Calibration Examples Documents

TIP Last Update: JeffMangum - 04 Jan 2008


Contents


Calibration Example Documents

Document Author(s) Version
2007/08/02
2006/08/28
2007/08/14
2007/08/02
2007/08/02
2006/07/18
2007/08/14
2007/11/01
2006/08/29
2007/08/06

  • Green = Complete
  • Blue = Reviewed / In Revision
  • Red = Draft

PDFLatex template for producing the above documents.


Calibration Examples Document Review Minutes


Comments on Example Documents

Comments by BaltasarVilaVilaro - 12 May 2006
  • General: Most documents below still contain 3% and 5% as relative calibration numbers.
  • General: Several documents do not have any references at all...some should be added to lead the readers.

Amplitude

See also comments from the Amplitude Calibration Example Review.

Comments by JeffMangum - 22 Dec 2005

Perhaps I am biased, but this is the model for what these examples documents should be. Only a few minor changes needed:
  • Goals: For clarity, should note that point 1 is relative calibration, while points 2 and 3 are absolute calibration. This document only currently deals with point 1.
  • Goals: To provide a solution to the absolute calibration problem, might want to add the CARMA (Gibson and Welch) absolute gain calibration scheme to this document.

Comments by BaltasarVilaVilaro - 12 May 2006
  • Absolute Calibration: I like Gibson and Welch's solution (quite elegant), but I still have some worries on the technical feasibility that perhaps should be dealt with in a Telcon. Specifically:
    • Number of Required "standard" horns (a single one will not do).
    • Targets: Some calculations on our side suggest that only a few planets could be used in terms of the fluxes, but most end up being resolved, which adds several complications to the measurements.
    • Required RX: Our calculations suggest that a cooled RX will be needed. This poses some problems with the RX locations, etc.
    • Horn Gain Measurements: Will there be a need for a calibration device for this?.
    • LO distribution and Backend lines: How will both be implemented?.
    • Other possibilities: Is the Horn method better than say, a dedicated small antenna?.


Phase

See also comments from the Phase Calibration Example Review.

Comments by JeffMangum - 21 Dec 2005
  • Goals: I think that the phase calibration spec, as spelled-out in the calibration requirements, should be stated here.
  • Goals: Add to this discussion the techniques (WVR, FS, etc.) used to meet this requirement.
  • The general musings about the phase calibration spec should be moved to an "issues" appendix item (other example documents have used an "issues" section to point to possible problem areas).
  • Section 2 (Calibration Strategies): I would move the last paragraph of this section to your "issues" appendix. Draws more attention to it.
  • Section 3: Point out that all of these phase calibrator measurements made using something other than the target observing band will be done without reposition of the subreflector (to maximize gain on the target observing band).
    • Added to Sections 4 (Calibration Steps) (JeffMangum - 16 May 2006).
  • Sections 4.x: The actual measurement steps should be spelled-out in these subsections. Use the Amplitude and Flux Calibration Examples document as a template.
  • Section 4.2.7: Does this section belong in your "issues" appendix?
  • Section 4.2.8: Another one for the "issues" appendix?
  • The references are just a placeholder and don't relate to the text (formatting issue caused by JeffMangum).

Comments by EdFomalont - 08 Feb 2006 (with rebuttal in italic by MarkHoldaway)
  • I agree that we should 'calibrate the phase as well as we can' is the best specification.
  • The phase calibration strategies for weak, moderate and strong sources are touched on in the write-up, but I think they impact quite a bit on the strategy of phase references, and perhaps this discussion should be put near the beginning of the memo. The way I look at it is the following. Given a phase calibration strategy and a quick check of the expected residual phase fluctuations, you can estimate the expected dynamic range for a observation run over a certain period. For weak sources (say <10-sigma peak brightness), an expected dynamic range limitation of 20:1 is okay. At the other extreme, strong sources which have sufficient SNR to determine phase solutions over one or two seconds, in principle, need no external phase calibration. But, if you want to know the position of the source and its flux density with reasonable accuracy, some external calibration is necessary. By the way, the 0.1% brightness noise-limited criterion is a bit strange. If the source is strong enough so that 0.1% levels will be above the noise level, then I expect you could self-calibrate the source on a few seconds times scale which gets you pretty close to SNR with little external calibration. Also the WVR phase noise versus the source phase noise is an important parameter of which way to go.
    • The "Noise-limited down to 0.1% of peak brightness" is the ALMA imaging specification. It is way too agressive. We will have images which have phase artifacts at much higher than this level. If the source is bright enough to have DR of 1000:1 in minutes, then yes, we can self cal. If it takes 24 hours of observing to reach 1000:1, then no, we cannot self cal out the atmosphere as we don't have sufficient sensitivity on the few-second time scale we need.

