Agenda/Minutes for meeting Tuesday, 23 November at 4:00 pm EST.
Date: 23 November 2004
Time: 4:00 pm EST ( 2:00 pm Socorro, 2:00 pm Tucson)
Phone: Phone: (434)296-7082 (CV SoundStation
Premier Conference phone 3rd floor). No Video planned.
Past minutes, etc on MMA Imaging and Calibration Division Page
| 15 Nov
|| AMAC & ASAC Reports due to Board
| 2 December
|| ALMA Board Telecon -- ASAC Charges to be discussed.
| 14 December
|| ASAC Telecon
| 5-7 Jan 2005
|| UNSC URSI Boulder meeting Commission J Agenda
| 11 Jan 2005
|| ALMA Town Meeting, AAS San Diego; Agenda
| 27 Jan 2005
|| ALMA Board Telecon . Dates for 2005
| 24 Feb 2005
|| ALMA Board Telecon. Rebaselining to be discussed.
| 5-6 April 2005
|| ALMA Board Face-to-face meeting, Pasadena, CA.
Science IPT Statements of Work
Ground Based Radio and Submillimeter Astronomy Planning Group
NA Workshop topics 2005?
My Tuesday lunch Presentation
NRAO Algorithms Working Group Meeting
Question: What receivers should be employed for interferometer tests at the ATF. Until January 2006, only evaluation receivers rather ALMA preproduction receivers will be available. The problem is that the evaluation receivers weren't originally designed for interferometry (what were we thinking?). Thus a modification of these receivers is necessary--an extra IF conversion stage, with its LO needing phase locking--is in these receivers. Recall that there are two bands, one at 3mm with two polarizations and one at 1.3mm with one polarization in this receiver set. Do we need both bands? If only one, which one? If two, how easy should it be to switch between them?
AW proposes: *Abandon the current plan to ship the first preproduction receiver to Chile in October 2005; this is a silly plan driven by an outdated milestone. Deliver it
to the ATF, followed by the second in early 2006. This suggests that whatever we do at the ATF with the evaluation receivers will last from first fringes...say April at the earliest...until mid-winter 2005/6. So if we were to operate at 1.3mm, it would only be with advantageous weather from say October until say Feb, and we'd have one ALMA preproduction receiver during that time anyway with a second in mid-winter 2006. So my inclination would be to support 3mm interferometry with the evaluation receiver from first fringes until delivery of the second preproduction receiver, at which point we would support interferometry in all usable bands at the ATF. At that point bugs should be ironed out of most systems at 3mm and we'd be fully prepared to iron out bugs in the preproduction receivers and to pay attention to the 1.3mm band and the WVRs. Astronomical holography will be debugged at the ATF in preparation for tests on the production antennas at the AOS or OSF.
In a document called "Project: Chilean Systems Integration" item AIV_15.3.3 reportedly proposes that single-baseline interferometry should be done at the OSF. Should interferometry be supported at the OSF? If so, why? If so, when?
Mangum notes: We need to know how the production antenna behaves in many orientations. This requires all-sky (astronomical) interferometric holography. Thus a correlator is needed at the site at which this is to be done. This could be done at the AOS; a very good reason for not doing it there is that any antenna adjustments would best be done at the OSF for safety reasons (hovering close to the reflector in a cherry picker under low oxygen high winds and threat of an occasional earthquake is not a good idea). The ALMA correlator is only available at the AOS. It makes sense to use the prototype correlator for these measurements at the OSF. This suggests a schedule under which the ATF would cease operations in the summer of 2006 just before the second antenna was to arrive at the OSF so that the correlator could be shipped and installed at the OSF. In the best of all worlds, the methodology would be tested at the ATF for interferometric holography and the experts who contributed to that testing would be available at the OSF to reproduce the methods successfully employed at the ATF at the OSF.
Baseline Ripple and quantization noise
Darrel wrote: One of the things I've been working on, as a background task in the last few weeks, is a simulation of quantization noise (QN) and what its spectrum is. I've been doing this with Dick Thompson. EVLA memo 83 assumed that QN is essentially flat, but Dick and I wanted to follow up to see how good an assumption that is. The plan is to submit this as an ALMA memo - we already have a near-final draft.
The answer is that it's a fairly good, but nor perfect, assumption. However, coming out of this is the implication that, if the input passband before the ALMA (or any other) digitizer has gain ripples in it, then after correlation and renormalization to calibrate out the gain variation with frequency, you'll be left with a variation in noise (or s/n) with frequency.
So, the question is, how much variation of s/n across a spectrum can we tolerate? This sets a requirement for the backend folks. For example, if a gain ripple across the passband before digitization is to produce a s/n variation, after calibration, of no more than a factor of 2, then the original passband gain ripple must be no greater than about 10 dB, with ALMA's sampling scheme. This is probably ok for the engineering, but if the science demanded a constancy of noise as small as (say) 0.5 dB, then the original ripple needs to be smaller than 2.5 dB, which would be extremely hard to do.
