SUMMARY OF ALMA RECEIVER JDRG MEETING 20-21 MARCH 2000 CHARLOTTESVILLE, VA S.T. MYERS (NRAO, Socorro) These are my take on the issues discussed at the RX CoDR meeting. Text marked with o are open issues and important questions. Text marked with * are my own comments and suggestions. ------------------------------------------------------------------------------- W.Wild - Summary of SAC decisions Bands: Priority 1 = 3 ( 89-116 GHz ) - can this go to 86GHz? 6 ( 211-275 ) 7 ( 275-370 ) 9 ( 602-720 ) Priority 2 = 1 ( 31.3-45 GHz ) 4 ( 125-163 ) 2 ( 67-90 ) - drops to 3 if 3 can cover 86GHz SiO Priority 3 = 5 ( 163-211 GHz ) 8 ( 385-500 ) 10 ( 787-950 ) - wait for better technology o Can we have 4 first light bands? o Can we get (perhaps at reduced sensitivity) 86 GHz SiO line on first light band 3 to help with testing and holography? * This list is loaded with high-frequency difficult bands at Priority 1 (eg. bands 7 and 9) which might be easier to do when better technology is available and might be better when array is fully checked out - could band 9 be traded for a different band at first light? Total Power: spec is 1E-4 in 1 second, but performance should not be sacrificed for stability Calibration: 1% adequate (though maybe too "agressive"?) o we assume this means amplitude. is this at all bands? Sidebands: allow DSB (with sideband separation/rejection at correlator) esp. at high bands. Choice to be made on band by band basis based on cost and sensitivity. Expect likely to have some bands of each, so correlator system will need to handle sideband separation. o are we doing sideband separation or is rejection allowable? WVR: goal 15um path in 1 sec, may need to be cooled to attain calibration goal of 1% (use WVR to determine emission in addition to opacity phase). Needs to be aligned to 3' (10' at low bands) o what are the required sensitivities (spec) so cooling can be evaluated Polarization: all bands should have polarization, but one (band 7 at 345 GHz) should be optimized for polarization. Ideally circular but will consider linear if it can be demonstrated to work well. o do we need stable single-dish polarization performance and at what level? o what are the cross-polarization and purity specs? o how uniform need the polarization be across the primary beam? Larry D'addario: his take on SAC meeting and questions Bandwidth: nominal spec is 16 GHz total (8 GHz per sideband/polarization) but 4 GHz allowed if both sidebands available simultaneously in each polarization. o it was unclear just what this means, make it clear! * can we trade sqrtBW for Tsys (eg. 4 GHz OK if Tsys at least 30% lower than with 8 GHz?) * the groups (SRON, IRAM) that favor Quasi-optical designs are driving this relaxation (note correlator will handle 4 x 2 GHz bands) - what are the cost and sensitivity breakpoints where we relax to 4 GHz RF/IF1 bandwidth? Frequency switching: 10ms for <0.03% change, 1.5s between selected bands (for fast switching), 15min(?) for large band changes Calibration: cold load not required for calibration o is this true in general to reach 1% cal spec, or is this assuming WVR is doing emission corrections? Polarization: what are polarization effects of tertiary optics? o just what are the polarization specs? Bands: is band 3 HFET or SIS? o need hard experimental numbers on HFET & SIS total power stability WVR: do we need cooling? LO: what are the power requirements? ------------------------------------------------------------------------------- W.Wild - planning Scheduled PPR for optics, dewar, cryocooler, 1st light bands in Dec 2001. Preproduction prototype in 2002. Release electronics for production in Apr 2003. Produce 1/Rx per month each band 2005-2009. o Is this schedule possible? o When do we freeze designs for the 1st light bands? o Do we need parallel agressive and fallback designs for the bands 6-9? * Since production runs are ~12/yr, can we do production upgrades to allow better designs to be utilized after 2007 for example? J.Payne - decisions "taken" One dewar for all bands astronomical bands (exc. possibly WVR). Use pulse-tube 1st & 2nd stage plus JT 3rd stage. Each receiver band in an independent "cartridge" Reflective optics (no lenses) for beam matching Band changes by pointing or mirrors (not subrefector) o Do we do band changes by pointing or mirror redirection? * In other words, are aberrations mild enough to just treat the dewar front as focal plane, or do we need pickoff mirrors to put beams more on-axis? J.Lamb - Sidebands DSB vs. SSB drivers: lines in image band -> 15-20 db rejection Tsys calibration single-dish operation consequences: LO control correlator (phase switching) IF design system software and analysis software system sensitivity (including Rx and backend electronics etc) SB sep/rej in Receiver or mixer: various implementations image rejecting mixer - reactive termination - tuning - narrow band - no noise in image sideband - demonstrated to 370 GHz image separating mixer - complex (junctions + hybrids) - low noise from image sideband - rejection depends upon ampl/phase balance (but relatively weakly) - demonstrated in 1mm band image diplexer and DSB mixer - matched termination of image sideband - 2 mixers - operated DSB - losses in diplexer - must be cryogenic (cooled terminating loads and optics) SB sep after mixer: implementation pros and cons phase switched LO - takes >1sec to switch through Walsh functions for 64 antennas! - must switch 1st and 2nd LO of each antenna - removes 1 sideband - works only in interferometer mode (not single dish) phase switched LO plus correlator phase binning - separates sidebands - noise from sidebands not separated - demonstated with 40db rejection - works only in interferometer mode (not single dish) frequency swept LO - sweep 1st and 2nd LO each antenna - smears image sideband across sweep range - best at rejecting narrow features in image band by smearing - demonstrated to about 20db rejection - works in single dish mode - limitations in 1st LO sweep range of ~10's MHz in current design Conclusions: - we will REQUIRE some post-mixer separation/rejection - this will mean phase-switching LOs w/separation in correlator - image separation NOT justified above 600 GHz, where it will likely be better to put effort into DSB sensitivity (this assumes trade-off of losses in separation vs. Tsys and atmosphere at site in these bands) o what are the single-dish sideband separation requirements? o do we need frequency sweeping and if so how do we widen the 1st LO sweep range? o are there cases of settings above 600 GHz (eg. with strong lines in image sidebands) where DSB wont work? (Bernard Lazareff presented a table with breakdown of Tsys for DSB and SSB cases) ------------------------------------------------------------------------------- Cryogenic System Design - presented by RAL (UK) group Brian Ellison: design goals - versatility to handle different Rx designs and optics - high capacity for cooling (esp at 4K) - compatible with antenna design (and focal plane) - reliable, maintainable, and producable proposed design - pulse-tube + JT ( 70,15,4 K stages) - single dewar for 10 astronomical bands (w or w/o WVR???) - reimaging optics to put beams through smallest windows - internal 70K plate for secondary stuff (IF amps, etc.) - receiver cartridge system - LO injection into cartridge or by reference signal only Mark Harmon: design changes after rumors from last weeks SAC - amend cartridge sizes down (to accomdate band 1 in dewar) other design considerations - robust cartridges - flexible thermal links - need for test program - cryocooler (thermal) connections flexible - dewar roughly 1 meter in diameter - two sizes of cartridges, 100mm dia (3x) and 170mm dia (7x) - hard mount 15K finger - aluminum dewar and shields (*why not copper shields?) - projected 2.5man/mos (*likely huge underestimate!) -> L74.5K Anna Orlowska more design issues, plus general discussion of desing - what is the total load at 4K? (exp 12mW dissip per ch) - temperature stability goal 10mK (in 1min) - radiation loads? 22W @70K, 8W @ 15K, 10mW @ 4K (*are these even close?) - design spec: 40W @ 70K, 10W @ 15K, 1W @ 4K - possible alternative 2 stage 40K/4K ? - JT or not? JT is temperature stable and buffer against refrig. cycle - operating MTBF GM refrigs seals 10kh compressor 20kh pulse tubes should be longer (?) - separate JT supply will greatly eliminate gas contamination and better buffer against the refrigerator pressure cycles - need high efficiency heat exchangers, servicing ~20kh - need estimates of 4K heat load, window loads, and 15K temp. stability - 4K cryo systems using separate JT supply should be reliable RAL and IRAM finding >9mos running time o what are the dewar orientation effects? o what temperature should mixers run at: 4K? 4.5K? o what is the mass of the 4K plate? o what is the total window area (will dominate load)? o what is cool down time? o what is the pump out time? Optical Design - SRON Matt Carter - baseline f/8 system - dewar 900mm dia -> use 820mm dia x 450mm deep with 9 Rx assumed WVR + band 1 outside - symmetric design (one size cartridges around periphery) - optics pass 5W, mirrors only - bands 2-6 use NRAO OMT bands 7-10 use quasi-optical (QO), probably need Martin-Puplett interferometers (MPI) for sideband rejection - bands 2-6 need 2 internal mirrors (one flat, one shaped) bands 7-10 need 3 internal focusing mirrors - original desing had cold load (now removed?) - response to rumors of SAC decisions: band 3 (345GHz) on-axis -> all bands on axis using mirrors no cold load -> cold load removed DSB bands 7-10 -> no MPIs band 1 included -> separate dewar WVR 3'-10' align -> pick-off mirror * this seemed significantly different than the focal plane presented by RAL cryo, need to hash out between them * I stated that for much of the time early on dewar will have many empty slots (only 4 1st light bands) so perhaps band 1 and WVR could obscure some ports (or even go in the dewar) LO and IF Design - NRAO Richard Bradley (and others) LO Multipliers - doublers to 250 GHz - triplers to 660 GHz (band 9 problematic) - unknown band 10 - use JPL HIFI design - PLL source bands: 27-33, 60-76, 74-94, 87-104, 99-122 (GHz) - tunerless, planar varactors, cooled above 120 GHz - doublers 3db bandwidth df/f 17% (71-84 GHz) LO Sources - primary problems: phase drift (>1s), phase noise (<1s), amplitude noise - noise budget 90% coherence @950GHz -> 77fs -> 23um equivalent - allocate 31fs (electronics) + 70fs (everything else) - graceful degradation: 0fs = 95% coh, 63fs = 85% coh - expect 14 fs fiber dist, 14fs YIG loop, 10fs multipliers (x4) - need RF shielding (50Hz!) o what is stability of photonic mixer (adds in)? o what is thermal phase drift stability? o need clean-up loop on reference? LO Power Amps - 65 GHz an up with JPL (GaAs to 120 GHz?) o LO chain total noise (1K/uW? 10 photons for 10hf/K spec) Specific Receiver Band Designs ------------------------------------------------------------------------------- John Payne - (NRAO) Bands 3 and 6 - pretty much the standard NRAO design (HFET band 3?) - see Tony Kerr's presentation later on for band 6 (and band 9) Baryshev & Wild - SRON Optical Design: - dedicated optics for each band (use pointing to choose band) - designed for 3' WVR alignment of any band SRON band 9 Design: - nominal 602 - 720 GHz - MPI plus grids -> sideband and polarization separation - advantage over DSB operation, but expensive - possible to incorporate balanced mixer design - try to image secondary on horn mouth o What is relative alignment of sidebands/polns on sky? o What is the projected cost? Sekimoto Band 8 Design: - nominal 385-500 GHz - quasi-optical (QO) sideband separation (MP) ~ 10db rejection (15db possible) - dual polarization + folded Fabry-Perot Interferometer (FPI) for LO diplexing - efficient LO coupling (-2db, ~20uW) - IF 4-8 GHz - adopt QO sideband-sep as baseline ASTE - f/8 10m primary - test band 8 prototype - Sumitomo GM + helium pot -> <2mK temp stability - 3-4 hf/K for 340-360 GHz test Rx using this design - install antenna at Atacama site end of 2001 3 candidate cryocoolers - Daikin GM+JT $70K - Mitsubishi 3stg GM $50K - Sumitomo dual 3stg GM $50K 1W@4K,5W@20K,12W@80K 2 cold heads + compressor Chalmers + IRAM Band 7 - newly prioritized by SAC, no details given ------------------------------------------------------------------------------- General Receiver Discussion at end of Day 1: o consensus that there is a need to make progress (faster) o desire for fixed (frozen) specifications from SAC and/or other groups (responsibility of AEC?) o need to evaluate total power stability of SIS vs. HFET (eg. band 2-3) o need to complete optical design soon o can the needed 70 receivers in each band actually be manufactured by the existing groups???? o when do the optical,cryo,1st Rx designs need to be frozen -> W.Wild suggested end of 2001 o should there be staged design decisions and freezes (eg. interfaces first, then optics, then cryo, etc.) ------------------------------------------------------------------------------- Summary of JRDG Closed Session (held 1st evening) AEC should make decisions - turn SAC recommendations into decision for baseline changes - decisions need to be published First decision should be whether WVR is to be used for anomalous refraction correction: In or Out? ALL BANDS (except possibly WVR) should go into single dewar - WVR in main dewar if cooled, but possibly 15K in sep dewar - this means band 1 should be designed into main dewar unless WVR is in second cooled dewar - size of WVR pickoff mirror depends upon whether anomalous refraction correction needed JDRG will approve baseline designs for each band - optics group models these baselines - if cartridge designer wants to change baseline, then needs approval of JDRG PROPOSAL: bands will be either DSB or SSB with sideband separating (or rejecting) mixers -> no quasi-optical sideband sep/rej * the second day it wasnt clear that this still held. yes or no? Other deadlines - need total window area for cryo load calculation asap! - need polarization properties of proposed designs Band 1 pickoff should be as large as needed with its cartridge in main dewar - it may obscure other bands, then must be retractable Need better communication - email exploder for notification of design changes - central Web site with receiver documents? - email directory of ALL ALMA related personnel ------------------------------------------------------------------------------- Second Day - More detailed recever designs Tony Kerr - Sideband Separating Mixers image rejection of only 10db is sufficient balanced mixers are preferred - ampl and phase balance requirements are NOT stringent - eg. 30degree phase OK o is the cost of balanced SIS mixer (with extra mixers plus hybrids) less than comparable unbalance QO designs? - balanced QO designs are possible (eg. Andrei & W.W.) - even better QO designs with circular pol. LO injection (eg. Sakimoto) sideband separating balanced waveguide mixers - slotted quadrature hybrids available - fairly easy machining (though harder at small wavelengths) - waveguide losses < 0.03 db/wavelength to 700 GHz MMIC circuits - capacitively loaded co-planar waveguide (CLCPW) - OK to put on thick substrate, conductors on same side - Nb CLCPW losses ~0.2 db/wave -> keep paths short - 211-275 GHz (band 6) balanced sideband-separating mixers produced by NRAO/UVA/JPL - 600-720 GHz (band 9) fully-integrated mixers produced by Jonas Zmuidzinas (Caltech) Implications for ALMA IF Requirements - isolator needed if IF amplifier outside mixer block - max isolator bandwidth ~octave -> IF 8-16 GHz for 8GHz BW - 4K termination of rejected sideband adds noise - can integrate IF preamp into mixer block (cf. Padin design) if parasitic L & C small Vacuum windows - < 0.1 db loss (should this be a spec?) - teflon 0.045" -> f2/f1 ~ 1.29 (0.1db width), f2/f1 ~ 1.18 (-20db) - quartz alone too narrow band - teflon-quartz sandwich better, but still less than full band - epanded teflon- teflon- quartz sandwich layers give very wide band o Should we check with mm/submm bolometer groups on what they use for IR blocking windows? o Would grooved polyethylene work? Claims of 0.05db losses? Chalmers Developments - Victor Belitsky Niobium facility (new) installed Band 7 Design - 275-370 GHz - SSB fix-tuned mixer, sideband separating - unbalanced, -15db LO coupling - 10 GHz scale model works well - 100 GHz demo receiver for Fall 2000 - ALMA band 7 mixer prototype Fall 2001 - 4-8 GHz IF o Can we get MMIC technology to push 4-12 GHz IF? o Will NRAO supply all the IF amplifier needs (even the integrated IF amp / mixers)? o Should we adopt the narrower (4GHz) bandwidth external IF amplifier as a fallback? SRON Developments - Andrei Baryshev & Wolfgang Wild Band 9 design (for JCMT) - 600-720 GHz (ALMA band 9) - 1 GHz IF bandwidth for JCMT - fixed tuned waveguide mixers - Rx temp 130-150 K, measured 320 K total DSB on JCMT Cold Plate breadboard receiver rig for testing - 500-1200 GHz - MPI, plus selectable blackbody loads - 2 mixer mounting ports Band 10 design for FIRST HIFI mixer - 800-1120 GHz RF - 4-8 GHz IF - waveguide (WG) and quasi-optical (QO) designs - WG = 270K @ 900 GHz QO = 185K @ 884 GHz for 2.3K bath -> 300K at 4.5K temp - mixer block manufactured by outside company - QO direct response 800-1050 GHz, 79% power main lobe (-11db) 2% total power in cross-polarization Integrated receiver (some of band 8) - 450-520 GHz RF - 0.5 GHz IF bandwidth - QO mixer, 95K @ 500 GHz for 2.5K temp - also waveguide version designed - on-chip LO injection - array configuration 5 x 4 pixels Band 9 design (for ALMA) - based on JCMT design - 600K DSB with 2 MPI (cf. 300K DSB w/o) - likely limited to 4 GHz IF bandwidth * note that this particular design seems to be one of the drivers for the SAC relaxing the 8GHz IF BW requirement. Should this be a fallback for band 9, or the prime design? DEMIRM/Paris and IRAM - Bernard Lazareff Band 8 Design - 480-640 GHz RF - helicoloidal feedhorns (double threads) -30db sidelobes, low x-pol! - version 1: image-rejecting, backshort tuned - version 2: DSB, fix-tuned Band 7 Design - 275-370 GHz RF - 4-12 GHz IF - full-height waveguide -> simple, low loss - single junction - probe-coupled, parallel tuned - simulation w/2GHz IF BW, but expect 8GHz IF BW to be OK - close to 7db or better image sideband rejection over 4-12 GHz IF BW (similar to MPI) - based on design as version 1 above, also a version 2 possible o When is the deadline to choose between competing (Chalmers, IRAM) designs for Band 7 (and any others)? HFET Amplifiers - Pospieszalski - MAP Q-Band give 25K or better - MAP 75-110 GHz and 60-90 GHz around 50K or better - also 3-13 GHz and 8-18 GHz HFETs for IF amplifiers - 1/f gain variations est. ~f^-0.45, 1E-3 at 1 Hz Use total power in band (bandpass filter defined, before detector) assume noise power equiv. to gain fluctuations. * Would actual gain measurement of narrow-band signal give same result? o What are the variations of 1/f amplitude (at a given frequency) across band? o Are there comparable measurements for SIS? - correlation receiver design for MAP presented - gives flat noise spectrum, ~2mK out of 60K TP -> 1E-4 spec met! * Can we afford to go to a correlation receiver design to meet total power spec? My guess is no, and if we adopt an HFET design we try to use passive (eg. temperature control) methods of gain stabilization only. o HFETs possible for bands 1-3. Band 1 certainly HFET. Band 2 lower priority. Thus, question is whether HFET or SIS for Band 3. o Can Marion's HFET for Band 3 make it down to 86 GHz? Possibly, though certainly at reduced performance. o What is the temperature stability vs. cryo temp fluctuations? Estimate 0.01 db/K? At 15K vs. 4K? - BIMA uses HFET LNA and gets 10mK stability -> 1E-4 OK More General Discussions: * If Band 1 goes into main dewar, should the HFET be at 4K (or 4.5 K) for stability? JT should be stable, and could regulate even better by raising with heater. o CRITICAL decision: will Bands 2 and/or 3 be HFET or SIS? Makes difference whether IR dumps onto 15K or 4K stage (respectively) THIS DECISION CANNOT BE CHANGE LATER AS IT IMPACTS THE CRYO! o What are the gain fluctuation properties of a balanced SIS design? (This keeps coming up, clearly need to do tests.) ------------------------------------------------------------------------------- Wolfgang Wild - Action Items Item Who? When? ---------------------- ------------- ------------- o QO vs WG / DSB vs. SSB James, Bernard 2 weeks o specs total power stability JDRG,ImCal,Larry 2 weeks o specs poln,cal,other WW,Victor,ImCal,Larry 2 weeks o agree of "final" specs JDRG + system grp ? o optics design, IF filters Matt,optics grp 3 weeks o dewar and cryo Brian,cryo grp 5 weeks o JDRG telecon JP,WW 3 weeks o email exploder, www site WW 1 week o band 3 to 86 GHz? Webber ? o LO multipliers Webber 3 weeks Other open questions o Do we have fallbacks for the LO (esp. the photonics)? o Is 100 uW for 2nd and 3rd LO OK? o Will bands 7 and 9 be DSB? If so, beamsplitters, MPI, or hybrids? o MUST HAVE SOME EARLY PRODUCTION RECEIVERS IN CHILE WHEN ANTENNAS ARRIVE. Which ones necessary? Which can be delayed? What is the hard date? Frequency bands baseline responsibilities o Band 1 Payne o Band 3 NRAO o Band 6 NRAO o Band 7 OSO/IRAM o Band 9 SRON John Payne - his open questions and missing info 1) Receiver specs needed -> scientific -> interface between cartridge & system (eg. LO) -> specs internal to cartridge (eg. load) 2) Exact definition of heat loads 3) Scientific polarization requirements (Larry and ImCal grp) 4) Complete Optics design 5) Polarization analysis of optics 6) DSB questions (cf. LO switching and correlator) 7) Band 6-10 LO system: delivery, power, bandwidth 8) Band 6-9 LO system: what fallbacks are acceptable 9) Total power stability 10) More detailed schedule - what are the milestones and deadlines? 11) Band 7 - which design to choose and when (same for band 9)? 12) IR filters ------------------------------------------------------------------------------- Final Discussions and political issues o What are the criteria for choosing between competing designs for the various bands? o Are parallel developments (eg. the band 7 and band 9) both charged to ALMA? * Should we encourage parallel developments, especially one risky and one fallback design? We may have to given schedule! o How do we deal with non-WBS sanctioned design work? o How do we take into account the Japanese??? Bob Brown Only 2 deliverables in Phase I - the scope and cost of the project! My final questions * How do we make for better communication between various groups, and with the SAC? * The SAC has given its priorities for the first light bands. These are loaded heavily to the highest frequency bands, which ironically have the riskiest, expensive, and volatile band technologies! SHOULD THE SAC BE GIVEN THE CHANCE AFTER SEEING THE DESIGNS AND PROJECTED COSTS TO RE-ORDER THE PRIORITIES? * In particular, it may turn out Band 9 needs extra development - could this be deferred? * What are the budgets available for the bands? Some of the presented designs, especially bands 8-10, appear to be complicated, expensive, and do not meet the 8GHz band requirement. Should there be a point where the go-ahead (with either the risky or fallback designs) or deferrment decisions are made? * Some of the lower bands (eg. Band 1 and Band 2) may be so easy to build that they could easily be a "fallback" for more challenging higher bands. Should this be considered? * I can easily envision some of the more extravagant high-band designs dominating the budget, effectively cutting out later bands, meaning that the SAC priority is not just an order but the bottom of the list may be lost (Band 10!). We need a reasonable costing of the various option ASAP. We also need a mechanism (AEC? AAC? JRDG?) for juggling cost / performance / availability issues!