  • The medium strength sources, which are too weak for selfcalibration, but still can give >50:1 dynamic range images over a few hours observations, are the ones which need the most careful phase calibration strategy (this regime will probably be the most common).
    • Agreed.

  • I would change the reduction strategies to:
    • No WVR, only fast switching with cycle times of 10-60 sec.
    • WVR on 1s-times with fast switching cycle times of 10-300sec.
    • Rely on self-calibration techniques for phase corrections.
      • This works for me.

  • You might not like the last strategy above. In that case, only two are needed. The difference strategies between switching cycles of 10 sec or 300 sec with the WVR are minimal.

  • The WVR should deal with changes on at least 10-sec time scales. This is probably what you meant by 1-sec time scale.
    • WVR can deal with short time scale fluctuations, but will start to lose it on large time scale flucs -- do you mean 10s as a short time scale or a long time scale? (its a pretty "in-between" time scale, so I'm confused). The hope is that WVR will work on 1s, which for most applications (except pointing correction, though that seems to have been given up) is the shortest time scale we really need to worry about (dish crossing time).

  • Reobserving the possible half-dozen calibrators before an observation run is a good idea. Perhaps, a few dozen calibrators can be checked every day for the upcoming possible projects.
    • Caught an error in my text: Section 3.1.2: observations BELOW some critical frequency will usually be more efficiently calibrated at the target frequency.

  • Section 3.1.2: I think that the closeness of the calibrator to the target may be a more important factor than you think.
    • Well, this is a long term item of discussion between us. I do take its importance in one implicit way: the farther away the cal source, the more time it takes to get there, so far away cal sources are selected against --- maybe we need to include another term in the expression that we minimize to choose the optimal calibrator.

  • Besides systematic phase errors which depend on this angular separation, the short-term correlation of phase fluctuations of the calibrator and the target source will decrease with larger angular separation. In fact, you mention in 3.1.4, about actually using time-delay or time-advanced interpolation schemes based on the phase screen flow over the array. Yes, for nice laminar flow, this is possible. But for more turbulent flow, then the angular separation of calibrator and target may be a reasonably important factor.
    • I don't think it is so much "laminar flow" vs "turbulent flow" as it is "thin turbulent layer" vs "thick turbulent layer". But I take your point.

  • Using a calibrator source at a lower frequency than the observing frequency is still pretty scary to me. It may be necessary for high frequency ALMA calibrations, and, perhaps, it needs a special section. Non-dispersive tropospheric effects will be important,
    • Disperisive effects, rather. I suspect there will be large parts of the sub-mm windows where they are not a problem. While the SMA is going to be a pretty ratty instrument SNR-wise, it will be interesting to see if they see any dispersive effects in the sub-mm.

  • ...as is stated, but variable instrumental phase differences between the calibrator observations at the two frequencies may be caused by several effects if the secondary frequency-tying calibrator is far away and changing instrumental effects. You need this tie to a few degree accuracy. This will be a research topic during ALMA commissioning.
    • Yes. Larry seems to think it will all be OK if we can achieve the cross-band stability spec.

  • Section 3.1.3: If the phase solutions could be interpreted as a dynamic range limit over some period of time, that would be useful.
    • Yes, judging from the change in phase solutions, we can infer how well interpolation will work and what the residual phase errors will be like statistically, and can then infer dynamic range limitations.

