Chicago-2016: RedBlueYellow Team

From CMB-S4 wiki
Revision as of 13:05, 21 October 2016 by Galtsev (talk | contribs) (Created page with "== RED+BLUE+YELLOW TEAM mini-plenary on array configurations == Moderator: Steve Padin === homogeneous array issues, pros & cons, desirable features === 5m? telescopes - Ni...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigationJump to search

RED+BLUE+YELLOW TEAM mini-plenary on array configurations

Moderator: Steve Padin

homogeneous array issues, pros & cons, desirable features

5m? telescopes - Nils

Can focus resources on one design.

Economy of scale, but large telescopes are more expensive.

Robust to failures & change in survey strategy.

low l performance not proven/known. Gives up high l if we don't go to 10m class.

Monolithic mirrors, boresight rotation, shields all more difficult.

Can we all get behind one telescope design?

mixed array issues, pros & cons, desirable features

~1m + <5m telescopes, ~1m + >5m telescopes - Akito

Small apertures will do low l, large apertures will do high l.

Big aperture can be really big because we don't need to worry about low l systematics. Segmented mirror OK on large aperture.

Can do boresight rotation on small apertures. Can scan faster.

Do we need pol modulation on small apertures? How do we decide?

More apertures means more complexity.

Could have >2 aperture sizes. More angular resolution options.

Need to define lmin & lmax for small & large apertures. Figuring out lmin for large apertures not easy.

Right now, we are pretty much obsessed with the details of how to build CMB-S4. Need some focus on what it needs to do.

next steps

Science requirements

  • Need top level requirements. Can we do this now, e.g., sigma(r)<0.001, goal <0.0005?
  • Instrument folks need angular resolution range & sensitivity vs. frequency. This already exists for some of the science, e.g., BICEP performance-based forecasting.

Configuration-this is the most pressing issue for CMB-S4 instrumentation (but Berkeley cost models show that sigma(r) and sigma(Neff) are not very strong functions of D for fixed total cost; broad range of configurations may be workable; some disagreement because assumptions in the cost model may be wrong).

Need a framework to get from science requirements to configuration requirements and then to more detailed instrument requirements (i.e., systems engineering).


  • how far can we get with simple calculations?
  • one big model vs. many small models?

Effects that must be captured

  • bands
  • efficiency (be realistic)
  • receiver 1/f
  • atmospheric opacity & noise
  • scattering & sidelobes, general vs. specific models, can we say anything about stability?
  • cost (careful, can derail the project, cost assumptions change with time and are always wrong)
  • complexity
  • lmin, lmax vs aperture size (systematics & foregrounds)
  • criteria for pol modulation
  • staged build, deployment
  • field size & scan strategy


(Minutes from Suzanne Staggs)

  • RED TEAM (homogeneous array of ~5m telescopes)
    • Could one receiver/ optics tube design serve both the 5m telescope and small ~0.5 m telescopes?
    • See Nils's slides for the rest of the concise points the red team put together!
  • BLUE TEAM (heterogeneous array with both large and small apertures)
    • Nils asks: did you consider a BICEP strategy where each small aperture has only 1-2 frequencies? Answer: yes, that was one thought on reducing complexity.
    • Kam: is what is driving you is the total bandwidth of the system, or the complexity of the system; if it is just the total bandwidth, then having the 1-2 bands per telescope could be good.
    • Peter: an advantage of this would be you could compare detection of r from large and small aperture. Answer: blue team did not assume you could get r from the large aperture.
    • Why not consider more like 1.5 m? Could do that as a reflector. Nils: In general, did you consider the question of reflector vs refractor for the small aperture? Answer: we noted that was something to consider.
    • We need to get focused on what we need to do or we might do the wrong thing -- Steve says this is a quote from Ed Wollack, roughly. Specifically, we eed to know requirements on angular resolution range versus frequency, as well as sensitivity versus frequency
    • Surprising thing from today was that a pretty broad range of configurations may be workable; need to keep looking at this and its underlying assumptions
    • See list above on ``Effects that must be captured"
    • We need to simultaneously get the science requirements and do the simulations and calculations on the instrument size to know what is possible, to meet the requirements. Steve suggests breaking the latter into two parts -- configuration requirements first, and then more detailed instrument requirements
    • Jeff: could we build part of it and then deploy and test it, and then build the rest of it? Padin: yes, we could write the project requirements to say we have to do that, even.
    • Nils: we need to consider a profile for detector deliveries (ie, timing and whether the first ones will need to be replaced later as fab improves), because that could drive us perhaps to wanting to have a heterogeneous array so we could get some of the easier telescopes going earlier to prove the detectors, for example.
    • Vlad: it appears to me that the varying science goals have not yet been pulled in to form a single set of science requirements
    • Natalie: usually there is an iterative process where you define your science goals, then think of the surveys and instruments, and see what you get to on cost. Then, you figure out if you should revisit the science goals, either making them easier to attain or more ambitious.
    • Clem : OK, then could we agree on the target sigma_r, and perhaps sigma_Neff, etc? For this first iteration of instrument design and costing? Discussion: sure, this would be good, and was a goal for the Science Book.
    • NEAR TERM GOAL FOR GROUP: set the first target science requirements!