 100 hours ( ~ 100 hours of backup) Maintenance ~ 68 hours Test & Comm ~ 170 hours (cancelled ~7 hours and will probably cancel another ~ 24 because of severe azimuth restrictions) March ===== Astronomy ~ 336 hours (backup ~ 106 hours)* Maintenance ~ 146 hours Tests & commissioning ~ 266 hours * Numbers do not consider additional track repair time and assume the severe restrictions can be relaxed significantly by mid march. Other ==== Slight change to the dynamic scheduling procedures. The decision is now made every day including on weekends. Details available at http://wwwlocal.gb.nrao.edu/GBTopsdocs/GBTschedules/dynamicscheduling.htm -- RCB 6. AIPS++/Dish status o Observing Sig/Ref Problems There were some sig/ref incongruities encountered in Hogg et al. data. This is suspected as being a problem in the interaction of the observing interfaces which lead to a state which is different than labelled. Ron has sent a note out with a fix on the interface side. Bob has provided a fix on the filler side when data of this sort is enountered. Mis-matched phase problems This ended up being caused by aborted scans only and was not something systematic to the observing. This is better handled on the AIPS++ end. Regression test problems The regression tests were run with an installation of AIPS++ which had an unforeseen and unencountered behavior change in glish. The scope of the problem was restricted to cases where the dynamically created field of a dynamically cast record of a dynamically cast vector of records is dynamically cast to be a record. There were no instances of this in the AIPS++ code but it did exist in one instance of GO for setting the velocity of a source. We agreed to the need for a GO test in the assay of AIPS++. In addition, we clarified the procedure for testing; in this case, the stable version of AIPS++ should be used and would have avoided this problem. Coordination There was an AIPS++/GBT coordination meeting on Tuesday of this week. We agreed on the February targets. Discussed documenation issues - incorporating user sections into the cookbook, the use (and definition) of regression tests and several IARDS issues. In particular a second meeting was held to discuss the upgrades to IARDS and the improvements to the communication between IARDS and M&C. A further discussion of the upcoming changes in the FITS files was made. o New Stable Snapshot. For details see: http://aips2.nrao.edu/daily/docs/reference/updates.html o Targets joe migrate glish access utilties 40 28/2/03 bob gbtmsfiller: multibank data to single MS 80 28/2/03 joe gbtms average/decimate routines 40 28/2/03 bob gbtmsfiller: support weight spectrum 40 28/2/03 jim iards: display 1 spectrum/spw 40 28/2/03 jim iards: append to one ms 40 28/2/03 postponed bob gbtmsfiller: document 40 30/3/03 Calendar: 3/5-3/7 AIPS++ External Technical Review o User Support Hogg et al. Lockman issues - incomplete! o Defects resolved: Use http://aips2.nrao.edu/ddts/ddts_main to examine AOCso04322 broken for sdrecords selected by polarization AOCso04292 the tsys values in the dish header are incorrect AOCso04170 plotter labeling of polarizations should be more consistent AOCso03974 calibration of multi-IF scans failed AOCso04187 for OFFON scans the OFF scans are not calibrated. AOCso04096 multiple uses of import bombed -- JMcM Discussion. Steve Myers added that he had held a scientific requirements meeting in GB while visiting earlier this week. They are developing a specification to cover end-to-end requirements. 7. Software status Software Development Division #24 - Friday, February 14, 2003 The state of M&C v3.11 is "red" this week - unfortunately, not in honor of Valentines Day. Although we have completed week 4 of a 4-week development cycle, the intended release of M&C v3.11 on Thursday, February 13 was aborted due to an overload of messages into the system from the IF Manager when the system was in production use only. Although all of our unit tests passed, the errors that occurred would only have been caught through regression tests or during telescope operations. Since the regression tests planned for this past weekend were abandoned due to problems with the interaction of local software and glish (which were resolved soon after), the issues were not identified until v3.11 was moved into production. The key updates for this release are the inclusion of frequency and polarization information into the IF FITS files, a fix for the spectrometer duty cycles issue, updated active surface software which includes a fix of the 100ms problem, and addition