  • Section 3.1.6: I don't see why the calibrator gain solutions cannot correct for amplitude decorrelation pretty well.
    • Well, you don't see decorrelation when you observe the calibrator, it is just a 1s observation. Then, over the 20s ON TARGET, if we don't do WVR, the phase fluctuates enough to decorrelate down to about 0.90 (typically -- though at the higher frequency bands, we might see more like 0.80). The problem is that this is highly time & baseline dependent. So, one baseline might be well interpolated/calibrated, and be at 0.97, while a minute later, it could come in at 0.85. There are two orders of business. One is to get the flux scale right, and the other is to correct for amplitude fluctuations which can limit image quality. I worked on a scheme some 2 years ago (before Tucson got the AX) that looked at the statistical properties of the gain solutions in a running boxcar to infer a statistical baseline-based amplitude correction to account for this. The results: we could correct the flux scale very accurately, to like 0.99 or 1.005, so on average we were doing the right thing. However, in detail, we were doing the wrong thing more often than not correcting at all, and the dynamic range went down. At that point, I said "Hey, maybe we really DO need WVR."

  • For very large decorrelation with closure problems, then it cannot, although the gain closure solutions will also be useful. But, when this type of decorrelation occurs, ALMA should be operating at a lower frequency anyway.
    • Well, the atmosphere is not good enough to avoid such decorrelation unless the distribution of observing frequency is highly skewed towards low frequencies. If you assume that each band is about equally likely to observe, then decorrelation is something that is built into fast switching, at all frequencies.

  • Generally, I think that short-time scale decorrelation effects the phase closure much less than the amplitude closure---although the antenna-based phase will suck (see below).
    • My studies indicated that amplitude closure no longer held either -- but that is irrelevant in 80% of the cases, as self-cal can't work on these source as they aren't bright enough in 20s.

  • Section 3.2.1: You are correct that parameters associated with fast calibrator switching will depend heavily on the properties of the WVR corrections.

  • Section 3.2.3: Calibration of WVR emission fluctuations with real phase fluctuations using a nearby strong calibrator may be an important, common, calibration. Yes, it is a ALMA research project, but perhaps one of the most important. The presence of water and ice along the line of site may be the major cause of a changing proportionality, hence theoretical models of WV and temperature profiles may not work all that well.

  • Section 3.2.4: This sentence is hard to understand, and I am not sure what you are suggesting. What you will have is a set of antenna-based phases for about 5 sec of calibration data, every 30 sec as an example, all corrected by the WVR. Is the question, how to interpolate these phases between calibrator scans to apply to the target and the interaction with WVR measurements of the target? Good question, and it depends on the longish term stability (30-s) of the WVR. Another research project?
    • Here's the deal: when doing fast switching alone, 2-pt linear interpolation helps a fair deal. The exact amount depends upon the phase structure function exponent -- so a steep power law exponent means you have lots of large scale structure and little small scale structure, or fewer squiggles in the phase time series, so interp will work better. You can get an improvement of like 1.5. The interpolation basically tries to track the phase fluctuations between the cal observations, right? IF you are using WVR, you use the WVR to track the incremental phase fluctuations between the cal observations, so using WVR AND interplating between fast switching cal observations, you in effect take those fluctuations out twice, which means you are adding the negative of the phase fluctuation -- or you end up almost back where you started. IF you use nearest - neighbor applictaion of the fast switching AND nearest neighbor application of the WVR data (ie, using increments from the last cal observation or the next cal observation) it works out.

  • Section 3.2.6 can be combined with 3.2.3
    • OK -- I don't mention that the disperisve phase needs to be calcultaed and applied online, which it does. So these can go together.

  • Section 3.2.7: I am a little confused about various sampling times that are planned for ALMA. I thought that the individual frequency channels will be normally average over 10 sec or longer, with a wide-band signal sampled on a 1-sec time scale. This probably means that significant decorrelation of the individual frequency channels will have to be estimated from the wide-band signal. I don't know, and only question whether any of this is applicable to your write-up. Perhaps, this should be covered in section 3.2.8 where you discuss decorrelation.
    • The plan is to take data on 1s timescales (the relevant time scale for WVR phase correction -- the WVR was designed to have sufficient SNR in 1 s, but during the very best conditions at Chajnantor the phase fluctuations on timescales up to 5s are less than the WVR noise -- that happens like 5-10% of the time) and atmospheric phase fluctuations), apply the WVR data ON THE FLY, and then average up to 10s -- this avoids decorrelation and gives us much smaller data sets. This is an essential part of the plan that we should specify here as a place holder -- it will require software and much work -- specifically, we need to make the WVR delta T ---> delta phase conversion, and any dispersive phase corrections, work in NEAR REAL TIME. Research topic.

