Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.mrao.cam.ac.uk/~eic22/Papers/harp-acsis.pdf
Дата изменения: Wed Jul 22 18:59:12 2009
Дата индексирования: Tue Oct 2 04:24:59 2012
Кодировка:
Поисковые слова: m 103

Mon. Not. R. Astron. Soc. 000, 118 (2009)

Printed 22 July 2009

A (MN L TEX style file v2.2)

HARP/ACSIS: A submillimetre spectral imaging system on the James Clerk Maxwell Telescope
J.V. Buckle1 , R. E. Hills1,2 , H. Smith1 , W. R. F. Dent3,2 , G. Bell1 , E. I. Curtis1 , R. Dace1 , H. Gibson1 , S. F. Graves1 , J. Leech1,4 , J. S. Richer1 , R. Williamson1 , S. Withington1 , G. Yassin1 , R. Bennett3 , P. Hastings3 , I. Laidlaw3 , J. F. Lightfoot3 , T. Burgess5 , P. E. Dewdney5 , G. Hovey5 , A. G. Willis5 , R. Redman6 , B. Wooff6 , D.S. Berry4 , B. Cavanagh4 , G.R. Davis4 , J. Dempsey4 , P. Friberg4 , T. Jenness4 , R. Kackley4 , N. P. Rees4 §, R. Tilanus4 , C. Walther4 , W. Zwart4 , T. M. Klapwijk7 , M. Kroug7 , T. Zijlstra7 1
2 3 4 5 6 7

arXiv:0907.3610v1 [astro-ph.IM] 21 Jul 2009

Cavendish Astrophysics Group, Cavendish Laboratory, University of Cambridge, J J Thomson Ave., Cambridge CB3 0HE, UK ALMA JAO, Av. El Golf 40 - Piso 18, Las Condes, Santiago, Chile UK Astronomy Technology Centre, Blackford Hill, Edinburgh, EH9 3HJ, UK Joint Astronomy Centre, 660 N. A'ohoku Place, Hilo, HI, 96720, USA Dominion Radio Astrophysical Observatory, PO Box 248, White Lake Herzberg Institute of Astrophysics, 5071 West Saanich Road, Victoria, BC, V9E2E7, Canada Delft University of Technology, Faculty of Applied Sciences, Kavli Institute of Nanoscience, Lorentzweg 1, 2628 CJ Delft, The Netherlands

ABSTRACT

This paper describes a new Heterodyne Array Receiver Programme (HARP) and AutoCorrelation Spectral Imaging System (ACSIS) that have recently been installed and commissioned on the James Clerk Maxwell Telescope (JCMT). The 16-element focal-plane array receiver, operating in the submillimetre from 325 to 375 GHz, offers high (three-dimensional) mapping speeds, along with significant improvements over single-detector counterparts in calibration and image quality. Receiver temperatures are 120 K across the whole band and system temperatures of 300K are reached routinely under good weather conditions. The system includes a single-sideband filter so these are SSB figures. Used in conjunction with ACSIS, the system can produce large-scale maps rapidly, in one or more frequency settings, at high spatial and spectral resolution. Fully-sampled maps of size 1 square degree can be observed in under 1 hour. The scientific need for array receivers arises from the requirement for programmes to study samples of objects of statistically significant size, in large-scale unbiased surveys of galactic and extra-galactic regions. Along with morphological information, the new spectral imaging system can be used to study the physical and chemical properties of regions of interest. Its three-dimensional imaging capabilities are critical for research into turbulence and dynamics. In addition, HARP/ACSIS will provide highly complementary science programmes to wide-field continuum studies, and produce the essential preparatory work for submillimetre interferometers such as the SMA and ALMA. Key words: instrumentation: detectors instrumentation: spectrographs methods: observational techniques: image processing techniques: spectroscopic submillimetre

