From CMB-S4 wiki
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigationJump to search

Back to Argonne 2018 main page

Parallel session P1-2: Telescopes and Mounts (chair: N. Halverson)

Here’s the plan for the P1-2 Telescopes and Mounts parallel session on Tuesday:

Goal: Determine as much of the CMB-S4 large and small aperture telescope reference design concept as possible.


  • We need a design for which we can estimate sensitivity and perform independent costing
  • We need something concrete and credible for decadal. “Shovel ready.” We should not give options.
  • This is a reference design for the purposes of the decadal survey, not a technology down selection, DOE baseline design, or in any way a final design.
  • It is more important to make a decision then spend the next 9 months detailing our case for the decadal rather than getting stuck on design details or contentious decisions.

Concrete proposal to start discussion: For the large aperture telescope reference design, adopt a design similar to the SO crossed-Dragone telescope. For the small aperture telescope reference design, adopt a design similar to the BICEP-3 telescope.

For discussion during the hour:

  • If we do not have consensus on the proposal above, what are the issues that prevent consensus, and how can we get to consensus.
  • If we do have consensus on the proposal, what do we need to do to detail the design for the decadal:
    • Telescope design and aperture
    • Frequencies and number of detectors per telescope
    • Monolithic vs segmented mirrors (large aperture telescope)
    • Lens material and AR coating (small aperture telescope)
    • Number of optics tubes per telescope
    • Polarization modulation (boresight rotation, HWP, etc)
    • Ground pickup and sidelobe mitigation
    • etc.

In linked spreadsheet [1] discuss the telescope properties.

List viable options, and consensus choices. Indicate conclusions in spreadsheet.

We want as much detail as possible; feel free to add rows, listing new properties that are agreed to be important, or other properties where there is consensus.

Notes from session (Tuesday, March 6, 08:45-09:45)

Large Aperture:

  • CDT has said 6 meters is what we need for the diameter of the large aperture, so we do not need to revisit the question of whether we should go larger for point source science goals.
  • Segmented vs. monolithic primary for large aperture telescopes: Segmented is existing technology, so it might make sense to do this for reference design. Monolithic hasn’t been proven and can’t be flown to the South Pole, so it would have to be dragged. SO contacted many companies about this. Many have the ability to machine something this big, but maintaining temperature stability of the building during machining is hard, and most shops don’t have temperature control to do it. Companies will investigate but even investigation will be expensive. Panel gaps have been seen to be an issue in current experiments, and how they are dealt with will depend on the temperature range of the mirror. We have consensus that we will use a segmented baseline design.
  • Crossed-Dragone (pronounced Dra-goh-nee) has much more intrinsic cross-pol than either the ACT or SPT designs. That may or may not be an issue, but optical modeling should be used to verify that this isn’t.
  • For cost evaluation purposes, all large telescope designs will cost roughly the same, although the crossed-Dragone has a much bigger field of view.
  • From the SO experience, even though the field of view is large, you need to be able to take advantage of it (multiple receivers could hurt efficiency). Building a big receiver becomes challenging because the thickness of the front plate of the cryostat becomes large. This drives you to many small tubes but decreases your efficiency. (However, it is nice to have extra margin on your field of view.)
  • ACT and SPT designs dump stray light to the sky. The cross-dragone design dumps it to ambient temperature, which could pose a loading issue. This can be understood with modeling, and much of the work is already being started for other instruments.
  • Optics tubes per telescope on large aperture: SO has a single cryostat that’s 2.5 m in diameter with 13 windows. It’s not a fully populated focal plane area (could fit 19), but the number was kept low to reduce risk and complication. Individual optics tubes could make testing easier, but you lose the dead space around each tube (and the smaller your tubes are, the more you lose). Some of this space gets used by wiring, cryogenics, magnetic shielding, etc, so some amount of space is necessary. It’s an optimization between number of detector wafers per tube, size of tube optics, optical design, length of cryostat, etc. You also have to deal with the curvature of the focal plane in this design.
  • We might not have enough time between now and when the reference document is due to do a detailed analysis if we don’t adopt the SO design (or something very similar) as the reference design. The same could be said for baselining an existing design for the small aperture design. These designs are the closest to “shovel ready.”