  • Section 3.2.8: I'm still not convinced that decorrelation of the phase means very much. Yes, if the phase fluctuations are bad over an integration time, than the average amplitude is decreased. But, the average phase is still probably pretty good...
    • True.

  • ...It may not be reliable measure of phases in the future, but is still is a reasonable average.
    • Well, this is for 950 GHz data, and the decorrelation will be like 0.95 -- ie, this could eat up most of our flux scale error budget. The good thing is that the WVR phase errors will be generally Gaussian-noise-like (unlike the horrible time & baseline dependent decorrelation you get from fast switching), so they are easy to correct.

  • Section 3.3: As I mentioned way above, I don't think this goal is important enough to warrant a separate section. Yes, we will push the calibrator switching times as long as possible, but nothing much changes in strategy.
    • OK --- this is a defect in my way of doing things. I give too much importance to the particular evolution of my own conceptual understanding of a problem, and think just because I understood something in stages means that there is a fundamental division here. I agree in this case there is not. The most fundamnetal thing for the 20-40 seconds is that we have good calculations of fast switching on those time scales, and we have cross-band electronic specifications for the 300 s time scale.

I have tried to include all of Ed's suggestions in the most recent rewrite. The latest iteration is a major change from the previous (JeffMangum).

Comments by BaltasarVilaVilaro - 12 May 2006

  • Section 1: It should be about goals. There is too much text on spec problems and inconsistencies.
  • Section 2: No WVR, just fast switching: Under which conditions will this mode be carried out?. I thought the WVRs will be pumping out the data anyway. Is this for those cases when the weather is so good that actually applying the WVR would add noise? Needs to be specified.
  • Section 3.1.1: A definition of the properties of a �phase calibrator� may be needed here�(compact source, flat spectrum or else, etc,etc).
  • Section 3.1.2: Is people comfortable with a 1s integration for the phase calibrators???. We have been worried about this for a while on this side and our conclusion was that 1s might be a bit too short (from a technical point of view).
  • Same Section: frequencies �above� and �below� 300GHz instead of �above� and �above�.
  • Same Section: It is not quite clear WHEN will the calibrator measurements be done. Will this be at the start of the run?. Will this be an Observatory task?.
  • Same Section: Why is this section here?. I would think that a general discussion on the properties of the phase calibrators would apply to ALL the modes in the list.
  • General Comment: (for high frequency observations) when the author(s) talk about �monitoring the phase solutions�, what phase solutions are they talking about, those at the final observing frequency or only those at the reference frequency?. If the latter, cross-band solutions should also be monitored.
  • Obvious typos/etc:
    • Section 3.1.2: �about about� repetition
    • Title of Section 3.2.6: �Daley�
    • Section 3.2.8: �acocunted�
  • General Comment: Should the document mention the fact that for phase transfer we will have to have the focus fixed?

Tried to include these comments in the most recent draft (JeffMangum).


Bandpass

See also comments from the Bandpass Calibration Example Review.

Comments by JeffMangum - 21 Dec 2005
  • Goals, bullet 1: I don't follow what this one is saying. The last sentence suggests that this BP correction is not needed.
  • Goals, bullet 2: I don't follow what the last sentence is saying. Do you mean that because the measurements will be averaged to one number that only one frequency point for the BP cal correction is needed?
  • Would it make sense to summarize the Bacmann and Guilloteau BP calibration sequence in the "Calibration Steps" section? In any event, I think that the "Steps" section needs a bit more detail.
  • There are two reference [5]s.

Comments by BaltasarVilaVilaro - 12 May 2006
  • Section 2: It would be worth mentioning HOW the data will be used in the Phasing between basebands.
  • Same Section: Repeated mention of ALMA Memo 505. Should include brief summary of techniques, etc.
  • Same Section: traditional �bandpass� calibration: The method outlined is not clear. How will the effects of phase shifts added when the bands are switched and IF atts are changed, to match the required levels, be dealt with?
  • Two [5] References in the References Section
  • Typo: "pbassband" in second-last paragraph on Page 4.