E-mail: j.buckle@mrao.cam.ac.uk Present address: Oxford Astrophysics Group, Denys Wilkinson Building, Keble Road, Oxford, OX1 3RH Present address: Department of Astronomy, Columbia University, Pupin

Physics Laboratories, 550 West 120th Street, New York, New York 10027, USA § Present address: Diamond Light Source Ltd, Harwell Science and Innovation Campus, Oxfordshire, OX11 0DE

2

J.V. Buckle, R. E. Hills, H. Smith et al.
switching by using appropriately selected chop and node distances, thereby increasing the mapping speed of compact objects. The full ACSIS capabilities are described in Sec. 3. When used with HARP, ACSIS offers either wide bandwidths (up to 1.9 GHz for each of the 16 IF channels), or high spectral resolution (with a channel spacing as small as 31 kHz, or 0.03 km s-1 ). In addition, it can provide one or two sub-bands per IF channel, allowing simultaneous observations of multiple lines within the HARP 5 GHz IF frequency. One example is observing the J =32 lines of 13 CO and C18 O with two separate high-resolution sub-bands. ACSIS and the upgraded JCMT Observatory Control System allow for rapid observing and data taking. Fast scanning enables data to be taken continuously at up to 10 Hz, and it is possible to make fully-sampled maps with HARP of 1 square degree in less than 1 hour. As shown in Fig. 1, HARP has been installed on the righthand Nasmyth platform of the JCMT. An optical relay (Fig. 2) provides the field of view and imaging performance to match the HARP specifications. The relay optics also bring the beam down to a lower level so that the cryostat is conveniently accessible on the Nasmyth platform. A schematic diagram of the components of HARP is shown in Fig. 3. The path is further explained in Sec. 2.2. Each of the components is described in the following sections, with Sec. 2 describing the HARP components, and Sec. 3 describing the ACSIS components. Designing and building HARP was a collaborative project between the Cavendish Astrophysics Group in Cambridge, the UK Astronomy Technology Centre in Edinburgh, the Herzberg Institute for Astronomy in Victoria, Canada, and the Joint Astronomy Centre in Hawaii, with the Kavli Institute of Nanoscience at Delft contributing the critical SIS junctions. ACSIS was built in collaboration between the Dominion Radio Astrophysical Observatory in Penticton, Canada, the UK Astronomy Technology Centre and the Joint Astronomy Centre.

1 INTRODUCTION The James Clerk Maxwell Telescope (JCMT), situated on a high, dry site near the summit of Mauna Kea, and with a 15 m dish, is the largest submillimetre observatory in the world. The submillimetre band is rich in molecular lines, and so high-resolution spectroscopy at these wavelengths enables studies of space densities, velocities, chemical structure and excitation in the gaseous material of both galactic and extra-galactic sources. The majority of these sources are extended on scales larger than the JCMT beam size of 14 arcsec (at 345 GHz), and many astronomical targets are much larger, extended on parsec scales, which frequently need to be mapped to carry out relevant astronomical research. With submillimetre heterodyne receivers approaching background-limited performance, the scientific need for focal-plane heterodyne array receivers is clear. By building multiple detectors, mapping speeds can be increased, making possible programmes which observe samples of objects of useful statistical size. HARP (Heterodyne Array Receiver Programme) and ACSIS (Auto-Correlation Spectral Imaging System) have recently been installed and commissioned on the JCMT. HARP operates in the submillimetre band spanning 325 to 375 GHz, a frequency range which contains transitions from nearly all the most abundant molecules in interstellar gas. The key scientific programmes that are expected to be carried out with the new system are · surveys of the distribution and properties of molecular clouds in face-on and edge-on galactic disks, starburst and Seyfert galaxies, and interacting galaxy systems; · large-scale unbiased surveys of star formation and outflows in molecular clouds; · large-scale studies of hierarchical structure and clumping in molecular clouds; · studies of the chemical processing of molecular clouds by shocks and turbulence; · surveys of the dynamics and excitation of molecular gas in the Galactic centre; · deep, narrow surveys of molecular gas in Solar System objects; · studies of the chemical and physical state of comets near perihelion. Several large JCMT legacy survey projects have been developed to exploit this new instrumentation in the above scientific areas (Ward-Thompson et al. 2007; Plume et al. 2007; Matthews et al. 2007; Wilson et al. 2009). The combination of HARP and ACSIS makes possible science which is highly complementary to the widefield continuum studies of cold dust in distant galaxies and the earliest stages of star formation that have been carried out by SCUBA (Holland et al. 1999), and will continue with SCUBA-2 (Holland et al. 2006). It will also provide the essential scientific preparatory work for submillimetre interferometers such as the SMA and ALMA. In addition to the increase in mapping speeds, HARP/ACSIS offers supplemental benefits, the most important of which is probably the decrease in calibration errors and increase in image quality, since multiple detectors observe the source simultaneously. Relative calibration of individual detectors is more accurate, as data are taken simultaneously through the same atmospheric path. By making overlapping maps, pointing drifts between maps can be removed. For objects smaller than the array field of view, edge detectors can be used to make reference spectra through on-array beam

