DAQs for cryogenic detectors in Cosmology Gustavo Cancelo, FERMILAB - - PowerPoint PPT Presentation

DAQs for cryogenic detectors in Cosmology

Gustavo Cancelo, FERMILAB DAQ R&D Workshop 11 October 2017

Objectives

• Identify the DAQ and Trigger challenges that long-term future

experiments will face.

• Identify detector and DAQ synergies and common efforts

across our physics frontiers.

• Identify breakthroughs in detector and DAQ from outside the

DOE/NSF community that can facilitate our physics and computing.

10/10/2017 Presenter | Presentation Title 2

Identify the DAQ and Trigger challenges that long-term future experiments will face

• Detector development is focused on physics goals and
• pportunities.

– High-tech consumer products also generate detector breakthroughs.

• Detectors drive DAQ and Trigger architectures.
• Detectors, DAQ and Triggers for Cosmology overlap with
• ther areas of our physics such as Neutrinos, HEP and

computing.

• Cosmology is a very diverse field. I will highlight DAQ for:

– Dark matter – Dark energy

10/10/2017 Presenter | Presentation Title 3

Dark matter

• Almost everybody thinks that DM exists and it is cold.

– Currently and in the near future the attention will focus on WIMPs and AXIONs. – WIMPs and AXIONs are particles that interact through the weak force. – DM models cover energy scales from ueV to TeV. – The energy scale is unable to be covered by a single experiment. – Current DOE experiments are ground based and direct search.

10/10/2017 Presenter | Presentation Title 4

• WIMP search experiments look for nuclear or electron recoils in the detector mass.

– Atmospheric neutrino background is a potential problem. Characterizing the neutrino floor is important.

• AXION experiments try to detect the decay into a pair of photons. Measure power in a

resonant cavity.

Dark matter

• For decades DM experiments evolved with what the technology allowed.

– New technologies based on cryogenic detectors (e.g. noble gases and liquids, silicon devices at ~140K and superconducting sensors) have pushed the energy threshold to sub GeV scales. – DAQs have accompanied the trend by developing ultra low noise electronics.

• Not always “bigger is better”, DM experiments must beat the radioactive backgrounds.

– DM experiments have achieved a maturity that have allowed them to grow to multiple Kg or tonne of detector mass.

• DOE G2 experiments: LZ (Xenon), ADMX (axions), SCDMS (superconducting

germanium).

• Will these technologies carry over to G3?
• What other technologies are trying to gain space?

– Massive “zero noise” CCDs (skipper). – Low noise CMOS detectors. – Gas TPCs for directional DM search. – Superconducting detectors: MKIDs, Josephson junction, etc. – Multi cavities for Axion detection. – Superfluid detectors for low energy DM.

10/10/2017 Presenter | Presentation Title 5

Dark matter: Future DAQ example for massive CCDs

• Established detector technologies will grow in mass and number of detectors/channels.
• “Zero noise” silicon detector can see nuclear and electron recoils with 1.1 eV threshold

10/10/2017 Presenter | Presentation Title 6

Javier Tiffenberg arXiv:1509.01598

muons, electrons and diffusion limited hits. Next generation of CCDs achieve <0.1 electrons of noise !! 20,000 channel instruments for dark matter and low energy neutrino searches. Ultra low noise DAQs are needed (zero noise contribution).

Particle detection with CCDs

CCDs for DM and neutrinos

10/10/2017 Presenter | Presentation Title 8

• 6K x 6K pixels, 1mm thick = 20g of mass.
• Can operate at 140K. Dark current 10−3 e- pix−1 day −1. Could achieve 10 −7 e- pix−1 day −1.
• Radiopurity of the current package < 5 dru.

Dark matter: FNAL DAQ

• Very successful collaboration of FERMILAB, Univ. del Sur (Argentina), CNEA (Argentina),

UNAM (Mexico), Univ Asuncion (Paraguay)

10/10/2017 Presenter | Presentation Title 9

• 4 channel DAQ is working.

– The purpose of the 4 channel is CCD characterization in the labs.

• 20 channel/board DAQ being designed to target

detectors of up to 1000 channels.

• Experiments:

– DAMIC 1Kg, ~200 channels. – SENSEI: Skipper CCDs, 100g, 200 ch. – CONNIE 1Kg: ~1000 channels.

• DAQ challenges for a larger system (e.g. 1000 channels, 50 x 20 channel board)
• System noise: including PCB and EMI noise must be contributing zero noise to a “zero

noise” detector.

• Conductive EMI from AC line, vacuum and cooling.
• Each CCD channel requires 17 clocking signals at 10’s of volts and a video reading uV.
• CCD bias voltages must be clean.

5 x 5 inches

Dark matter: FNAL DAQ

10/10/2017 Presenter | Presentation Title 10

• Clock channels can be shared.
• Videos could be shared.

– Skipper CCDs may need to be readout continuously.

• Grounding and controlling EMI is crucial.
• Data bandwidth for 1000 channels: 1 GB/s all to disk.
• No triggers or data combiners needed.
• Sophisticated DAQ and slow control software needed.

