Minutes GBT Commissioning Meeting Friday, 10 January 2003 AGENDA 1. Spectral Baseline work -- Rick 2. PTCS project work -- Richard 3. Az Track work -- Bob A. 4. K-band and other commissioning -- Frank 5. AIPS++/Dish status -- Joe M. 6. Project and observing program scheduling -- John, Carl 7. Software status -- Nicole 8. Spectrometer status -- Rich 9. Any other business MINUTES 1. Spectral Baseline Report We have been working with Rich Bradley and Srikanth on a noise model of the FET and GBT OMT/feed combinations to quantitatively determine the likelihood that known impedance mismatches are the cause of spectral baseline distortions on continuum sources. Initial results look encouraging. Since substantial improvement in receiver design is not feasible in the near term we are investigating calibration strategies as reported last week. A look at continuum sources near the beam half-power point shows only a small difference from beam-center spectra some moderately extended source calibration looks practical. Norrod, White, Balser, and Fisher met to discuss the next move regarding the 2.6 MHz optical modulator gain ripple problem. We decided to recommend purchase of eight more modified units like the prototype recently found to give a 15 to 20 dB improvement in this gain ripple. Steve will request a quote from the manufacturer. In parallel with this we will plan tests to isolated the cause of gain or phase variation in the ripple of the current units with the intention of reducing the amplitude to this variation for additional stability improvement. Measurements of the 8-10 GHz (X-band) receiver with the new (interim) InP LNAs were made on Friday (03 Jan 2003). Only one polarization was available because of the 100 MHz comb in the second channel of the spectrometer. On-off measurements of the continuum source 3C48 (Flux density = 3.4 Jy @ 8 GHz) produced baseline structure similar to the previous LNAs. Baseline structure exists with amplitudes (peak-to-peak) ~5% of the source temperature on frequency scales > 100 MHz and ~1% of the source temperature on frequency scales < 100 MHz. We need to confirm that these are upstream of the first mixer and in both polarizations. On-off measurements on cold sky with the Y (axial) sub-reflector focus shifted wavelength/8 and wavelength/4 reveal ripple periods of 9.1, 29, and 54 MHz corresponding to delay paths of 33, 10, and 5.5 meters. The 9.1 MHz ripple (strongest in the wavelength/4 data) is consistent with reflections from the feed to the sub-reflector and has been observed at L-band. Initial measurements of the 12-15.4 GHz (Ku-band) receiver were done Wednesday evening. Wind conditions precluded movement of the telescope, and the aftereffects of the power outage earlier in the day made this run problematic. The rapidly changing equipment room temperature may have called attention to a particularly temperature-sensitive electronics module. An intermittent spectrometer anomaly in one of the 800 MHz modes was seen again and reported to Rich and Holly. G. Watts began linearity measurements on modules in the Converter and Analog Filter Racks. Initial results are good. Because of the many channels, Galen plans to develop an automated test set to perform these tests. S. White discovered "gain expansion", an unexpected non-linear behavior of commercial medium-power amplifiers causing the amplifier output to input power variation to be greater than 1dB/dB, well below the amplifier 1dB compression point. This has now been seen in amps from two vendors, and we continue to work to ensure we are not operating these amps in the gain expansion region. A 100MHz comb was encountered in various channels when doing spectrometer observations. Injecting noise sources at various points and swapping cables isolated the intermittant comb to the spectrometer itself. -- JRF [Phil asked what the prospects were for improved impedance matches in future designs of feed / amplifier systems. Rick noted that simply putting isolaters in did not make much difference, but that they are continuing to investigate longer term approaches.] 2. PTCS project work Kim continues to work on the high level system architecture design, and we now have a first draft which we feel is ready for public consumption. This will be presented to a PTCS project meeting next week. Some preliminary discussions with Rick, Brian and Dana to confirm (at least in general terms) that the design would not rule out any required control behavior were extremely positive. Some commissioning time on Sunday in poor conditions was used to investigate the repeatability of low-level pointing residuals as a function of time and position, by following a number of sources over the same trajectory in (az,el). Unfortunately the backlog of unreduced data continues to increase. The SDD group continue to work on the Active Surface, including a redesign of the 100ms control loop to allow correct operation. The AS laptop network connections in the Actuator Room have been upgraded to fiber, to remove coax segment (potential source of RFI). Work has also started on the Engineering Measurement System project charter. -- RMP [The temperature dependence of focus was brought up in discussion. Richard and Dana noted that the focus can be checked fairly easily, and that this should be done periodically during the day by observers. In the longer term, it may be possible to handle this through the FEM or by other methods using the PTCS and temperature sensors. This is on the list of tasks for PTCS, but is not at high priority given that empirical calibrations are possible.] 