2

HARP DESIGN CONCEPTS

A description of the key design features of the instrument, along with technical and organisational overviews, have previously been published (Smith et al. 2003, 2008). In this section we present a brief overview of the key features of HARP's specification, design and performance. 2.1 Overview

HARP comprises 16 detectors laid out on a 4в4 grid, with an onsky projected beam separation of 30 arcsec. At 345 GHz the beam size is 14 arcsec, and the under-sampled field of view of HARP is 104 в 104 arcsec, as shown in Fig. 4. In a single pointed observation, the map is under-sampled by a factor of 4.4 in area, and by a factor of 17.6 with respect to the Nyquist frequency ( /2D). Beam rotation, using a new K-mirror (Sec. 2.2.1), optimizes the observing efficiency. For objects similar to or smaller than the array field of view, beam rotation ensures no edge detectors miss the source emission. For larger objects, and particularly for scan mapping, the array can be orientated with respect to the scan direction to provide good sampling (Sec. 4). HARP is easily and rapidly tuned across the operating range of 325 to 375 GHz using automated control software in 1030 seconds (Sec. 3.4). The IF frequency is 5 GHz, and, in conjunction with ACSIS, HARP offers up to 1.9 GHz of bandwidth.

HARP/ACSIS on the JCMT
M4 Encoder Warm load Cal2 Cal1

3

Axis of rotation Cold load G1 Pupil on cold stop M5
Figure 2. Diagram of the relay optics from the encoder to the cold stop, through a series of labelled mirrors.
Warm IF Amplifiers ACSIS Correlator

Cold IF Amplifiers Imaging Array

K Mirror JCMT Tertiary Mirror

Calibrator

(Sideband Filter)

Instrument Cryostat

LO

Figure 3. A schematic diagram of the components of the HARP receiver system, adapted from Leech (2000).

Figure 1. HARP shown on the right-hand Nasmyth platform of the JCMT, and a zoomed-in image of HARP and the instrument electronics box.

HARP uses single sideband (SSB) filtering to minimize the system temperatures and improve calibration accuracy. HARP sensitivities have exceeded expectation, with the combination of receiver noise temperature and beam efficiency better than the required 330 K (SSB). Relative inter-receptor calibration, measured on continuum sources, is accurate to better than 5 per cent, and relative beam positions can be measured to an accuracy of 1 arcsec. HARP utilizes a number of innovative features across all elements of the design. The key features of these are outlined below. Further technical details and schematic diagrams are presented

Beamwidth (14" @ 345 GHz)

Beam spacing (30")

Undersampled field of view (104" @ 345 GHz)

Figure 4. Schematic diagram of the HARP grid layout.