Small Aperture:

  • Clarification on concrete proposal: BICEP Array is an array of BICEP-3s and there are 4 per mount.
  • The SO small aperture design is converging to something very similar to the BICEP-3 design, although SO will have a cryogenic HWP which makes it slightly smaller (because of the size of available sapphires).
  • It is relatively easy to change the small telescope design later if you learn something critical from an experiment currently on the sky.
  • The small apertures will be roughly 50 cm in diameter. Does this dictate the lens material and AR coating? Not obviously, but HWPs could be the limiting optic, because the biggest sapphires we can get are 51 cm. Ali will have alumina lenses with a 0.72 meter aperture.
  • Would you want different aperture sizes for different frequencies with the small apertures, and you might only want HWPs for some frequencies? The atmosphere falls off at lower frequencies, so you need it more at the higher ones. It is convenient that the frequencies where you might not want a HWP are also where you might want a larger aperture.
  • South Pole experiments have found so far that they don’t need a HWP, so it may only be necessary in Chile (and you might want to keep this as a difference between the two sites).
  • The window size will be related to the size of the ground shield, which is another reason why you might want to keep the apertures small.
  • Having a difference reference design at the South Pole and Chile increases the complexity
  • As you grow the aperture you can grow the throughput because you’re not limited by Strehl ratio, so larger telescopes really can have larger throughput.
  • What do we need to do to reach a consensus on the diameter of the small telescope design? (We don’t know what beam size you want to target at a given frequency). There’s a trade-off between science vs. risk of going to large aperture, and SO has a calculator that will become public this summer that can address this. Risk is that you might not be able to get an HWP big enough, and TRL of silicon is currently higher than alumina ( See further discussion of lens technologies below).

Technological maturity of different lens materials:

  • Silicon can go up to 46 cm in diameter. The TRL (Technology Readiness Level) is very good because these are in use on AdvACT (though not quite that big, but technology exists). 95/150 dichroic silicon lenses are on the sky and work well.
  • SPT is struggling with AR coating alumina, but that’s for triple band. Two seems to be much easier.
  • Alumina is lossier than silicon, but not so much that it’s a problem in current experiment, and alumina helps with filtering so you can potentially use fewer filters. Plastic AR coating for alumina at 95/150 is solved and works well, for other frequencies, challenges remain (sourcing materials and demonstrating it works). Alumina loss may become problematic at high frequencies (like 280 GHz).
  • BICEP Array is looking into both epoxy and plasma spray for alumina AR coating. Both are looking promising.
  • BICEP-3 is in the field with epoxy AR-coated alumina lenses, but single band per receiver. SPT-3G is deployed with a triple band AR on larger lenses, but has had issues with the AR coating and uses larger, they use a lower grade of alumina than B3 because of the larger size. SPT uses a 3 layer e-PTFE coating. It would be really helpful to have absorption and transmission spectra for the different technologies.
  • “It works” is a fuzzy statement, but we need to be more quantitative about actual transmission numbers because small penalties add up.
  • For any aperture less than 46 cm in size, one reason not the adopt silicon is that the AR coating takes time. We would need to invest in scaling up the production effort,and we would not need to do this to alumina. We will defer this decision to the next session.


  • We will need to decide how to handle the presentation of design risks, and in particular, how we present them to the committee in a way that is realistic but doesn’t steer the committee in the wrong direction. Risk assessment studies will be informed by the experience of instruments.
  • We will need to address the issue of using existing telescopes, because we will get push back unless there is a science case for why we can’t re-use existing ones. We don’t need an answer now, but we need to be sure to address it in the report. It is easy to write a case for why we can’t use current instruments like ACT, SPT, etc, but may need to address this for instruments that are more similar to the reference design like SO and CCAT-prime.
  • If we had serious options for alternative designs that we wanted to keep open, those options could be described in the appendix of the report.