3. Az Track work Az Track work during week of 1/6: Prep work continues for the bolt strain test. Materials ordered for the test pieces and the wipers on 1 bogey are being modified to allow passage over the bolts. --RA [Present plans call for painting the zinc-rich paint onto the base plate and wear plate late next week. Relatively mild weather, or a means to heat the surfaces for curing, are needed. The bolt strain test is presently scheduled for the week of January 20.] 4. K-band and other commissioning K-band commissioning over the last week consisted of testing 4-bank spectrometer modes at both 200 MHz and 800 MHz bandwidths. Jan 4-5: (project id: TKBASE_JAN04) The spectrometer was set up in 4-bank 200 MHz mode: 1W2--200 The frequencies going to the 4 banks were spaced so as to cover a 760 MHz span of the spectrum. Several continuum sources were observed, each at 7 rest frequencies so as to explore the entire frequency range of the lower band, namely 17.8 - 22.6 GHz. Jan 9: (project id: TSPECTEST_JAN04) Used 4-bank, 800 MHz mode: 1W2--800. Again setting frequencies to span the whole frequency range of the lower part of K-band. The spectrometer was re-booted at the start of the run, and had to be re-booted again part way through. There were problems with wind gusts, so no actual data on radio sources were taken, but we did some spectra on blank sky with the antenna at survival position. Data were taken with both beams. In general, the 4-bank data taking worked well. The reduction of the data in aips++ is somewhat awkward because each of the 4 banks must be processed separately. There is a problem with getting high enough power levels in some channels for the spectrometer to "balance" properly. Although there is adequate power registered in the IF Rack (1 volt levels), there is not enough further downstream. The problem was only in the RCP channels. In general we could balance the signals from the LCP channels. For the 800 MHz run (Jan 9th) I tried increasing the power level in the IF Rack for the RCP channel to 2, 3, and 5 volts, and was then able to get a good balance in all but one channel (J8). The only other problem noted with the spectrometer was a 100 MHz comb seen in the 800 MHz spectra in the channel of the J4 sampler. In general, the baselines in the (on-off)/off displays look fairly good. -- FG [Frank showed some spectra taken with different Spectrometer banks and polariztion channels. Some of the channels had pretty flat baselines, and several were very wobbly. The Baseline Project Team is taking a systematic approach to isolating these problems.] 5. AIPS++/Dish status o New stable functionality - available in /home/aips++/stable1 stable link will be changed. o Stable Snapshot page - off of main daily page (5th bullet) or: http://aips2.nrao.edu/daily/docs/reference/updates.html o Looking at K-band data Al's data - problems with bad PORT numbers rendering the data uninterpretable. Bob has created some augmented multi-beam data for testing. Calibration routines and NOD support underway. Some work with Kristy on continuum data. o January targets - Full targets listed under Stable Snapshot page. SD relevant items are: joe beam switched calib refinements 30-jan-03 joe beam-switched imaging 30-jan-03 bob gbtmsfiller: support bs data 30-jan-03 bob gbtmsfiller: support multi-beam data 30-jan-03 jim bs calibration refinements 30-jan-03 sanjay multi-beam imaging 30-jan-03 Multi-x targets discussed in separate meeting 1/14 at 1300 EST. o Defects resolved: Use http://aips2.nrao.edu/ddts/ddts_main to examine AOCso04253 error on fill of scans (not a bug) [The problem with the displaying the data that Al took in late December was discussed. At present, only the first IF can be displayed because of missing information in the IF FITS files. The bug that caused this is being addressed as a priority by the Software Group, and will probably go in as a patch in the near future. In the meantime, the problem with the existing data could probably be fixed by editing the IF FITS files.] 6. Project and observing program scheduling From John Ford: Project Scheduling Report: False start this week on consolidating the entire project scheduling matrix into one schedule for release to general audiences. Instead, project milestones are being distilled from Project into a text format containing the task, the target date, the actual date completed, in analogous form to the Quarterly Reports. This list will be the one posted to the web for public consumption. Baselines and spectrometer testing and repair continue to use all the time that has been allocated, currently 3 daytime periods per week. Baseline astronomical tests are also ongoing at regular intervals. -- JF Observing Report from Carl Bignell: Last Week ======== GBT02C-049 Radar Obs of Mercury (Failed due to Goldstone failure) TOO - New pulsar discovery confirmation Polarization tests - Troland/Heiles Next week ======== GBT02A-052 Pulsar observations GBT02C-049 Radar Obs of Mercury Polarization tests cont'd January ====== Astronomy - 39 hrs + 86 hrs of backup Maintenance - 113 hrs Tests+commisioning - 572hrs February ======= Preliminary schedule is now on the web Astronomy - 63 hrs (no backup yet) Maintenance - 106 hrs Tests+commissioning - 132 hrs Un scheduled - 362 hrs -- RCB [John added that the C-Band Receiver was coming off the telescope soon for replacement of the LNAs with the new, InP amps. The X-Band upgrade will be done later. We discussed the rudimentary dynamic scheduling system a little. Details are still being ironed out, but the system will be one in which the PI of the primary program makes the decision on a daily basis. Carl noted that we have lost quite a bit of observing time to high wind. Some of this lost time is due to the tighter restrictions in place because of the az track.] 