Polarization

Comments by JeffMangum - 21 Dec 2005
  • Goals: The Calibration Plan no longer exists. It is being superseded by these "example" documents.
  • In the description of the "preconditional calibrations" in the Goals section, it might be useful to list that the amplitude and primary beam calibration sequences, respectively, will take care of these requirements.
  • Section 2.1: Is the "measurement equation" Equation 1?
  • Section 2.1: Can you estimate how long these calibration measurements will need to be?
  • Section 2.1: For clarity, might be good to indicate that one is solving for dx and dy in this section (perhaps put this in the title for this subsection).
  • Section 2.2: Same comment as for Section 2.1. Indicate that delta_x and delta_y are being solved for here.
  • Section 2.3: Same comment as for Sections 2.1 and 2.2. Indicate that phi_xy is being solved for here.
  • Section 2.3: Estimate measurement time.
  • Obvious comment. Your "Further Issues" need to be dealt-with.

Comments by BaltasarVilaVilaro - 16 May 2006
  • Section 1: All matrices in Eq [1] should be defined here.
  • Same Section: A brief description of the gain measurements and a pointer to the Beam Calibration Steps is required.
  • Section 2.1: Some numeric estimates should be provided
  • Section 2.2: ��it follows the rules for (1) above��. Does this apply to the whole list in Section 2.1 or just point (1) in it. The latter does not seem to apply.
  • Same Section: Provide some numeric estimates, times, etc.
  • Section 2.3: Stress that for this measurement the ANGLE of the polarization needs to be known with a good precision.
  • Same Section: Other ways of measuring the angle have been explored??, hardware, etc?
  • General: Some discussion on the usable calibrators may be needed.
  • Section 2.4: All these items need to be dealt with
  • Appx. 2: Missing


Pointing

See also comments from the Pointing Calibration Example Review.

Comments by RobertLucas - 13 Apr 2006

  • Section 2 second bullet: details on the way to derive the refraction correction as a function of elevation for the current atmosphere will have to be precised for inclusion in the Control software. Can this be directly derived from the ATM model? do we need changes in the ATM software?
    • Good idea. Will look into this. (JeffMangum 10 Jun 2006).
  • Section 3.1 point 4. 'Monitor'
  • Section 3.2 'require many fewer sources' -> 'fewer'?
  • Section 5: Local pointing model can probably be roughly tested at the ATF but I believe its real efficiency can only be tested on the full array (sensitivity here is a key factor, we may not have enough nearby sources at the ATF).
    • Good point. Revised wording. (JeffMangum - 10 Jun 2006).

Comments by BaltasarVilaVilaro - 15 May 2006

  • Document does not mention interferometric continuum pointing explicitly. Other options being also considered? (line pointing, etc).
    • Good point. Yes, spectral line pointing is allowed (though likely won't be used much). Revised wording to clarify this. (JeffMangum - 10 Jun 2006).
  • Section 2: 0.6� spec for the antennas is up to 2 degrees. This affects some of the pointing modes mentioned later in the text and thus should be included.
    • Done. (Main.Jeff Mangum - 10 Jun 2006).
  • Same Section: ��[4] and Pardo should�� Ref for Pardo?
  • Section 3.1: Mention interferometric operation.
  • Same Section: Typo �Montor�
  • Section 3.2: Do not follow the phrase �For e.g 4 antennas only�.�. Are you talking about sub-arraying with 4 antennas only?.
  • Same Section: Should it explicitly state that the objective is a good AZ, ELV coverage for the range of antenna operations?.
    • Yes. Modified wording. (JeffMangum - 10 Jun 2006).
  • Section 3.3: For most high-freq Bands the pointing will have to be �referenced to Band 3�. What is the plan for synthesis observations?. In terms of phase issues we will not want to move the subref much at all, but if we point at 90GHz this becomes an issue. Is the plan to point /foc on the pointing calibrator at Band3 and then refocus to the sky-freq band and leave the subref fixed in the meantime?. Or are we going to build a �defocused� pointing model specifically for these modes?.
    • Clarified in text. Subreflector will be optimized at science band. (JeffMangum - 10 Jun 2006).


