UMICH-2015: Instrumentation I break-out session 2

From CMB-S4 wiki
Jump to: navigation, search

Wiki navigation

Return to main workshop page

Return to Instrumentation I sessions page


Goals: Analyze technology status and progress needed to read out direct detectors on the scale of CMB-S4. Please note that direct coherent amplification of CMB signals (with HEMT or other amplifiers) is covered in the detector session.

  1. Identify requirements for CMB-S4 polarimeter readout electronics, including both cryogenic and room-temperature components
  2. Review status of existing technologies with particular attention to assessing feasibility for scaling to the total pixel count of CMB-S4 (order of 500,000 total across multiple platforms) and cost
  3. Identify work that needs to be done in order to complete maturation of candidate technologies.

Please come prepared to describe readout techniques for CMB detectors as described above. If you have a slide to add, please post it on the wiki. If you think of additional questions or topics for debate, please add those as well.

Top-level requirements for discussion


  • O($100 / pixel) is too much
  • O($10 / pixel) is about right
  • O($1 / pixel) is past the point of diminishing returns to CMB-S4, and may not be worth the R&D cost

Noise lower than the sensor

Cryogenic components scalable to O[500,000] sensors

  • Physical size of filter and readout components
  • Power dissipation per pixel
  • Focal-plane interconnects
Can wirebonds from each pixel out of the focal plane work for CMB-S4?
Bump-bond hybridization?
On-wafer multiplexing elements? (LC resonators coupled to TESs or MKIDs)
  • Number of wires to 4K
  • Number of wires to 300K

Acceptable crosstalk

  • Nearest neighbor
  • Distant pixel

Room-temperature electronics scalability

The path forward

Statement: we have two mature readout techniques: TDM and FDM.

  • Is the fabrication scalable in both of these techniques?
  • Can the cost be scaled?
  • Can the wiring be scaled?

Statement: we have 4 next-generation readout techniques at different levels of maturity: Microwave SQUIDs, KPUPs, Microwave resonator TESs, MKIDs

  • These techniques have advantages in scaling and cost
  • How much do we want to push to next generation techniques for CMB-S4?


CMB Polarimeter cryogenic multiplexer technology

  1. Time-division multiplexing for TES Bolometers
  2. Frequency-division multiplexing for ac-biased TES Bolometers
  3. Microwave-resonator multiplexing for dc-biased TES Bolometers: microwave SQUIDs, KPUPs, or direct readout with quantum-limited amplifiers
  4. Microwave-resonator readout of MKIDs

Room-temperature electronics

  1. UBC MCE for TDM
  2. McGill digital feedback electronics for FDM
  3. ROACH2 for resonator TES, microwave SQUID, or MKID
  4. SLAC LCLS boards for resonator TES, microwave SQUID, or MKID
  5. GPU

Advantages/disadvantages, comparison of readout options

  1. Technological maturity
  2. Fabrication complexity, yield, uniformity
  3. Scalability to order [500,000] detectors
  4. Cost

Required Work or Studies

  1. What is technological readiness?
  2. What are the technical tradeoffs?
  3. What are unknowns?
  4. What is timeline for development?

Slides for Discussion

MUX overview: [[Media:Irwin.ReadoutStatusOverview.pdf]]

Time division SQUID multiplexing (Hannes Hubmayr): Media:Hubmayr_cmbs4_tdm.pdf

Adrian Lee: FDM [[Media:2015.09.22.Adrian.Lee.FDM.small.pdf]]

Microwave SQUID multiplexing (Hannes Hubmayr): Media:Hubmayr_cmbs4_umux.pdf

Ed Wollack: Superconducting bump-bond hybridization [[Media:CMBS42014Wollack_Superconducting_Interconnects.pdf]]

Peter Day: KPUPs [[Media:Day_-_KPUP_-_0815.pdf]]


two things in the field TDM FDM

MKID, microwave SQUID, KPUP, Direct TES readout


direct TES variable Q

microwave SQUID and MKIDs

O($100)/pixel is too much O($10) is about right O($1) is too little

TODO item

table with power dissipation/pixel, physical size,

number of wires to 4K or 300K

MCB, 10 person years of firmware.

cost of producing cold electronics, resonator arrays

how much energy goes into microwave techniques.

we could do both and muddle along


very mature.


300:1 MUX factor common bias for each detector. constrains fabrication. a lot of cold interconnect

so far not on the wafer. well developed warm electronics. cost per pixel ~$100.



33 channel MUX

ROACH2 architecture

LCLS multi slow ATCA crate, kintex

when do we have to decide?


kinetic inductance is modulated by current.

are people going to 300 MHz range...?

practical things - biasing etc.

when do we have to decide?

a few mW, could get 200 to 300 microW

hands on the options in the next year.

What do we do need to demo any detector/combo?

for some things may require a CMB test.

table of parameters action plan to TRL to sell to DOE

if you have a problem with yield, then individual bias good.

magnetic fields and RF sensitivity.

low frequency performance.... understanding of challenges and risk

need 100 mHz 1/f performance since we don't know if we will use modulators.


Kent's cost analysis

O($100)/pixel is too much

O($10) is about right

O($1) is too little

Three categories of MUX

1) TDM and FDM - Mature today for Stage III

Can these be the CMB-S4 readout technology with evolutionary changes?

TDM - hybridization

FDM - hybridization or direct integration

Costs of both are order $100/pixel, so too high at present.


Excellent scalability, and plausible to reach cost.

  • Need to demonstrate in experiment(s)*

Great progress on performance

1/f knee at 100 mHz required

3) Microwave Readout of TESes - MSQUIDs, KPUP, Direct coupling to resonator

Readiness varies.

MSQUIDs on telescope (MUSTANG)

KPUP in lab demo

Direct Coupling at idea stage, requires new amplifier.


When do we have to pick? CD2 ~2020?

Interactions with the rest of the experiment: is there a modulator -> 1/f requirement


Make a performance table

Power dissipation

Wire count


magnetic field sensitivity

RF sensitivity