of the K and Ku band receivers to the Configurator GUI. The current version of the software, M&C v3.10, was patched with the spectrometer duty cycles fix on Thursday afternoon to accommodate the work of commissioners and electronics. Our goal at the moment is to conduct astronomical tests as soon as possible, to ensure a solid release at the earliest possible time. As the SDD completes astronomical testing of these features with their sponsors next week, developers will begin working on commitments agreed to for the March cycle, which is targeted for March 12, 2003. Key efforts are Operations and Usability, focused on completing improvements for spectrometer reliability, and PTCS, with continued development on the EMS and discussions regarding new instrumentation. For details please review the SDD Plan of Record at http://tryllium.gb.nrao.edu/images/POR_Mar03.jpg. This cycle, 57 new unit tests were written, bringing the total number of unit tests run every day on the M&C system up to 141. Although unit tests do not catch every issue that may be seen when the telescope is in use, they do reduce the SDD's troubleshooting burden while using time on the telescope. -- NR Discussion. Paul Marganian gave the report in Nicole's absence. 8. Spectrometer status Three types of hardware problems occurred during the past week. Symptoms of the first include: 1 - Spikes in the auto-correlation function (acf) every 16 lags, 2 - Almost complete failure of the self-test. It shows problems in every correlator chip in the system, 3 - Intermittent Symptoms of the second include: 1 - Small steps of length 32 or 1024 in the acf, 2 - Ringing in the frequency spectrum, 3 - Does not occur until after the second integration. 4 - Seen primarily in the second quadrant, occasionally in the first, and never in the third and fourth. Symptoms of the third include: 1 - Occurs in 2-quadrant, one 800 MHz sampler mode, 2 - See a DC offset in the first half of the acf 3 - See a factor of 2 or 3 more noise in the first half of the acf. 4 - Intermittent. Due to limited test time and the intermittent nature of two of the problems, it has been difficult to make progress on these problems. When are problems bad enough to stop observing and do some intensive trouble-shooting? This may be worth discussing during the meeting. Rick gave final engineering approval to the following engineering modes: 1W2-001-800, 1W2-123-800, 1W2-001-800, 1W2-123-800, 1W2-245-800, 1W2-367-800. In addition the following modes were tested: 2W2-401-200 and 2W2-523-200 in multi-bank mode: looked good except for a few spikes in the spectrum probably related to the test signal 1W2-001-200 and 1W2-123-200 in multi-bank mode: looked good except for a few spikes in the spectrum probably related to the test signal 1W2-001-200, 1W2-123-200, 1W2-245-200 and 1W2-367-200: looked good except for spikes in the spectrum, probably related to the test signal, and except for a significant slope in sampler 6 (counting from 0) We also supported a pulsar spigot test run. A little progress was made testing scripts and checking serial communications between the spigot computer and the spigot cards. The spigot cards seemed to have a timing problem. Plans for next week (comments please): a) - Trouble-shoot problems identified above. b) - Re-design noise source so that it picks up less contamination in the test signals that it emits. These spurious signal always call the test results into question. We really need a cleaner test source. In addition, we will soon need a test source to check cross-correlation modes. This redesign needs to take this into account as well. c) - Test additional modes in multi-bank mode. d) - Work on the timing problem in the pulsar spigot. -- RL Discussion. We discussed Rich's question about how we detect subtle spectrometer problems that might be affecting baselines, etc., and how bad do they have to be before a fix is required. This needs some thought, but we may need to establish self-tests to be run on a frequent interval to detect these. The Spectrometer support scientist could then make a decision about the need for repair. 9. Any other business Glen asked if we should be doing some RFI checkouts to see if some of the low frequency programs could be scheduled, following recent improvements. The engineering group has not yet evaluated the recent changes to the feedarm servo, because they haven't had access to the telescope owing to bad weather. After this is completed, we should do some astronomical checkouts to see if these programs can be scheduled. PRJ 14 Feb 2003