Location

See also comments from the Location Calibration Example Review.

Comments by JeffMangum - 21 Dec 2005
  • Sections 2.2 and 2.3: List the actual sequence of observations for each of the measurements listed. Note, though, that things like a pointing run can just be referred to without details (since this will be described in another example document).
  • Section 2.4: How often is "every so often"? Good to have an estimate here.
    • During the review of this document it was pointed out that the antenna position stability requirement of 65 micron is over a timescale of 2 weeks (a fortnight). Therefore, to zeroth order, baseline runs will be required at least every two weeks in the beginning. (JeffMangum - 17 Jul 2006).

Comments by BaltasarVilaVilaro - 15 May 2006
  • Section 1: Mention w.r.t. what are the antenna positions defined.
    • With respect to ICRF (JeffMangum - 17 Jul 2006).
  • Same Section: Where does the 71 micron spec come from?
  • Same Section: A bit of more verbose intro may be useful here; too direct to the details.
  • Section 2: Just checking .Is the computing group designing things so that the baseline calibration data is "automatically" applied to the data?
  • Section 2.1: Pointing Issues: Perhaps mode than "blind pointing" it would be better to say "global pointing model"?
    • All-sky pointing can be done during a baseline run. (JeffMangum - 17 Jul 2006).
  • Same Section: Initial Antenna Positions measured with what?
    • Pad positions likely measured with a laser tracker. (JeffMangum - 17 Jul 2006).
  • Same Section: Phase Stability: quite vague. Elaborate a bit more.
  • Same Section: Calibration Sources. What catalogues will be used for that?. Will ALMA have to create one with that precision?
    • Existing catalogs (VLA, VLBA, etc.) will be used for seed. Ultimately, ALMA will need to develop its own catalogs (for most calibration observations, in fact). (JeffMangum - 17 Jul 2006).
  • Section 2.2: What is the answer to the "What if only one is moved?" comment in red?
    • Still need to do rough baseline measurement, independent of how many antennas are moved. (JeffMangum - 17 Jul 2006).
  • Same Section: Point 2, typo "re-established"
  • Same Section: Point 6. Should there be a number here for what would be "acceptable" for day-time measurements?
    • I don't think that we have any choice here. It will be as good as it is going to be when antennas are moved. Since we only require something like 1cm accuracy for the crude baseline solution, I doubt that this is an issue (JeffMangum - 17 Jul 2006).
  • Same Section: Most arrays also monitor the gain variations with periodic observations of a given calibrator. Will we be doing the same?. If so, state that explicitly.
    • Part of phase calibration, I think (JeffMangum - 17 Jul 2006).
  • Section 2.3: Point 2. What is "excellent" phase??, we will need a bit of harder numbers in this statement, since it may depend on baselines, etc.
    • Have quoted best quartile STI phase information as guide. (JeffMangum - 17 Jul 2006).
  • Section 3: Does this whole section belong here?
    • I have replaced some of it with an "Issues" section (JeffMangum - 17 Jul 2006).