7. Software status We are in Week 4 of a 5 Week Development Cycle, preparing for the release of M&C v3.10 on January 16, 2003. This week has focused primarily on tying up loose ends in preparation for the release of M&C 3.10, including the following: 1. Data Export Improvements (Filler). Green. Incremental work continues; may be ready to start validation by end of January. 2. Baseline Investigation Support. Green. Joe and Mark have completed testing of this facility, after a hardware modification was performed to reset counters at the start of a scan. DCR will now be able to collect data as a slave while spectral line data is being collected. 3. Antenna Control Unit Improvements. Green. Testing complete but awaiting sponsor testing prior to rollout for snow dump/stow functionality. 4. Multibank Capability. Green. Debug file generation has been patched into 3.9 and will be fully tested and validated for inclusion in the new 3.10. 5. Software Project Office Online. Yellow. Nicole is working to get this out ASAP so that it can be used to funnel Requests for the release planning complete date of February 11. 6. Configurator Development. Green. Melinda has added the K and Ku band receivers to the Configurator. The changes are awaiting acceptance/sponsor testing by R. Maddalena. 7. Engineering Measurement System. Green. Project Charter underway; end January completion goal. 8. Active Surface Improvements. Green. Project Charter has been reviewed and comments are currently being incorporated into a final product. Amy and Melinda have begun work on redesigning the 100ms control loop based on the requirements specified in the MR. Amy expects to lead a code review of the entire AS code base during the week of Jan. 13. The team is now prepared for the first round of regression tests to begin, which will be performed by Toney on Monday. Release Date Update: To better accommodate astronomical and other scheduling, the M&C system will be released within +/- 1 day of the posted 2003 scheduled dates. Note that the 1st release of 2003 will come out on Thursday, January 16, which is one day after the posted date. -- NR 8. Spectrometer status All Low Speed Samplers have been modified for flatter bandpass. All, including spares, have been tested in the system. Test data is not fully analyzed yet. Preliminary analysis indicates a few higher than desirable offsets and some spikes in the spectra of a few samplers. The data needs to be fully analyzed and anomalies explained and/or fixed. A 100 MHz comb has been seen in 800 MHz spectra. It has been difficult to isolate because it vanishes when we try to isolate it, only to re-appear during commissioning runs. One difference between our tests and the commissioning observations is that we observe in total power mode while the commissioners generally switch the CAL signal (should not make any difference right?). We can try this in our tests. If this is not successful, we may need to interrupt commissioning work when the comb is seen to try to isolate and fix this problem. A glish function has been written and tested. It's purpose is to look at all the individual integrations in a scan and flag those with large steps. It will be used in conjunction with the freshly delivered (thanks Amy, Eric and John) code to store raw lags to fix the problem with large autocorreltion function (acf)steps. Plans for next week include installing and testing slightly modified Correlator Control Card firmware. The modifications should improve the performance of the pulsar spigot with respect to the "ghosting" that has been seen. Also, we need to set up for and support a pulsar spigot test scheduled for Tuesday. Finally, several signal integrity issues are plaguing the spectrometer at this time: large steps (0.01)near at the middle of the correlation function, small (0.00001) steps and sawtooth features in the acf in at least two bandwidths, an intermittent 100 MHz comb in the 800 MHz data, ringing at the low frequency end of the spectra of one High Speed Sampler, one-lag and 1024-lag glitches in the acf seen during multi-integration scans. My reading on the priority of these is that the highest two are the 800 MHz comb and the large steps in the acf. Barring any guidance these two will be attacked first. -- RL [In discussion, the suggestion for the two priority problems to address, the ACF step and 800 MHz comb, were affirmed.] 9. Any other business Following a suggestion by Richard, since all the commissioning group leaders are contributing written reports for inclusion in these minutes, we will now ask for these contributions by ~4 PM ET each Thursday afternoon. Phil will compile these reports before the close of business on Thursday into a "pre-minutes" document and place it on the web at http://www.gb.nrao.edu/GBT/gbtstatus/minutes.html (or by going to www.gb.nrao.edu, clicking on "GBT News," "Commissioning Status and Information," then "Minutes of the Friday Commissioning Meetings"). The reports can then be reviewed by participants prior to the meeting. During the meeting itself, we will hit only the high points, leaving more time for discussion and items toward the end of the agenda. This scheme will start next week. Phil will send out a reminder next Thursday morning. PRJ 10 January 2003