MMA Imaging and Calibration Group

Minutes for meeting Mon, 2 August 1999 at 4pm EDT.

Date: 2 August 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

Minutes

--------

MAC Meeting - All

Carlstrom pointed out some of the advantages of an ALMA Q-band system:

  • ALMA has half the VLA collecting area but twice the aperture efficiency, making it more sensitive than the VLA for point sources and much faster for surveys, owing to its broader beam.
  • Sensitivity should be about 3 microK per hour for continuum, making it excellent for distant cluster objects, etc.
  • ALMA's southern hemisphere site complements the VLA's northern site
  • The cost should be only $50K/telescope with compressor for a separate dewar, only 1% of the cost of the project.
  • ALMA's design offers many shorter baselines than the VLA, a distinct imaging advantage for extended objects.

    Richard Hills had some interesting comments about the sensitivity of the 183 GHz WVRs:

    You might like to pass this on to others for comment. This is deliberately somewhat one-sided because there are strong advocates elsewhere for other approaches - e.g. putting full reliance on fast-switching and using cooled receivers for radiometry. I am sure that an agreed overall strategy which allows for the best use of the various techniques will eventually emerge but clearly information on what are the real cost drivers is needed before we can get the debate onto a really rational footing.

    Just to follow up a few points from today's telecon:

    1) Dispersion in the refractive effect of water at submm wavelengths:

    This is due to the strong water absorption lines: there is a component of the refractive index which is proportional to the frequency derivative of the absorption. (I think this is the Kramers-Kronig relation - spelling?) This means the refractive index goes up on one side of the line and down on the other. The effect is not large except close to the centres of the very strong lines where we can't observe anyway. I believe that it will be possible to calculate it with quite adaquate accuracy by means of a simple model.

    Note that even if we use fast-switching, we would also need to apply this correction to the phase measured on the calibrator unless we observed the calibrator at the same frequency as the source (which will generally be difficult when we are observing at submm wavelengths). If we calibrate at a different frequency we would need to have a way of separating the atmospheric and instrumental phase errors, because the correction for dispersion should of course only be applied to the atmospheric component.

    2) Sensitivity

    Here are the rough numbers for the 183 GHz case:

    We use a double-sideband mixer with the LO frequency at the centre of the line, so the IF frequency determines how far out in the wings you are. You get the maximum sensitivity by choosing the points where tau = 1, i.e. where the apparent brightness is about 170 K. This is at about 0.5 GHz for the very driest conditions (0.6 mm of water), 1.5 GHz at w = 1 mm and 4 GHz at w = 4mm. The sensitivity numbers are then about 10 microns of precipitable water (i.e. about 63 microns of path) per K of brightness temperature for w = 1mm and 250 microns of path per K at w = 4mm. Thus, if you can measure the atmospheric temperature fluctuations to 0.3 K, you get errors of 20 microns when w < 1mm - i.e. 22 degrees of phase at 900 GHz - and if you are still observing when w = 4mm (!) you get the same phase accuracy at 225 GHz.

    By comparison a mixer with Trec = 2000 K (DSB) and 400 MHz of IF bandwidth should give an rms fluctuation of 0.2 K with a 1 second integration time. This includes the factor of 2 for Dicke switching. The ideal would be to switch against a load at about 170 K, so that gain instabilities are removed, or perhaps alternate between two loads, one at 270 K and one at 70 K. A continuous comparison receiver could also be used.

    There is of course more to it than this - we need to worry about pressure and temperature terms and the effects of clouds - so I am sure that we will want to use a number of IF channels. The present experiments - Martina Wiedner's devices on JCMT-CSO and the ones we are using on Chajnantor - are being done with three channels at 1.2, 4 and 8 GHz. The point of the above is to demonstrate that a simple uncooled radiometer is just about good enough: we do not need to devote one of the precious slots in the dewar to the radiometer. The 2000 K is achieved with a mixer working on the second harmonic of the LO - we would presumably do better with one driven at 183 GHz directly.

    3) Fast-switching specification

    I continue to fret about this because I think the contractors will find it difficult, and therefore expensive, to meet this spec. The comment today was to the effect that we get the switching for free from the fast raster scanning requirement. I am sorry that I haven't been paying attention, but do we have a specification for the fast rastering and if so how was it derived?

    I agree that unless we install chopping secondaries on all the antennas we will need to be able to waggle them around quite fast to obtain continuum total power data. I would point out however that the requirements for making a raster scan - particularly where you simply record the actual pointing positions along with the data and regrid to form a map later - are very different from the ones for fast switching, where you have to accelerate as hard as you can, then deccelerate again and settle on the source to high accuracy, all in a very short time. (Is it 1.5 seconds?)

    Even with the water radiometers it will of course still be necessary to switch to calibration sources quite often. The point is that even if the radiometry does not work as well as some of us expect, it must make it possible to soften the switching specification significantly.

