Agenda for meeting Mon, 10 May 1999 at 4pm EDT.
Date: 10 May 1999
Time: 4:00 pm EDT (2:00 pm Socorro, 1:00 pm Tucson)
Phone: (804)296-7082 (CV SoundStation Premier Conference phone).
Past minutes, etc on MMA Imaging and Calibration Division Page
' The question came up in one of the receiver engineering meetings on how fine a step would be acceptable for the first local oscillator of the MMA. In one scheme, by the time you get to 800 GHz, it could be as coarse as 1600 MHz.
Are there limits on how coarse it can be? As you know, at the 12 M we do all our Doppler tracking on the first local oscillator, with a fine step of a kHz or less. In principle, provided the first local oscillator step is less than the IF bandwidth, maybe very coarse steps are ok. Do you have any thoughts on this?
I think 1600 MHz for a LO step is probably too coarse, but could easily be reduced by a factor of 2 or 3 or so. But if a very fine step (e.g. a few MHz or less) is needed, it might mean a fundamental change in the current hardware local oscillator generation plan.
' After some discussion we realized that we didn't have sufficient information to determine if the coarse steps proposed would allow fully flexible tuning across the band. John Lugten offered to pursue this with Darrel. Bryan Butler will attend the LO PDR on 17 May for followup.
Pisano writes: ' Brian Glendenning mentioned that I ask you about integration rates for the MMA correlator. He mentioned that the MMA science working group had specified a minimum integration time of 10 milliseconds.
According to M. Rupen's MMA Memo # 192 regarding short integration times, he summarizes:
- OFT power mapping requires dump times of a few milliseconds for auto-correlation spectra
- Large area surveys need dump rates of 10's to 100 milliseconds of both auto- & cross-correlation spectral channels
Where does this 10 millisecond number fit in? The reason that I ask is that the current design for the correlator is 1 ms dumps of the auto-correlation channels and 16-ms dumps of the cross-correlation channels.
' Al noted that we had set a specification. Brian noted that it should have propagated through the system but clearly had not. The important thing is that it does so so that the system architecture is not designed so that for example 10ms dumps cannot be achieved in the future.
Division Heads meeting stuff. MOC meeting plans, MMA Brochure
MOC: The viewgraphs which Radford will present for me are available in Adobe Acrobat format . See also Radford's site summary viewgraph and Simon's Gantt chart viewgraph for Division 11.
Sargent, Boss, Scoville, Crutcher, Spergel, Millar, Jewitt, Blain and Evans have accepted our invitation to speak. Hartmann will be at a conference in the Canaries but suggested his co-worker Nuria. Steve Beckwith is involved in the ST refurbishment mission and cannot attend. I suggest Koerner as a replacement--other suggestions before I pass this along to the SOC? Ishiguro has also nominated Ohashi to give a talk. Hollenbach has declined, owing to workload associated with the McKee report, and also Glassgold's 70th birthday party. Neufeld as a replacement?
Pat is getting bids for production of an MMA brochure whose form was copped from the SIRTF brochure NASA has.
Review of emails past week and figuring out where we are (via GPS II)?
Tamara pointed out the current status of the configuration studies:
the four nominal arrays are something like:
(1) D array: Filled array with d=150m.
(2) C array: Donut array with dmax/dmin=0.5; dmax=420m, dmin=210m.
(3) B array: Donut array with dmax/dmin=.75; dmax=1110m, dmin=830m.
(4) A array: Donut array with dmax/dmin=0.9; dmax=3000m, dmin=2700m.
and the 10km/20km arrays would be some kind of ringlike arrays, subject to the topography constraints.
Helfer has been modeling observations of M51 with existing arrays, using a somewhat similar model source. 'I've been running some simulations that you might like to see -- they are simulated observations of a spiral that has varying width from 10"-90", observed with a 7-field mosaic from BIMA (one config), OVRO (two configs), and IRAM (2 configs). Unlike Mark's simulations, I don't include total power for any of the observations. OVRO and especially IRAM show significant errors even for arms as narrow as 10-20", whereas BIMA does relatively very well -- for arm widths of 50", BIMA recovers 80% of the peak flux of the original source. Although some of the difference might be due to the poorer uv sampling, it's clear that a lot of the poor performance comes from the size of the central hole. In current ALMA thinking, total power data will be needed for essentially every project. My simulations suggest that if you can get short enough baselines, there should be many projects where total power data is not necessary. Of course that would really argue for small dishes (since BIMA's 6m did so much better than OVRO's 10m).'