I think there are already specs on passband ripple from Larry, but I don't think they've been expressed in the form of a variation of s/n across the calibrated spectral baseline before. What's acceptable? We may need to revise those engineering specs, assuming they exist. BaseRipple
The Specs are:
271 272 Gain flatness, each baseband channel (2
Effective b.w. due to anti-aliasing filter > 90% of nominal.
Max. pk-pk variation across channel due to all other components, any tuning: 6dB
This requirement is allocated 3 dB to BE and 3 dB to FE subsystems.
In Memo 452 Larry recommended 3 dB edge-edge for slope and 2.45 dB pk-pk for ripple. From TMS Section 7.3, Table 7.1, this results in loss of 0.98 and 0.98, respectively.
The adopted value of 6dB, no spec on ripple or slope, gives something like 0.93 I think. The 2% loss from slope and 2% from ripple recommended has been compromised into
a single 7% loss. Since each 1.6% loss is equivalent to one antenna, it is worth going to some expense to improve performance.
It is proposed to change the antenna specification, reducing the maximum acceleration in azimuth and elevation --- 10 deg s-2 instead of 18 deg s-2 in azimuth and 5 deg s-2 instead of 9 deg s-2 in elevation. It is also proposed to extend the distance to a calibrator source from 2 deg to 4 degrees in the specification of offset pointing and repeatable residual delay. This change allows the ACA array with small collecting area to find a strong calibrator and to perform calibration with sufficient accuracy.
It is proposed that the scientific needs of the ACA suggest that the ACA 12-m antennas should be optimized for the total-power mapping in the OTF mode; the ACA antennas may be rarely required to do fast switching in the same manner (cycle time as short as 10 second cycle) as the 64-element antennas, and that the ACA array is expected to spend only a small fraction of time on cross correlation/calibration with the 64-element array.
I agree with these statements and with the calculation that shows that the cost in efficiency during the period when the ACA antennas are correlated with the main array is only a few per cent. During what fraction of time will this occur? The fraction of ALMA data needing ACA data is essentially unknown; it has been argued that it will be 25%; this number was used to develop parameters of the ACA. In this case, the ACA used with a 60 element array will be in use all of the time. The fraction during which the ACA 12m antennas will be used with the main array will then be very short and the efficiency impact of the requested change will be quite small, probably less than one per cent. Let us examine a scenario in which ALMA has, for whatever reason, fewer antennas availablefor specificity let us speculate that only 50 antennas are available. In this case, during constant observing time, the ACA will be relatively over-sensitive since its specifications were developed to match a 64 antenna ALMA. The ACA could provide total power data for 40% of the projects in the main array. Or, if the fraction of projects needing ACA data were in fact 25%, the ACA antennas would be available the remainder of the time for incorporation into the main array, increasing its sensitivity by 15%. The period of time during which this would happen is still low, only 15% or so of the total available time. Even for this case, the impact on array efficiency is small, on the order of one per cent. We would concur with the analysis here, that the scientific impact of changing the specification is small. The change in distance to a calibrator source is an increase in specification, somewhat ameliorating the already small impact of the first change.
Say one wanted to image the known submm galaxy J02399 (z=2.808) with ALMA. van den Bout and Solomon proposed to image J=3-2 CO in a 0.1" beam to a
5 sigma sensitivity of 0.33 mJy/beam in four hours. The velocity resolution is
not mentioned, but with the tunable filter banks it will be possible to place spectral windows
on nearby spectral regions where emission from other lines is to be expected; alternatively one could just use a wide band. One could also
integrate simultaneously on the CS7-6 line at 90.043 GHz, the HCN 4-3 line at 93.106 Ghz, the HCO+ line
at 93.68 GHz and the H2CO
5-4 K=1 line at 92.376 GHz at no extra cost. While these lines
would not likely be seen at the noise level in a 0.1" bin, the uv plane at 3km is completely
sampled. One could degrade resolution by smoothing and get to sensitivities of
2.5 microJy in a 1" beam. Even if the emission in these lines is 100x weaker than CO, one would
get a usable measurement of all of them simultaneously; interpretation would be greatly informed
by the resolved CO image.
Dusty and Molecular Universe 27-29 October 2004, Paris
IAU Symposium 227 Massive Star Birth: A Crossroads of Astrophysics May 16-20 2005, Acireale, Italy
Astrochemistry throughout the Universe: Recent Successes and Current Challenges 2005 August 29 - September 2; Asilomar, California
Protostars and Planets V 24 - 28 October 2005 Hilton Waikoloa Village, The Big Island, Hawaii
URSI General Assembly 23-29 October 2005; New Delhi, India; "Mm/submm Techniques and Science" session 25-26 Oct.
- 18 Nov 2004