Comments by EdFomalont - 05 July 2006
  • Section 1: Generally agree with Baltasar comments that the introduction needs more substance.
    • The position precision is related to minimizing systematic phase offsets between cal and target.
    • 71 micrometer gives 2 deg phase offset at 300 GHz (scales with frequency) for cal-target offset of 5 deg. I assume that this specification (or something like this) is defined somewhere.
    • As far as I know, phase referencing phase considerations drives the antenna position accuracy. Decorrelation with time and over frequency or inaccurate u-v values are much less significant.
  • Section 2: Antenna Position (Baseline) Calibration Observations: The main point is that the best method for the determination of antenna positions is to analyze the phase/group delay from appropriate short scans of many calibrators over the sky. How and when these observations are made will be developed as ALMA is debugged. However, some specific concerns might be:
    • Individual antenna structure peculiarities (egs.~non-axes intersection) may have to be understood and monitored. Any significant elevation or azimuth dependence of the electronic phase must be determined and removed.
      • This is part of "Antenna and Electronic Delay Calibration" (JeffMangum - 17 Jul 2006).
    • The array clock error and astrometric/geodetic model (Calc) can impact on the apparent antenna position error. This and the above effects may determine how often baseline observations have to be made, even with no configuration change.
    • Interaction of the baseline observations with WVR observations and tropospheric modeling will be important, and can only be determined with the first few years of ALMA observations. As with present VLBI, global tropospheric properties may be determined directly from the observations more accurately than meteorological monitoring.
    • For ALMA astrometric proposals (eg. determination of the accurate absolute position of faint stars), the proposal itself probably should include more accurate baseline calibration observations to supplement the normal ALMA baseline monitoring.
    • We already have a catalog of suitably bright point-like calibrator sources with sub-mas positions (ICRF + friends). I think that a calibrator with 0.1 Jy flux is sufficiently bright for antenna-position determination. Using as wide a spanned bandwidth (about 20 GHz rather than 8 GHz) will improve the astrometric group delay accuracy.
    • When antennas are reconfigured, baseline observations with a nearby fixed antenna or two give the highest accuracy.
    • It is possible that the accurate position of reconfigured antennas may not be well-known for several days. While the expected baseline errors (after an initial quick baseline run) will not be large enough to impact most projects and to hamper the quick-look analysis of the data or the initial pipeline pass, phase referenced image accuracy will definitely improve with better baselines. Thus, the interaction of the pipeline results with the best antenna location values (and other post-observation calibrations) must be well-thought out.
    • ALMA will probably be used to search for suitable calibrators for relevant future targets and determination of the flux density of quasars. These searches could be combined with antenna position determination observations.
  • Section 3: A quick summary here could be:
    • The best observing strategy with WVR and tropospheric monitoring will be developed in the early year of ALMA.
    • The frequency of baseline observations (other than the obvious when antennas are reconfigured) depend on the structural stability of the antennas and other electronic phase effects which are associated with the pointing location of the antennas.
    • Coordination of a-postiori baseline corrections into the ALMA pipeline must be developed.
      • Have inserted all of these as "issues" (JeffMangum - 17 Jul 2006).


Delay

See also comments from the Antenna and Electronic Delay Calibration Example Review.

Comments by JeffMangum - 21 Dec 2005
  • Section 1: Might want to refer to the Calibration Plan as "defunct".
  • Section 2.1: It might be more instructive to bullet the steps listed in each of the current numbered bullets of this section.
  • Section 2.1: Can you list the calculation used to derive the delay from the calibration source measurements?

Comments by DarrelEmerson - 23 Feb 2006
  • There is an additonal measurement that needs to be made to correct for LO and timing signal errors. The measurement would would involve measuring an astronomical phase, then change the LO frequency by some amount (several MHz?) to establish the dphi/dfreq value. This has to be done separately for the 1st LO and 2nd LO. Timing also has to be established for the signal path delay and the digitizer clock, but this is already included in the current version of this Examples document. Note that, for each antenna at each antenna pad, separate constants need to be measured for the 1st LO delay, the 2nd LO delay, the digitizer clock delay and the signal path delay. (I think digitizer clock and signal path can be rolled into 1 number). -- RobertLucas - 13 Apr 2006 I'm not sure I understand this fully ... I thought remeasuring the phases on a calibrator for even a small change of frequency would be easy enough with ALMA sensitivity. Further more very small frequency changes are obtained by digital synthesis of an additional LO signal (FTS). DO you think we need to calibrate these devices?
  • In the section "How often?" I would add another bullet, "whenever there has been a system reset or power failure." This addresses the 8 nanosecond uncertainty we heard about.
This bullet would be unnecessary if the hardware could be made to provide unambiguous phase/timing under all conditions, which, since I'm not the one to have to design it, I think it should.
  • In the list "What quantities need to be archived" I think we need to add something like:
    • TauLO1 the effective propagation delay of the first LO to each antenna.
    • TauLO2 the effective propagation delay of the second LO to each antenna.
  • TauLO1 and TauLO2 will be similar, but will differ slightly because of cable lengths inside the antenna and indeterminate propagation delays through various electronic modules. The parameters are required, with a precision of ?10 ns or better, in order to compensate for potential phase changes while Doppler tracking the local oscillators.
  • I think the digitizer clock delays will be incorporated into the TauIF delay parameter.