Dewar for 2 Kg mass 4 quadrants 24 CCDs each

10/10/2017 Gustavo Cancelo | CONNIE electronics 11

DAQ electronics for 24 CCDs

VIB

CCD DAQ FPGA

Ethernet CLK Generation Based on 40 channel 12 bit DACs Video readout Based on 20 MHz 18 bit ADCs . . . . . . . . .

DC power

SPI

DAC_H DAC_L

Analog switch Matrix

Bias voltages Generation Based on 40 channel 12 bit DACs

Individual

• r grouped

Power management Linear DC Power management Linear or switched

. . . . . . . . .

Bias 1 Bias N Analog MUX

The most challenging part is that all the data must be acquired with the level of

• quality. If the noise is not uniform it is hard to achieve low energy thresholds and to

calculate detector efficiencies. The DAQ must work flawlessly and stably during the life of the experiment.

10/10/2017 Gustavo Cancelo | CONNIE electronics 12

Physics reach

• New physics:
• Dark photon
• A’ boson
• Neutrino magnetic moment
• Large number of theoretical

models can be tested

DAQs for Dark energy and the evolution of the universe

10/10/2017 Presenter | Presentation Title 13

DAQs for Dark energy and the evolution of the universe

10/10/2017 Presenter | Presentation Title 14

• Science reach:

– Dark energy – Large scale mass. – BAO. – Inflation – Neutrino masses – Light relativistic species. – Etc.

DES CMB

• The present:

– DES – SPT3, ACT

• The near future:

– LSST – DESI – Simons observatory

• What is in the longer term future

for CMB and optical surveys?

Optical surveys and CMB highly complementary Other probes can also contribute

Electronics and DAQ

DAQs for Dark energy and the evolution of the universe

10/10/2017 Presenter | Presentation Title 15

LSST will generate >1 billion galaxy catalog. Opportunities for spectroscopic surveys!!

• CMB future: CMB S4

– A collection of CMB telescopes at the South pole and Atacama – Superconducting detectors: Frequency Multiplexed TES or MKIDS.

• Optical surveys:

– High and low resolution spectroscopy. – 100,000 channels high res spectrometer? – Low res MKIDs based instrument? Could cover the near infrared spectrum!

DAQs for Dark energy and the evolution of the universe

10/10/2017 Presenter | Presentation Title 16

• Superconducting detectors
• Frequency multiplexed.
• RF electronics.
• Almost the same warm

electronics.

• Challenges:

– High number of channels per RF feed to minimize thermal load and detector wiring. – Low noise in a multi GHz RF environment with noise sources coming from mostly digital electronics. – Cost: few dollars/channel. – High input and output bandwidth.

A/D, channelizing, digital filtering, DAC: signal generator for thousands of channels 1pps, and 10 MHz reference. Rubidium clock, frequency synthesizer Computer MKID HEMT

Cold Warm Warm RF IF IF

Fermilab DAQ: this is the most advanced electronics today 10 K pixels crate

10/10/2017 17

ROACH2 Fermilab electronics

Gustavo Cancelo | Scalable 10 to 20 Kilo-pixel MKID Signal Generation and DAQ for Cosmology

RF out IF in RF in IF out LO

To MKID from MKID

Up conversion, amplification, attenuation and filtering Down conversion, amplification, attenuation and filtering

To MKID from MKID To/from ROACH2 MKIDs for optical require a detector with a BW of ~250 KHz. CMB ~100Hz. (More channels per ADC and more resolution).

FMESSI (Frequency Multiplexed Electronics for Superconducting Sensor Instrumentation)

10/10/2017 18 Gustavo Cancelo | Scalable 10 to 20 Kilo-pixel MKID Signal Generation and DAQ for Cosmology

• Applications:

– CMB-S4. – Quantum computing. – Low resolution spectroscopy for cosmology.

• DARKNESS: 10,000 pixel instrument
• 2 observation runs at Palomar
• KRAKENS: 30,000 to be deployed at KECK 1

Mauna Kea coming up Oct.2017

DARKNESS at Palomar

The next step is to apply the FMESSI DAQ to CMB sensors working in frequency multiplexed mode. CMB channels are low bandwidth. 5000 ch/board is doable with current DAQ. Cost: $1/channel (5000 channels)

Performance of the RF electronics 4-8 GHz

• Noise is not the main issue.

– HEMT noise is 2K and has a gain of 40 dB, the input amplifier of the warm electronics has a noise temperature of 360K. 20 dB less than the HEMT output noise.

• The main issues are:

– gain flatness over 4-8 GHz, low harmonic distortion when receiving 2000 channels packed less than 2 MHz apart. – Keeping EMI and LO noise down. – Being immune to high frequency switching noise from the ADC/DAC and Roach boards.

10/10/2017 Presenter | Presentation Title 19

This is a plot of 1000 tones. Equal powers/tone. 6db flatness before filter roll off

RF out IF in RF in IF out LO To MKID from MKID

RF board performance

• Tallest harmonics are due to the finite table size generating the 1000 tones. In this case

generate a harmonic 7.6 KHz from the tone.

