Transition from the VLA Correlator to WIDAR 12-17-08 MMM I. Transition from WIDAR prototype correlator (PTC) to WIDAR0: March 1, 2009 A. Enough station boards will be installed in WIDAR0 for 10 baseband inputs. 1. We will have 10 station boards and 4 baseline boards. We will receive 6 additional baseline boards by Mar 13. 2. The number of baseline boards is adequate for OSRO. 3. We could use 10 antennas x 1 baseband each, or 5 antennas x 2 basebands each. B. The WIDAR PTC will be abandoned at this time. Boards will be moved from the PTC to WIDAR0. C. Priorities for WIDAR0 are 1. Prepare for OSRO with software and systems integration. The two WIDAR modes needed for OSRO are pseudo-continuum and spectroscopy in two tuneable sub-band pairs. 2. Testing of high speed samplers and wideband data path. E. WIDAR0 will have no more than 10 antennas until the VLA correlator is shutdown. F. WIDAR0 might grow as 1. 5 antennas, full polarization; Feb - Apr 2009 2. 10 antennas, single polarization; Apr - May 3. 10 antennas, full polarization; Jun - Aug 4. 10 antennas, full polarization, 2 independently-tuneable sub-band pairs; Aug - Dec G. We need to define what we want to achieve with these configurations and identify the software needed for them. H. Engineering tests 1. There is some desire to have a WIDAR0 available for separate tests of hardware and software, in addition to on-the-sky (OTS) tests. 2. Current preference is to time share between OTS and engineering test, but coordination problems have occured in the past. I. WIDAR0 will not be a general user instrument. II. Shutdown VLA Correlator: January 11, 2010 A. Assumptions. 1. VLA correlator shutdown on Jan 11, 2010, a. OSRO (VLA emulator mode) begins after hardware is moved and software is adequately tested. b. Shutdown date could be earlier, but probably not earlier than Nov 1, depending upon WIDAR board delivery and what we want to do. 2. We want to test WIDAR with as many stations as possible for one to two months prior to the beginning of OSRO. 3. Time is needed (~2 weeks) to move deformatters (deform) and demultiplexers (demux) between correlators. B. Moving deform and demux between correlators. 1. Work required a. Disconnect demux and fibers. b. Add components, primarily a new connector, to each deform board. c. Mount a deform on each station board. d. Install each station board in WIDAR racks. e. Reinstall demux and fiber. f. Test the station board. 2. Work schedule a. Jan 11: Proposed date for shutdown of VLA correlator b. Jan 11 - 25: Move 15 sets of demux and deform from VLA correlator to WIDAR as described in IIB. (15 sets needed because there will be 25 VLA antennas by late Jan and 10 sets already installed) c. Jan 26: Begin 25-antenna, NRAO-internal, WIDAR-only commissioning observations. d. Jan 26 - Feb 9: Decommission remaining VLA demux/deform hardware and move to WIDAR. C. Consequence of shutting down VLA correlator in Jan 2010 is that stress tests of the last 4 EVLA antennas must be made with WIDAR. IV. Deployment of 3-bit, high speed (4Gsps) samplers (HSSs) A. Four DTS modules with HSS will be available for installation on antennas by mid-Feb 2009. 1. Current default plan is to install 1 sampler in each of 4 different antennas. 2. At lease one baseband, and preferrably all, in each of these antennas must have: a. Upgraded T304 downconverter with wideband filter and gain slope equalizer b. Second set of L302 frequency synthesizers (?). 3. Each antenna must be connected to WIDAR. B. The installation of T304s started in Jun 2008, and is proceeding at rate of roughly one antenna per month. Initially, these should be installed on antennas that are connected to WIDAR0 to allow wideband tests with it. C. Beginning in Mar 2009, HSSs and the second set of L302s will be installed at the rate of one antenna per month (i.e. there will be about 9 antennas with HSS capability by the end of CY09). 1. As with the T304s, the initial installation of these modules should occur on antennas connected to WIDAR0 for wideband testing. 2. The HSSs are installed in the DTS module, and 4 DTSs are installed in each antenna. (This is the current default plan, although alternatives are being discussed). V. Configuration of deformatter racks. A. Currently, the deformatters and demultiplexers are configured such that 20 sets provide signals to the VLA correlator and 10 are dedicatd to WIDAR0. B. Using existing spare hardware, at little or no additional cost, the number of sets to the VLA correlator will be increased to 22. The number of sets to WIDAR0 will remain at 10. VI. Total antennas to VLA correlator in 2009. With the deformatter rack configuration described in V, no more than 22 EVLA antennas can send their signals to the VLA correlator. The number of antennas sending signals to the VLA correlator will decrease when more than 22 antennas are converted to the EVLA design. A. Feb 15: 27 total (20 EVLA, 7 VLA) B. Apr 15: 27 total (21 EVLA, 6 VLA) C. Jun 15: 27 total (22 EVLA, 5 VLA) D. Aug 15: 26 total (22 EVLA, 4 VLA) E. Nov 15: 25 total (22 ELVA, 3 VLA) (doesn't happen in Oct because of bearing change on antenna 20). F. Jan 15: 0 per proposal to shut down VLA correlator on Jan 11 In Aug 2009, the most recently converted antenna (#7) should be connected to the VLA correlator for stress testing in place of a well-understood EVLA antenna. This latter antenna will sit idle until full WIDAR is used (~Feb 2010). Same is likely to be true for antenna #20 in Nov. VII. WIDAR board delivery A proposed schedule for the delivery of station boards (SB) and baseline boards (BB) follows. The minimum quantities are those required for OSRO. The desired quantities are those needed to complete WIDAR installation. Date SB(min) BB(min) xbar(min) SB(desired) BB (desired) xbar(des) ----- ------- ------- --------- ----------- ------------ --------- Mar 1 10 4 3 10 10 ? Jun 1 20 4 3 32 10 ? Aug 1 40 4 3 64 32 ? Dec 1 96 4 3 128 128 all The delivery of SBs is obviously the priority. BB capability: 1 BB provides one sub-band for 32 antennas. 2 BB for 2 sub-bands thus needed for full polarization. VIII. Correlator backend computer (CBE) A. We will purchase 2 CBE nodes and 1 small switch for WIDAR0 in the next few weeks. 2 CBE nodes already on hand. B. We want to test the nodes to see how their required number scales with data rate. Target is 4 BB per node. Look to make significant node/switch purchase in late 2009. C. Number of nodes basically limits integration time. ~/EVLA/correlator/WIDARtransition.txt MMM, 12-17-08