Control Electronics LO Injection Unit

Cold Optics
Ў Ў Ў Ў Ў Ў Ў

Chapter 5: HARP-B Calibration Unit

4

J.V. Buckle, R. E. Hills, H. Smith et al.
M1 M3

Virtual Image

Real Image

M2

Encoder

K-mirror

Figure 5. Schematic of the K-mirror, showing the three mirrors, M1 to M3. The JCMT tertiary mirror is just to the left of the real image.

TMU

JCMT 'Y' AXIS

Figure 7. The calibration system (from Williamson 2004).

Figur0.5 per cent at of thewavelength, which unit layout. The line source the e 5.1: Drawing this HARP-B calibration will contribute 3 K to is injected into the page.

Figure 6. Schematic overview of the K-mirror in the JCMT receiver cabin. The JCMT `Y' axis is the elevation axis.

elsewhere (Smith et al. 2003; Leech 2000; Williamson 2004; Smith et al. 2008; Bell 2008).

2.2 Optical system The optical system for HARP has three sections. The K-mirror is inside the receiver cabin and described in Sec. 2.2.1. The warm optics, comprised of the relay optics and calibration system, are exposed on the Nasmyth platform, and are described in Sec. 2.2.2. The cold optics are contained inside the cryostat, and are described in Sec. 2.2.3.

2.2.1 K-mirror The K-mirror (Fig. 5) consists of three large powered mirrors with modest curvature. The system acts as an image and polarization rotator, and forms an image of the secondary mirror at a point near the elevation encoder. An overview of the K-mirror in the JCMT receiver cabin is shown in Fig. 6. The K-mirror also maximizes the field of view available to instruments on the Nasmyth platform, which otherwise would be severely limited due to the relatively small hole in the encoder axis. The K-mirror rotates as a unit about the elevation axis of the telescope, and has a rotational range of ±57.5 degrees, limited through control software to ±55 degrees, with the effective rotation of the image and polarization twice the K-mirror rotation angle. The K-mirror can rotate at a rate of 5 degrees s-1 , with a rotational position accuracy of ±0.1 degree. The design allows for a field of view of 200 arcsec. The system is designed to have very low losses and aberrations. Spillover loss is designed to be less than 1 per cent, and in fact is expected to be < 0.5 per cent at 850 µm. The losses per mirror should be less than

system temperature from each mirror. In order to meet the needs is cut of possible future JCMTees causing the forward edge to behave at an angle of 20 degr instrumentation, the mirror surfaces horizontal. This help been finished so that they are compatible with the highest operato reduce the effect of standing waves. tional frequencies at the JCMT of 870 GHz. Rotation of the K-mirror is controlled by the Telescope ConThe heating source is provided by a 12 V, 18 Wcontinuous rotation of is placed directl trol System (TCS). The TCS commands a power resistor which the K-mirror to compensate for the is via in parallactic angle 11 on the rear surface of the load. Cooling changean off the shelf 12 V,of W Peltier coolin the source and the The Peltier is linked to the load telescope. device from Supercool. rotation of the elevation axis of the via a small I beam and hea When the telescope is commanded to slew to a new position, the sink compound. A drawing the temperature loads is shown in Figure 5.2. TCS computes an optimal orientation of the focal plane relative to the tracking coordinate system to give the longest tracking time on The thermal beforeol of the unit is provided by an off-the-shelf Eurotherm proportiona source contr hitting one of the K-mirror rotation limits. This optimal ferential (PID) controller The output from which which integral diforientation corresponds to .an initial K-mirror angle to is passed through a soli it is which controls the 12 volts required to drive both the heating resister an state relayslewed and then the TCS continuously updates the angle to maintain a fixed orientation between the focal plane and tracking coordinate systems throughout the observation. The pointing offsets due to possible misalignment of the Kmirror axis and its axis of rotation have been calculated, and in166 cluded in the JCMT pointing model. The effective pointing of the telescope on the sky is changed if the mirrors are displaced or tilted, so a requirement of the K-mirror is that total pointing errors are less than 10 arcsec. Residual pointing errors, after subtraction of the known correction terms, are a few arcsec, partly due to the number of new terms in the pointing model introduced by the K-mirror. 2.2.2 Calibration