• Noise floor at -120 dBm.

10/10/2017 Presenter | Presentation Title 20

Tone power -70 dBm Spur due to DAC table size = -120 dBm

Other tones

RF out IF in RF in IF out LO To MKID from MKID

RF board performance

• After the mixer the sideband suppression is 38 dB before fine tuning.
• Fine tuning can bring sideband suppression to 50 dB below signal.

10/10/2017 Presenter | Presentation Title 21

Sidebands are created by amplitude and phase imbalances in the I Q mixer and IF amplifiers

ADC/DAC board

• This is a very complex board. 2 ADCs and 2 DACs sampling at 2Gs/s. Virtex 7 FPGA.
• The communication between the ADC/DAC and the FPGA is via JESD interface (challenging

layout and firmware).

• All signals must be synchronized to a 10MHz reference (external rubidium clock).
• The 12 bit ADC performance is critical because we must acquire 1000 channels.
• The effective number of bits and SNR depends on the signal “crest factor” which is the peak

voltage of all combined 1000 tones to the voltage of a single tone.

• There are algorithms to optimize the crest factor.
• For 1000 tones you can get crest factors lower than 70.
• If we use 100 to be safe, we get SNR of 64 dB and 10.5 ENOB per channel.

10/10/2017 Presenter | Presentation Title 22

DAQ for CMB S4

10/10/2017 Presenter | Presentation Title 23

• CMB S4 requirements

– Channel BW 100Hz. – DAC excitations -30dBm per tone for 4000 tones. – Signal to noise ratio better than 90dBc. – Phase noise:? – ENOB?

• The current electronics can serve >4000 detectors.

– DAC excitations -30dBm – Signal to noise ratio better than 120dBc – Phase noise better than -125 dBc. – ENOB > 14 bits – The limitations will be on the crest factor and on the FPGA processing resources. – Basic crest factor using phase randomization is

DAQs for Dark energy and the evolution of the universe

10/10/2017 Presenter | Presentation Title 24

• Technology is progressing fast.

– Soon we will be able to design a DAQ with 10 fold the capacity of the one we have.

• ADC and DACs built into the FPGA.
• The limitation will be on the FPGA processing.

– FPGA design becoming increasingly more complex.

• New RF challenges on integration of multiple 1-10GHz channels.
• New LNA (Low Noise Amplifiers).

– A 2K (noise temperature amplifier) costs $7K, we will need thousands.

• PCB design challenges.

– Low noise at blasting speeds. – 33 GHz GLINKS – 4 Gs/s multi ADCs and DACs. – High speed DC/DC conversion and low EMC at the same time.

• Where to put resources.

– LNA design could save us money. – FPGA design requires real experts as much as ASIC designers. – Engineers who can do RF and high speed analog/digital – PCB designers. – DAQ software such as OTS for 500K channels.

Silicon Photomultiplier detector DAQ

10/10/2017 Presenter | Presentation Title 25

• SiPMs are single photon and high

precision timing detectors.

• Applications all 3 frontiers of our physics

– Neutrinos: DUNE, SBND, – Cosmology: DM, Intensity interferometry, Pierre Auger, pulsars. – HEP: LHC fast timing

• Large photon detector planes for room temperature or cryogenic environments (LAr,

XENON, etc)

• Timing resolution few ps.
• Spatial resolution < 1cm2.
• Single photon counting capability.
• Active ganging of SiPMs to cover large areas with fewer readout channels.

DAQ synergies

10/10/2017 Presenter | Presentation Title 26

• CMB S4, Optical surveys and quantum computing:
• CCDs and CMOS detectors for large scale DM and Neutrino experiments.
• Silicon photon detectors for neutrino, DM, Intensity interferometry, and HEP.

Where to put effort and money?

10/10/2017 Presenter | Presentation Title 27

• Large scale, ultra low noise electronics for DM and Neutrinos.
• Low Noise cryogenic amplifiers for DE, CMB and QC.
• Multi channel RF electronics embedded in analog/digital systems.
• Highly complex multi GHz analog and digital design.
• Complex FPGA design.
• High speed and optical links of over 30Gb/s.
• EMC design.
• Fast timing for Cosmology and HEP.
• DAQ software interfaces.

Spare slides

10/11/2017 Presenter | Presentation Title 28

Performance of the fMESSI electronics. ADC: AD9625

• Raw data at 2Gs/s, unfiltered.
• SFDR reduced by the generator harmonics

10/11/2017 Presenter | Presentation Title 29

2000MSPS 53MHz at -1dBFS SNR=59 dBFS SFDR=-72dBFS

Single tone

Performance of the fMESSI electronics. ADC: AD9625

• Raw data at 2Gs/s, unfiltered.
• SFDR reduced by the generator harmonics
• Nonlinear distortion below -70 dB
• Two independent signal generators with their close in noise.

10/11/2017 Presenter | Presentation Title 30

2000MSPS 52.9 Mhz and 55.9Mhz at -7dBFS SFDR=-72dBFS to 2nd harmonic