    I'm not at all sure I have the origin of the total power mapping constraint on the antenna speed straight either. Let's discuss this.

    --------

    Configurations - Helfer, Radford

    See my notes emailed separately. The strawman configuration which Tamara and Simon have devised since the Tucson meeting will be discussed as a prelude to discussion at the URSI meeting next month.

    Simon and Tamara will compare the 64 element optimizations Leonia has performed and match these to existing site maps to produce a strawperson layout similar to that produced for the MMA Reference Design. The results were plotted for that 36 10m antenna design. The circles at the antenna positions are not to scale, but provide leeway for topography, etc.. Leonia's optimizations for the B, C and D arrays should be easy to mate to the site, but those for the A array and the 10km array will be more difficult, owing to incompleteness in the terrain maps in some regions.

    The question of whether it is operationally desirable to have a crew committed to moving antennas always will be discussed at the 9 Aug DH meeting.

    --------

    WVR - Wootten, All

    See my notes emailed separately and since updated. Someone has changed the decision date to 1 August. I propose the following recommendation:

    A review of the state of water vapor radiometry was held in Tucson on 7 June 1999. The following facts arose from this meeting:

    1) Continuum water vapor radiometry has been employed at IRAM for some years to correct data for the excess path length caused by differing amounts of water vapor along the line of sight to astronomical objects. Simultaneous 1.3mm and 3mm observations are used for this, resulting in improbed images about 70% of the time during which no liquid water vapor is present. A similar system is under development at Nobeyama and the technique has been explored at BIMA.

    2) Water vapor radiometry in the 22 GHz line has been employed with good results using a system developed at OVRO, primarily by Dave Woody. Path length variations are corrected down to about 200 microns in virtually all weather, except for rain or snow. Similar systems are being constructed for use at the VLA, the AT, IRAM, and BIMA. They have proven stable, the radiometry shows good correlation with phase, coherence correction is straighforward and previously unusable data has been recovered. Residual delay errors are at the 100-200 micron level at OVRO. The system under development is expected to improve this to below 50 microns.

    3) Water vapor radiometry in the 183 GHz line has been demonstrated to provide good correction at Mauna Kea by Wiedner, Hills, Yun and others (see ALMA Memo No. 252 and Wiedner's thesis). Path length corrections better than 60 microns were demonstrated under conditions which apply at Chajnantor 80% of the time. This is within a factor of a few of the ALMA goal of achieving 0.15 radian phase calibration at 230 GHz. Substantial improvement should be achievable on the performance of the receiver which would allow ALMA realization of the goal.

    We conclude that water vapor radiometry in the 183 GHz water line will provide a most promising technique for correcting interferometric data for phase noise due to water vapor path fluctuations in the troposplhere. We recommend that construction of a water vapor radiometry system operating at 183 GHz for this purpose become a part of the ALMA project. The systems currently stationed at Chajnantor should be further improved to develop into the final more sensitive system.

    --------

    URSI - Wootten

    Update on URSI: There will be a 17 Aug meeting with the European DHs at URSI. Kawabe has circulated a tentative agenda for the 14 August workshop. Mangum will speak on amplitude calibration of the MMA. Yun will speak on the strawperson configuration. Radford will give a short presentation on future site testing. Wootten suggested other topics to be covered might include:

  • Criteria for simulating imaging performance of configurations.
  • Problems and techniques in combination of single antenna and interferometric data into ALMA images.

    --------

    Data Rate - Glendenning

    The data rate figure I've been carrying in the project book is 1MB/s average, 10MB/s max sustained. This comes from the Scott et. al. memo (#164) which in turn apparently came from a Tucson science meeting (40 antennas). Mark Holdaway should know about this number. Rupen in a later memo comes up with a science case for 100's of MB/s (OTF synthesis surveys). Note that the NRAO correlator design is capable of dumping data from the LTA at a rate of 1GB/s.

    What is the science-driven rate?

    --------

    ALMA Science Meeting

    I will update everyone on the status of this event.

    Estimate of number of attendees:

    32 members of SOC (includes MAC members)

    5 members of LOC (not included above)

    28 non-MAC speakers

    65 WWW registrants (some overlap with above)

    Estimate about 120-130 registered attendees. Auditorium holds 400.

    Estimate ~ 50 poster papers. Posters in reception hall, 57 x 29 feet.

    Kate has posted these to the WWW.

    --------

    M/C technical document

  • Please see the document at Brooks' homepage.

    --------

    Total Power Mapping --'Holdaway Holdover' Memo - Butler

  • The audit committee had a lot to discuss on Holdaway's Protomemo (postscript) on total power mapping.

    Progress is slow owing to a lack of suitable simulation software. We discussed a) breaking the memo up into parts describing 1) Noise in e.g. EKH chopped and restored maps and 2) Noise in OTF maps. A qualitative description with numbers can be prepared for the project book, which aleady has 1/f noise constraints in a table.