Al suggested that these studies might include the ALMA configurations, which are designed to have closer spacings than OVRO or IRAM and may give somewhat different results. Tamara agreed to look into this. Mark brought up the question of the total power data, which should be included in realistic simulations. There was discussion on including this general topic at the Toronto URSI meeting, at which Peter Napier will chair a special workshop devoted to millimeter arrays. Bryan reports that Napier intends to have a few invited talks in the morning, then split up into working groups for the afternoon. This will be planned further in Tucson in June, and it may be well to discuss the ability of ALMA to recover short spacing information. --------
Min's simulation of M51 Long, long ago will be discussed when he returns from far, far away. Min has returned and will be proceeding with these simulations.
7 June it is? 4-6pm? Kurtz will be asked to invite Richer and Guilloteau, at least. Others who might be interested include Brown, Butler, D'Addario, Emerson, Greisen, Hasegawa, Helfer, Ho, Holdaway, Ishiguro, Jewell, Kogan, Liszt, Maddalena, Mangum, Morita, Mundy, Napier, Plambeck, Radford, Welch, Wiedner, Woody, Yun and me. That is quite a few.
We should also review amplitude calibration. Bock will also be present. See Bock's photos of the amplitude calibration device at BIMA.
Andy Harris sent an interesting overview of the WVR project at the MDC:
In overview, the project is a BIMA-OVRO collaboration to understand how to reliably measure phase shifts due to fluctuating atmospheric path length differences, and apply them to correct astronomical maps. I should note that the funding we receive from the MDC covers only a fraction of the work I'm describing, but I'd like to cover what we're doing in general rather than sticking to the fraction that we've promised the MDC. The overall project's main goals are to:
* Use spectroscopy of the 22 GHz water line to measure pathlength fluctuations to to changes in water vapor. We're concentrating on the 22 GHz line because we need to do real end-to-end tests on actual data, but our theoretical work is nearly equally balanced betweeen the 22 and 183 GHz lines.
* Build enough equipment to test our theoretical models and correction schemes in the field.
* Use the field measurements to improve astronomical data. This requires:
- Finding a robust scaling between phase monitor data and millimeter wave phase. This scaling must be insensitive to changes in the water vapor's altitude, air temperature, and to very small water droplets that are often found in, for instance, haze. The system needs to work under many weather conditions, working well under the best conditions but also making 3 mm observations possible in marginal weather.
- Developing methods to transfer the phase of the calibrator across the sky to the phase of the source -- we want to "phase up" on the calibrator, then measure the change in atmospheric pathlength as we move to the source.
- Find appropriate ways to incorporate the phase data into the observations. Some questions here are: On what time scale do we need to make the measurements? How do we synchronize the phase monitor data with astronomical data files? Can we find a way to correct the LO phase, and integrate the astronomical data for long periods of time, or are short integrations better, each with its own software phase correction?
So where are we?
* We've taken and extended the atmospheric modeling code Dick Plambeck wrote, which is based on analytic approximations to the atmospheric absorption. We compared it with Ed Sutton's far more complete model, and decided we would do our design work with the simpler model -- it seems adequate for exploratory work. We may need to go back to Ed's model as we try to convert measurements into phase.
* Lee Mundy and Peter Teuben have taken the modified code and packaged it in a form that makes it easy to use and interface with the MIRIAD package. This has let them start on modeling more complicated situations -- fluctuating numbers of layers, and so forth -- to explore at least the theoretical limits of various systems.
* Based on the early modeling and a certain degree of common sense, we believe that a system with a number of spectral channels is important, whatever the wavelength. The simple way to understand this is that we need to account for lines with slightly different shapes from different altitudes, and separate the line emission from the atmospheric quasi-continuum and spillover.
* We target the 22 GHz water line because it is always optically thin, and its width is small enough (about 8 GHz) that we can measure points across the line and somewhat into the quasi-continuum with existing equipment. I'll come to the equipment in a moment. The penalties are that it's weaker and harder to measure and that the optics and beams aren't well matched to the mm-wave equipment.
* The 183 GHz line has a different set of problems which I believe are more serious. This core of this line is optically thick essentially all the time at any site, so one has to work in the thinner line wings. Spectroscopists will instantly recognize that accurately measuring column densities well out in the line wings is a difficult proposition; it requires a high degree of knowledge of the excitation conditions in the line. Our concern is that the 183 GHz line is not a robust estimator unless the atmosphere is basically static. There will certainly be a good correlation between emission in a wing channel and phase, but the conversion of brightness temperature to microns of pathlength will be variable, depending on the altiude of the water (the line wings are produced by pressure broadening), number of layers of water at different altitudes, temperature, and so on. The wing emission is also enormously broad, and measuring the line shape to try to make fits to the data requires continuum-measurement performance over tens of GHz -- with no baseline structure to confuse the emission. Even if line wing excitation information from the lineshape alone is sufficient, building the receivers that can do this is going to be very tough -- no one has ever built a receiver with total power stability performance close to what's required. The technical tradeoff is mK stability over 8 GHz vs. tens of mK stability over 30 GHz -- I think the former's more reasonable.
* With these considerations tempered by ALMA's engineering difficulties of getting cryogenic 22 GHz systems into each telescope, we've also spent some time thinking about using a single 22 GHz system, and possibly a 60 GHz system, to probe the atmosphere. If one can do that well enough well enough to measure the average atmospheric excitation independently, there's some hope of using that information to interpret the data from simpler 183 GHz systems in each telescope.
* Although we haven't gotten too far into this part (we're concentrating more on trying to get real measurements at this stage!), it doesn't look too easy to "invert" the 22 GHz spectrum to get at the detailed atmospheric structure. But we'll see -- this is an important part of this summer's measurement campaign.
* Marc Pound has started archiving the BIMA satellite phase-monitor data to provide a data base of real atmospheric phase measurements. I've begun some work to try using this for getting short-term statistical behaviour to make "smart" software predictor-corrector filters that would do better than individual measurements. So far, this doesn't look too promising: at the scale I'm interested, the fluctuations are not statistically stationary. But this is really preliminary, as I haven't had time to really work on a lot of data -- I have to find a student to work on this, I think.
* We're in the late stages of building up a single cryogenic, 16 channel spectrometer for 22 GHz line measurements at OVRO this summer. David Woody has designed the optics and cryostat, which must be pretty much done by now -- they were well underway when I visited OVRO in late March. The cryostat will house one of NRAO's cryogenic K-band amplifers. This should give us a system temperature low enough that we can surpass our target of measuring pathlenghts to 35 microns rms in 10 seconds. We're in lab tests of the backend spectrometer at Maryland: this is a compact version of a 16-channel WASP spectrometer board. Once we finish with the readout electronics testing (next week?) we will verify that its band is the same or larger than WASP's 750-4000 MHz coverage. We have the components to amplify the 22 GHz line and then mix it with an LO at line center: the spectrometer channels then cover both sides of the folded line out to pretty low levels. Our modeling indicates we can separate the line from the quasi-continuum and a linear slope from spillover -- we can probably get by with fewer than the 16 lags we have (14 equivalent frequency channels), but the incremental cost is small (and we're prepared to discover that our models aren't all that correct!).
* Our plan is to merge the front and back ends at OVRO, with a target date of sometime in June, although we may slip into July. David plans to move antennas to favorable positions for testing before the summer shutdown.
* The goal for the summer's measurements is to see that our new system works well -- that it is stable and tracks the water vapor well, as far as we can tell with only one many-channel system. We'll be looking carefully at the spectra to see how well we can separate different layers and measure their heights -- we'd like to find where the seeing layer or layers is or are. We also believe that we can improve the results from David's 3-channel system by measuring the actual lineshape with the many-channel system, then using that information to break some of the linewidth vs. continuum level degeneracy in David's measurement.
* We'll use these data and opportunity to continue OVRO's work on folding phase correction into astronomical measurements. In addition to Steve Scott's effort on getting our data into the OVRO system, we're following John Carpenter and Nick Scoville on understanding how to couple the phase measurements with the OVRO astronomical data.
* Once we've convinced ourselves that the new system is what we want, or close enough, we'll go ahead and build a second system this fall. The bottleneck on a second system is the first amplifier in the system. Both David and I have been in touch with NRAO about this, and we'll see what happens.
* Past that, I hope we can add more systems. Dick Plambeck has laid out the optics and found mechanical refrigerators so we can add to this effort using the BIMA system. We're raising the hardware costs (and much of the personnel costs) from operating budgets and other semi-related projects as it is -- there's no way to go to any kind of production without specific funding from somewhere. I don't know how much other sites have spent, but Maryland's spent about $130k on this project so far, including hiring a postdoc specifically for this project, with $50k coming in from the MDC development funds to date.
UPCOMING REVIEW: On 7 June (?) 1999 we review results from the OVRO and BIMA phase correction systems.
DECISION: Congfigurations--where are we? DECISION: 183 GHz or 22 GHz phase correction?
DECISION: Is a nutating secondary necessary?
DECISION: What is the effect of 1/f noise in the HEMT amplifiers of SIS receivers upon our ability to combine total power and interferometric images into a faithful representation of the sky?
MEETINGS: IF PDR 17 May MEETINGS: MOC meeting 12-13 May 1999, Tucson. MEETINGS: MAC meeting 19 May 1999 at noon. ------
A. Wootten: 28 May - 3 Jun AAS 5-9 Jun I99 9 - 18 Jun CSO 19 - 30 Jun Japan.
M. Yun: AtCSO this week.