A schematic overview of the calibration system is shown in Fig. 7. The calibration unit was built as an independently mounted unit containing the two mirrors, Cal1 and Cal2, the warm and cold loads, and a spectral-line calibration signal (Fig. 2). Cal1 is strongly concave and fixed, while Cal2 is flat, and can turn from the warm load to the cold load in under 2 s. The loads are isothermal cuboid cavities lined with absorbing tiles that have high efficiency at the HARP operational frequencies. A thermally isolated conical reflecting baffle at the input limits the IR coupling. The positions of the loads were chosen to minimize convection. To make calibration measurements, a lower relay mirror (M5,

HARP/ACSIS on the JCMT
Fig. 2) is tilted mechanically to point to the calibrator unit, where the Cal2 mirror focuses the beams on to either the warm or cold load. The Cal2 mirror is actuated by a standard four-position "Geneva" mechanism, while the M5 mirror is actuated by a novel Geneva design of a two-position type. The Cal1 mirror then rotates to switch between the loads. The warm load is heated by resistors to temperatures 40 K above ambient. The cold load is cooled to 10 K below ambient temperatures using a Peltier cooler, with the intention of providing a load that is at approximately the same temperature as the water vapour in the atmosphere. Two- or three-load calibration measurements can be taken in a few seconds. A twoload calibration, which measures the power from the sky and the power from the cold load, should then accurately remove atmospheric attenuation. The three-load calibration, which additionally measures the power from the warm load, provides a measure of the receiver temperature. The Cal2 mirror has a third position which allows it to point at the spectral-line calibration signal. This coherent signal is used to check the tuning of the interferometer (Sec. 2.2.3) and to optimize the signal-to-noise ratio when adjusting tuning parameters such as the bias voltage and local oscillator (LO) level. It uses a YIG oscillator and a multiplier to produce a narrow line which can be tuned to lie anywhere in the HARP operational frequency range. The optics are arranged to couple this signal into all the mixers at the same time with a reasonable degree of uniformity. The night-to-night calibration accuracy has been measured to be within JCMT guidelines of 20 per cent, and is usually better than this (Sec. 5.4). Work on improving the accuracy and stability of the combined HARP, ACSIS, telescope and software systems continues at the JCMT. The receptor-to-receptor calibration accuracy was measured with observations on the nearly full Moon (Fig. 18), and indicate maximum differences between levels less than 5 per cent. Any variation of calibration with ambient temperature is at levels below both the night-to-night and receptor-to-receptor calibration accuracy. 2.2.3 Cold optics The cold optics are located inside the cryostat, and operate at temperatures 60 K. Four powered mirrors (Fig. 8) between the encoder and the array are optimized together to take advantage of aberration balancing. Utilizing a reflective, slightly curved cold stop inside the cryostat, the system forms an image of the sky with low aberration at the detector array. The smallest mirror, C1, forms the cold stop, while C2F and C2M create an image of the sky at the mixer array. A polarizing Mach-Zehnder interferometer is used for the sideband separation and the image sideband is terminated in a cold load, inside the cryostat. C2M is moved to tune the interferometer, and the full range of movement for this tuning can be achieved in under 10 s. The HARP IF frequency of 5 GHz sets the path difference requirement for the two beams through the interferometer of 15 mm. Leakage and cross talk levels between the beams were looked for during on-sky commissioning, and were not detected at the 1 per cent level. The interferometer is tuned so that the path difference is a whole number of wavelengths for the image frequency. This means that the path difference of the signal frequency, once it has completed transmission through the system, is an odd number of halfwavelengths. This results in the signal in the desired sideband being coupled to the mixers and in the other sideband the mixers are coupled to the SSB load. The tuning range achieved covers 324 GHz to
Mixer array

5

C2F

G4 G3

Cold Load C3 G2 G1 C2M Beam from relay

Window

C1

Figure 8. Overview of the cold optics, showing the four mirrors (Cn) and the four grids (Gn) of the polarizing Mach-Zehnder interferometer.

