SQUID-based Readout Schemes for Microcalorimeter Arrays

Transcription

1 SQUID-based Readout Schemes for Microcalorimeter Arrays Mikko Kivirata Heikki Seppä Jari S. Pettilä Juha Hassel he echical Research eter of Filad Ja va der Kuur Piet de Korte Marti Frericks Wouter va Kampe Piet de Groee he Space Research Orgaizatio of the Netherlads Motivated by the XEUS missio by the ESA Sigle-pixel sigal path Absorber rasitio edge SQUID amplifier No-fedback bare parameters: Room-temperature amplifier Power gai adwidth bath 1 α ( ) πω L SQ j G Iput coil resoace a be very large a be up to GHz s or more Dyamic rage G 4k 9.8L 3/ 4 SQ Φ 1/ 4 j k ~ 1 9 with stadard aalog circuits Use egative feedback to trade gai for badwidth & dy rage Use positive feedback to trade W & dy rage for gai

2 he desiger s job Power level (Absorber melts :) X-ray maximum eergy / rasitio o-liearity SQUID o-liearity R amplifier saturates Gai Dyamic rage Gai Johso oise Gai SQUID oise R amplifier oise Phoo oise floor ake care of the badwidths, too. F modifies iput & output impedaces (oise matchig) Stadard arragemet for the feedback paths... + _ + _ Electro-hermal Feedback (EF) Flux Locked Loop but there s a umber of other possibilities, eg. : +_ emperaturelocked loop +_ +_ Local F for the SQUID

3 What if we have a large umber of pixels? Direct readout: Feasible (compare: MEG devices) Heat leak through the wires omplex ad fragile orrelatio-based schemes: Noises are summed - bad Acceptable oly whe SNR ca tolerate summatio D-array: 3 x 3 pixels Wirig to mk stage (.8 µwh coolig capacity [XEUS]) 3D-array: x x pixels Multiplexig: Figerprit sigals by multiplyig by a orthogoal set of fuctios f 1 (t), f (t) ( sies & cosies; Hadamard fuctios; wavelets ) Sum to a sigle wire Detect the sigals by multiplyig with the same set f 1 (t), f (t) ad itegrate over all times Multiplier: (i) ES, (ii) SQUID, (iii) some extra device (a) (b) (c) (d) (e) Hadamard codes (Karasik & McGrath) Sies & cosies frequecy MUX imeslot fuctios time domai MUX Modified timeslot fuctios for ehaced duty cycle (4ch versios are symmetrically bipolar) urret through ES oscillates creatig oscillatory R- respose (oly oe half show) ESes as modulators I(t) G(t) U b (t) oductace G carries the sigal ias voltage carries the modulatig fuctio No direct thermal respose: average RMS heatig Magetic oliearity? Oly N wires from ES chip to SQUID chip for N M pixels Oly N SQUIDs

4 SQUID as modulator Sigal is multiplied by SQUID respose fuctio I MI ES I/ Φ. I/ Φ is a o-liear fuctio of U b Works best with two-level mod-fuctios m wires from ES chip to SQUID chip, if caot be itegrated moolithically Noise foldig Widebad oise is added after the modulator he oise from a give pixel aliases ito frequecy bads / timeslots / codes of other pixels (i) Provide gai so that oise summig ca be tolerated. (ii) Use frequecy-preferrig / timeslot-preferrig / codepreferrig oise blocker. I case of freq. MUX, the blocker is just a L resoator With other MUX schemes, active elemets ad exteral clock sigal feeds are eeded Without oise blockers With oise blockers Noise gets folded to other pixels No oise foldig

5 Filter implemetatio L is set by stability requiremet 8 H fits i ~.. mm implemetability sets lower limit to f ~ 5 MHz Magetic cross-couplig demads ~ 1 mm filter-to-filter separatio: (i) crosstalk betwee differet colums (ii) limits total W available to a colum ad separatio - Oly to avoid oise foldig - hael cofusio: take care by post-detectio filters X-ray absorbers ad ESes Noise-blockig L filters ω 1 ω ω ommo iductace i a colum Magetic cross-couplig (example) appears as commo series iductace, like L p Parasitic iductace L p i wirig: reactive part tued out with c, L p limits the badwidth. rasformers ramp up the impedace level, to help with parasitic L SQUID iput iductace L i ca be screeed away with egative feedback Feedback by flux ijectio or curret ijectio Wirig parasitics ueout cap SQUID iput I fb 1 ε L Quatum-limited badwidth: I i RI 4π f XEUS: R 1 mω 4 hbar for 3 chas separated by khz 1 mω -1 log - [R] -3 4pH i series, ot tued out APLA 6.4 User: V Automatio Jul log - [R] -3 4pH i series, tued out APLA 6.4 User: V Automatio Jul log - [R] -3 H i series, tued out APLA 6.4 User: V Automatio Jul mω M 4M 5M 6M 7M Freq [Hz] Z(w) -5 3M 4M 5M 6M 7M Freq [Hz] Z(w) -5 3M 4M 5M 6M 7M Freq [Hz] Z(w)

6 Dyamic rage ES curret: SQUID: Limited by self-oise I I pp.36 E max E FWHM Φ / Φ 3/ 4 Φ 9.8L SQ τ i 1/ 4 j k ~ for XEUS ~ for 1 K, j.5 pf, L SQ 4 ph (ε ~. hbar) SQUID: Limited by cable oise & R amplifier Φ / Φ Φ 5.3L 3/ 4 SQ 1/ 4 j k ~ 8 1 6, whe 1 K + K (for 3MHz R amp + cables) Need some more dyamic rage for liearity? Harmoic productio by a evet? (No, falls above the sigal bad) Mixig betwee a evet & imperfect idle curret balacig? (Probably ot) Mixig betwee two coicidet evets? (Not likely if pixels are scattered) Gai stability? (Probably yes) Dyamic rage - how to improve? Alleviate DR requiremet? Icrease itegratio time ( filter settlig time), still retaiig thermal stability coditio. Filters agaist oise foldig... replaced by duplex filters Array SQUID for sqrt() -fold DR improvemet? Log egative feedback at carrier freq. through R ot feasible, but... (i) F through low-dissipatio MOS amplifier at K? (ii) F through R at basebad rather tha carrier frequecy?

7 otal system: a sceario Adaptive idle curret cacellatio: grayed out