Comments by BaltasarVilaVilaro - 13 May 2006
  • Section 1: Typos ��the delays offsets �.�, ��delays have to been compensated��, ��band width��.
  • Same Section: From point 2 on I think it is not actually a �Goal�. A new section should be created.
  • Same Section: Points 3,4,5,6 should be gathered in a subsection or else.
  • Same Section: Just before the Table on point 3��there are in fact two kind of��. Is it 3???, note typo too ��two kind of��
  • Same Section, point 3: Do not follow the meaning of the text right below the table. Please rephrase.


Optics

Comments by BaltasarVilaVilaro - 03 August 2006
  • ALMA Template Re-formatting done (file at the end of this page as attachement)

Comments by RobertLucas - 17 Aug 2006
  • 2.1.2. Transmitter holography:
    • There is no requirement to use a power of 2 for the number of rows. In reduction software we extend the map to a power of two anyway (we used 180 rows for antenna evaluation).
    • The uncertainty due to the measurement of the sigal feed phase pattern is important and should be mentioned.
  • 2.1.3. Celestial holography
    • It is very convenient to use more than one reference antenna to improve signal to noise ratio by \sqrt{N} and also to use mainly short baselines if we are not in the most compact configuration (or have too much shadowing in the most compact configuration).
  • 2.2 Subreflector positioning
    • Focus curves are a must for single dish, but once interferometry is available it is convenient to do small celestial holography maps (e.g. 16x16) to get in a single measurement the 3 focus coordinates and the illumination offsets.
  • 2.3 Rx Feed setting
    • Single beam alignment
      • for an accurate measurement of the pointing offsets between receiver bands a full map is not required: Pointing scans (crossed scans or five points) are enough.
    • Add measurements of illumination offsets. This is largely independent of the alignment (correspond to a rotation of the feed rather than to a translation). This is done by small holographic maps (see above). Note that the illumination offsets can also be used in data processing to correct the u,v values for mapping purposes.


Primary Beam

Comments by JeffMangum - 21 Dec 2005
  • Section 1: The current text in this Goals section should probably be moved to an "issues" appendix. Replace with the current PB calibration specification wording at CalPrimaryBeam.
  • Section 2.1, Point 4: What is the lower-limit to the SNR to get a useable beam map?
  • References: The existing single reference is just a formatting placeholder (due to JeffMangum). Proper references, if needed, should be added.

Comments by BaltasarVilaVilaro - 16 May 2006
  • Section 1: There is no longer this problem with beam specs. The current phrasing for this should be adopted here, as per SRR and other documents.
  • Section 2: Typo "AMLA".
  • Section 2.1: "We don't need to worry about flux or amplitude calibration". This should be rephrased to mean "absolute" calibration of these parameters.
  • Same Section: Clarification Needed. Has the need for "spectral line holography" been passed along to the pertinent IPTs?
  • Same Section: Point 7. Typo "we will be abundant SNR".
  • General: Will the polarization beam observations be summarized here or should they go into the "Polarization Calibration Examples"?


ACA Calibration Issues

Not yet reviewed.


Calibration Dependency Matrix

Related to this "example" calibration document work is a matrix which connects various kinds of calibration. The current version of this matrix is based on a spreadsheet which Masao Saito started a while back. Please comment.


Topic attachments
I Attachment Action Size Date Who Comment
ACA_ALMA_cal.pdfpdf ACA_ALMA_cal.pdf manage 60 K 2006-09-01 - 10:34 JeffMangum  
BandpassCalSteps.pdfpdf BandpassCalSteps.pdf manage 202 K 2007-08-14 - 06:26 UnknownUser Bandpass Calibration Steps (revised version)
DelayCalSteps.pdfpdf DelayCalSteps.pdf manage 194 K 2007-08-14 - 11:02 UnknownUser Delay Calibration Steps (revised version)
Topic revision: r78 - 2009-03-02, DickSramek
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback