1. CARMA, and the CARMA WVR effort Alberto Bolatto Associate Research Astronomer U.C. Berkeley Astronomy Radio Astronomy Lab Dick Plambeck (UCB/RAL), Dave Woody (Caltech), Leslie Looney, Yu-Shao Shiao (UI), Douglas Bock (CARMA) WVR workshop Wettzell 2006

2. Outline What is CARMA? The OVRO experience The RAL correlation radiometer What next?

3. + UChicago SZA 8 3.5-m antennas Berkeley-Illinois-Maryland array 10 6.1-m diameter antennas Caltech array 6 10.4-m antennas CEDAR FLAT

16. Cedar Flat – elevation 2200m June 2004August 2005

17. 21 Jul 2004 – lifting off the first reflector

29. panel adjustment surface error determined from holography before adjustment: 127 μm rms → 75% loss at 225 GHz after adjustment: 28 μm rms → 7% loss at 225 GHz

37. all antennas assembled 10 Aug 2005

38. Comparison with other arrays CARMA + SZA SMAIRAMALMA elevation2200 m420025005000 antennas238650+ baselines25328151225+ diameter10, 6, 3.561512, 7 area850 m 2 22610605600+ max baseline 1900 m500 m400 m14 km

39. Comparison of u,v coverage 6 hr track on source at decl +10º OVRO E, 15 baselinesCARMA D, 105 baselines

40. Synthesized beams 5% contours

42. E, D configurations Now 1.6 km baselines 8–150 m 1mm beam: 2”

43. E, D, C configurations for Winter 2005 1.6 km baselines 8–350 m 1mm beam: 0.8”

44. E, D, C, B, B+ configurations for Winter 2006 1.6 km baselines 8–1700 m 1mm beam: 0.2”

45. E, D, C, B, A configurations for Winter 2008 1.6 km baselines 8–1900 m 1mm beam: 0.13”

46. 225 GHz zenith opacity %taumm H2O SSB Tsys 25<.12<1.8<290 50<.16<2.4<350 75<.28<4.3<520 Tsys computed for 1.5 airmasses, Trcvr(DSB) = 45 K

48. OVRO WVR

49. Sample phase improvement

50. It can work, but… Can it work reliably? It’s easy to improve very bad tracks, but good tracks can be worsen Only works for ~40% of the data Y.-S. Shiao et al., SPIE, (2006)

51. Correlation WVR at 22 GHz Correlation receiver: less sensitive to amplifier gain variations, no moving parts, built-in absolute calibration. Fast control of temperature of reference for nulling: ultimate stability. Weak points: complexity, sensitive to spurious correlations

52. Expected performance Measured amplifier performance based on Hittite commercial HMC 281 GaAs mmic ($40): T noise ~55 K, G ~23 dB, BP ~16-36 GHz Expect T sys ~ 140 K, or RMS ~5 mK in 1s in 1 GHz hot spill~3% (9 K), input w.g. loss~0.5 dB (32 K), hybrid+w.g./coax loss~0.3 dB (4 K), 2 nd amp stage~5-10 K Assuming canonical  ~4.5 mm/K @ 22.2 GHz expect path RMS ~20 mm in 1s Performance will be degraded by control of load temperature, thermometry, spectral baseline removal, etc, but there is a safe margin – /20 goal is ~60 mm

53. Block diagram x2 CARMA X-band CTL CANbus uP + DAQ 22 GHz optics 180 hybrid/magic T BIMA dewar HMC 281 cryo amps DITOM D3I1826 DUAL HMC281 NARDA 4017C-10 MARKI M1R-0726L 18-26 GHz ASTRONOMY IF MCL SLP-550 NARDA 4317B-2 D0612LA 9-13.5 GHz YIG OSC. DETECTOR 4-q multiplier 180° PHASE SWITCH WR42 th. gap + window 12 K stage 40 K stage

54. The Receiver K band cryo amp (x4) Controled temp. load Magic-tee hybrid Thermal clamp to 12 K stage Input (to horn) K band cryo amp (x4) Thermal clamp to 12 K stage Controled temp. load Magic-tee hybrid

55. The Controlled Temperature Load Cernox sensor chip on top of inverted 50 Ω alumina resistor 10 mil 50 Ω quartz μstrip Heater biasing wire Load + sensor mass is 3 mg: fast temperature response Once mounted, sensors are calibrated against standards S 11 ~-20 dB Brass pedestal

56. The Amplifiers Hittite HMC 281 12 amps put together by Dusty Madison, a freshman summer student who learned to assemble and wirebond them

57. The Dewar “Insert” Minimum impact on existing BIMA dewar No internal screws/electrical connections: just plugs in Special 2-port test dewar designed and fabricated

58. Other Hardware LO/downconverter IF/Multiplier Microcontroller Signal conditioning

59. The Complete System WVR dewar “inserts” LO chain and downconverter IF chain, AGC, multiplier, phase switching, and filters Signal conditioning and control electronics XAC uP, ADC, DAC, and CANbus

60. Nice idea, but it has proven difficult to make it work Tests looking into heated cryogenic waveguide load in 2 nd dewar Non flat passband –Slope is caused by imperfect hybrid –Central feature is from CTL wg adaptor –Edges not quite understood –A few K of “extra” correlation, probably reflections in hybrid Status May 2005

61. Nice idea, but it has proven difficult to make it work Spurious correlation due to internal coherent reflections –Could be mitigated with input isolators Even without moving parts, calibration is not repeatable enough –Difficult to attain the mK calibrability goal Further work? Status May 2005

62. What next? Revert to basics – Simple is beautiful Implement a Dicke-switch radiometer –Room temperature: use noise diode –Cryogenic: use controlled temperature load LO CTL DET e.g. AD8309 Dicke-WVR assembly using CTL

63. Conclusions Phase correction schemes improve correlation for a fraction of the tracks, but not all the time. Atmosphere or engineering? Nulling correlation radiometers are nice in theory, very difficult in practice. Large part count and complexity makes them unattractive for (university based) interferometers. Dickey-switch type schemes are considerably simpler, and more attractive if stability of 1:10,000 can be attained. Partial successes at PdBI, VLA, and ALMA/SMA suggest they are viable.