    Pending resolution of the problem of a lack of suitable simulation software, I propose that a table of target total power sensitivities be made similar to that which we have for interferometric sensitivities. For specificity, I propose that these be made for a 5' x 5' region at all bands under median atmospheric conditions.

    Correlator -- Webber

    Given the desire to achieve up to 10 kHz frequency resolution in the test correlator, Ray has re-examined the difficulty of providing various modes and concluded that the following are feasibly without excessive effort. However, he asks that we choose wisely and minimize the number of modes which actually needs to be written. The following table gives the number of frequency channels/resolution in kHz for the feasible bandwidths and polarization products:

  • 2 polarizations, 2 polarizations, 1-pol
  • Bandwidth with cross-products no cross-products
  • --------- ------------------- ----------------- ------------
  • 800 MHz 512/1563 kHz 1024/781 kHz 2048/391 kHz
  • 400 1024/781 2048/391 4096/195
  • 200 2048/195 4096/98 8192/49
  • 100 4096/49 8192/24 16384/12
  • 50 8192/12 16384/6 32768/3

    This is a total of 15 modes. Please determine which are really needed. Once a bandwidth is chosen, the other polarization modes are not too much extra effort; however, changing bandwidth requires starting nearly from scratch each time. For example, if we could live with just the 800 and 100 MHz bandwidths, we would get both the requested extremes--max bandwidth for looking at passband ripple effect, and frequency resolution of 12 kHz in a single polarization. Another choice could be 800, 200, and 50, which could achieve overkill in resolution and would be 50% more programming effort than just 800 and 100. What do you think?

    Darrel proposes:

    My inclination is to say that just 2 of these bandwidth modes will be fine - 800 MHz and 50 MHz, or possible 800 MHz and 100 MHz - but each with the 3 options of polarization. Remember, this is just for the antenna and system evaluation in Socorro, and has nothing to do with what might be installed in Chile.

    ------------

    Review of C&I Problems -- Wootten

    Bob asked for a review of the status of current problems at the DH meeting. C&I review of current problems, issues

    Major issues are definition of the configuration, radiometry as a calibration tool, and imaging.

    Kogan and Helfer have specified a strawperson configuration set, discussed at the review on 9 June in Tucson. They are in the process of mating the three smaller configurations to the site. This process in larger configurations is limited somewhat by lack of accurate topographic maps near the Bolivian border; Radford is addressing this problem. The strawperson configuration is then confronted with astronomical images to test its qualities.

    Problems include the software packages for the latter test (Helfer uses miriad, Kogan AIPS, Viallefond uses Gypsy, Holdaway used SDE. AIPS++ has agreed to provide a standard package in a several month timescale. C&I has discussed qualities a 'good' imaging configuration should have, with some input from the MAC.

    European array designs have supported spiral designs, which we do not feel offer enough short spacings; we have emphasized dual ring designs to provide these.

    European ideas have included the continually reconfigured array, in which antennas are moved continuously. This poses an operations question which will be considered at the 9 Aug DH meeting: it requires a crew to be present at the high site devoted to this each working day.

    Simulations so far suggest that even the optimized 64x12m array recovers (interferometrically) only about 20% of the flux from a mosaiced image.

    Helfer will leave the ALMA/US project next week.

    A review of water vapor radiometry on 7 June resulted in a recommendation that the array employ 183 GHz water vapor line radiometry to correct phases for the varying vapor content along the path to the source.

    Problems include:

    The state of the art at 183 GHz is in its infancy. Twin receivers at the ends of the phase monitoring interferometer are in place at Chajnantor,

    and are being improved.

    Wilson is building a 183 GHz radiometer to work at the SMA in conjunction with the CSO and JCMT.

    The character and location of the receiver and the definition of its backend are still under investigation.

    Imaging techniques were reviewed at a workshop at the AOC a few weeks ago.

    Problems include:

    Total flux recovery into an accurate image still appears difficult with the 64x12m array alone.

    Total power models lack the robustness of interferometric models of array performance, in part owing to the lack of simulation software.

    Measurements of 1/f noise on prototype receivers, or on any other receiver, are difficult and lacking for most relevant cases.

    Frequency coverage:

    Problems include:

    US Scientists favor a 7mm band, reiterating this view at the MAC meeting last week, while European scientists are unconvinced.

    ----------------------------------------------------------------------

    --------

    Action Items 2 August 99

    DECISION: Configurations--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: MAC meeting 25 August 1999 at noon. ------

    --------

    Travel

    If your travel isn't on here you have not sent me a travel authorization.

    ------

    T. Helfer:

    A. Wootten: 13-19 Aug URSI.

    J. Mangum:

    M. Yun:

    B. Butler:

    S. Radford: