Summary from 16-17dec02 Penn Array meeting at UPenn 03jan02 bsm Agenda: discuss & resolve issues associated with detectors (primary agenda item), data acquisition, & electronics; mutual debriefing on data analysis and cryogenics (and other) progress. Determine who (between Penn & GSFC) is buying or making what when. (I have not written down the details of the acquisitions-type transictions) DAQ Software ============ -The Goddard/NIST system uses Instrument Remote Control (IRC), a java-based DAQ application developed by GSFC. [see http://pioneer.gsfc.nasa.gov/public/irc/] -The software is fairly sophisticated: it supports data display, inspection, and archiving amongst other things. This will be useful at least in the lab, and possibly on the telescope. DAQ Computer ============ -The detector groups at NIST & GSFC currently run the IRC software on a dual-processor pentium (1 GHz) with 1 GB of RAM. We are taking this as a nominal spec based on their experience. They run windows98 (but the drivers are in linux); and are migrating to linux -The DAQ computer currently uses as custom fiber IO board to read the multiplexed data. There are a total of ~ 9 fibers, with > 1 MB/sec bandwidth. Penn is checking out the availability of commercial equivalents; otherwise NIST/GSFC will supply. -ACTIONS: *Goddard will work on assembling a working linux setup; when complete, Penn will buy an identical system. (?) *BSM has passed the GB "online hardware" recommendations to both parties. These were: -EBX or PC/104 form factor (to keep it compact and more easily RFI-shielded) -parallel port (one pin of which is for 1 PPS, which with NTP is used for timing) -Fiber ethernet connection -capable of running linux. The currently supported version at the observatory is RedHat 7.2 -At least 2 PCI slots for DIO board and GPIB card (not a GB reqmt) -2 cpu's running at 1 GHz or faster (not a GB reqmt) -1 GB Ram (not a GB reqmt) -hard drive (not a GB reqmt) *I note: The currently contemplated packaging of the DAQ computer is at odds (I think) with our recommendation, but packaging may be driven by the "Big Electronics Box" (see below). DAQ Electronics =============== -The DAQ electronics are accomadated in a) the "cold electronics": a small package which lives in close proximity to the detectors (within the dewar). b) "the tower": a small box that in the current design sits *on the front of the dewar* (ie, looking at the 2ndry). This is nominally RF-tight but testing will be required, as MHz signals on copper go in and out of this box. c) "The Big Electronics Box": a 19'' rackmount unit. Copper goes in and 8 fibers come out. Currently powered by 2 12 V batteries; shares ground with the tower/cryostat. d) the DAQ computer: described above. -ACTIONS: *Penn is coordinating with Goddard on acquiring these components. *There is much work to be done on coordinating and procuring cabling; this is a top priority issue for GSFC & Penn. *NRAO needs to work with Penn on packaging and RFI. DETECTORS ========= -The GSFC team has made 3x3 mechanical test arrays. The 100 um bars between detectors are too narrow and break. Dominic will try 300 um and 500 um spacings (rather than shrinking the detector area, which may lose sensitivity). With the current optics this will place the beams > 4.2'' apart on the sky, which would then be well over Nyquist (3.44'' at the top of the band; 3.89'' at the 82-95 GHz band center). [see also optics below] -Simon stated the following desiderata for the detectors, which we all agreed on: * time constant < 5 ms (faster than earlier spec, to allow for scanning closer to the GBT limit) * detector noise < 1.2e-5 pW rtsec (consistent with the past nominal goal of 1e-17 watts rtsec noise) * saturation power > 8 pw (which corresponds to his estimate of the loading at tau_90=0.15, ZA=65 degrees, with a 30% margin) -Simon and Dominic's loading calculations disagree, and they will work to resolve this. -In GSFC's test setup, there is a white noise floor a factor of ~ a few below the detector noise (for "typical" loading levels?); this also pushes us towards a faster time constant as the white noise floor will limit the degree to which you can deconvolve the sky signal. OTHER PENN PROGRESS =================== -FILTERS: Mark Devlin visited Cardiff and they will soon provide an 82-95 GHz filter. This is intended for lab tests. It may eventually find its way into the real receiver as a lowpass filter over the array (there will also be a bandpass filter in the optical path further from the array). *ACTIONS: BSM suggests a conservative strategy (eg 86-94 GHz) for the first "production" filters. -CRYOGENICS: as previously reported, progress has been good with the test dewar, pulse tube cooler, and He4 fridge (in particular the He4 fridge has been made to work in the test dewar with the PTC). There were some difficulties with the He3 fridge (leaks and scoring) but these have mostly been resolved. -OPTICS: as previously reported, there are two working designs. Simon has not found a vendor who says they can put an anti-reflective coating on the quartz lenses, and lab tests have proven difficult; however the Cardiff group claims to have done this. If this pans out, and we can decide on the actual optics (pixel spacing), we can proceed with the optics. 08jan03: BSM & SDicker spoke on the phone. We agreed that given the primary observing strategies for the Penn Array, it is ok to be somewhat more than 0.5 flambda apart on the sky, especially as you acquire more photons this way (get more of the airy spot on a detector). Dicker will use the 3.25 mm pitch detector design (300 um bars) as a basis for a 0.5 to 0.6 flambda design and proceed. This decision should be reviewed by a wider audience. -TIMELINE: (1 quarter slippage: started 1 quarter late, but we're not being billed for that) *Cryogenics: on schedule *Optics: nominally ready to go; can still make Q1 2003 start of optics construction *Detectors: late start but good progress. Production start 1st half 2003? (but initial GC shows v. long production run?) *Cold electronics: in good shape (target: done Q2 2003). *Warm electronics: we need to work to wrap up interface/DAQ by end of Q1 2003. *Receiver: overall design done-- mechanical interface with GBT needs to be done ASAP to stay on schedule (Q1 2003) *CDR-> sched 12/13+1 Quarter --> We need to plan for this. DATA ANALYSIS ============= -BSM reported on progress locally (simulation pipeline and our thoughts on the approach to analysis). We agreed that the analysis is going to be computationally challenging, and on the importance of trying to bring in some extra expertise to this. Devlin would first like to talk to some of the people at Penn. We reiterated the importance of producing a pipeline usable by the average astronomer. -We agreed that we must not let the quick-look pipeline slip through the cracks, and this must be ready when we are commissioning the instrument. This could be based on Chapin & Hughes IDL pipeline, or a similar implementation in AIPS++; or BoA (MPIfR generic bolometer analysis package under development).