376 GHz. The grids are made from gold-plated tungsten wire with diameter 10 µm and spacing 25 µm. G1 and G4 are rectangular, while G2 and G3 are circular. The two sets of grids are aligned to split the signals into ±45 degree components. The image sideband dump provides a termination for the beam transmitted through G1, and consists of a concentrator mirror, C3, which operates at 60 K, and a thermally-isolated cold load (the SSB load), which is cooled to 12 K. The details of the design and laboratory testing of the cold optics systems are fully explained in Williamson (2004) and Bell (2008). The reflective cold stop, C1 is surrounded by cold absorber, which is typically at 18 K. The cold optics also truncate the sidelobes of the feed pattern, so they see cold absorber inside the cryostat rather than thermal emission from the region around the elevation encoder. The interferometer is the critical item in the cold optics, and provides the datum position to which the other components are referenced, with the cryostat window providing the interface to the rest of the system. Sideband separation commissioning tests were carried out across the observing bandwidth, and were measured to be better than 19 dB on average (Sec. 5).

2.3

Imaging array

The imaging array unit comprises an array of mylar beam splitters for LO injection, two decks of eight horn-reflector antennas holding 16 SIS mixers with air-cored coils for Josephson current suppression, and 16 HEMT cold IF amplifiers with isolators and bias tees (a T-shaped multiplexer). The horn reflector antennas consist of split-block, directly machined corrugated horns with ellipsoidal reflectors. The corrugated horns were designed to be easy to mass produce using direct machining into a split block. The novel horn design features constant depth corrugations and delivers good beam patterns across the required bandwidth while avoiding difficult to machine /2 deep corrugations near the throat of the horn. Ellipsoidal reflectors were chosen to provide efficient coupling of the horns to both the telescope fore-optics and the LO meander line. The imaging array unit is shown schematically in Fig. 9. The array is mounted inside the 20 K heat shield.

6

J.V. Buckle, R. E. Hills, H. Smith et al.

Figure 11. Photograph of the assembled LO coupler, also showing one mirror of the LO relay and the stop. Figure 9. The HARP imaging array.

LO Horn
LO input

Feed horn boresights
1 2 3 4

ўЎ

Upper level horns
14 mm

#$

ё

ё¤

5

6

7

8

©

§

§Ё

Ґў¦ Ґ

Lower level horns
ў

ў

9

10

11

12

13

14

15

16

Upper level horns

!"

absorber

14 mm

Parabolic corner mirrors

Figure 13. Overview of the radial probe SIS mixer devices.

Figure 10. Plan view of the LO coupler.

2.3.1 The local oscillator coupler The LO coupler uses a `meander line' to provide the efficient coupling necessary to pump 16 mixers (Fig. 10). At each crossing point there is a 45 degree beam-splitter made of thin mylar film. The mylar in the beam-splitters was pre-tensioned to ensure it remains tight when cooled. In order to prevent the beam from diverging due to diffraction as it passes through the meander line, the roof mirrors that are used to turn the beam around after passing in front of each set of 4 mixers consist of off-axis paraboloids. These produce beam waists at the mid-points of each path. This arrangement has a large RF bandwidth, and injects LO power in a highly efficient manner (Leech 2000). For ease of maintenance and upgrades, this unit is independent of the cold optics, with the LO plate mounted on the door of the cryostat. The LO is generated outside the cryostat (Fig. 3), and injected into it through a small polypropylene window. To minimize adjustments with the operational system, a relatively slow beam is generated with optics on the LO plate, which is refocused with a mirror attached to the array unit before it enters the meander line. To provide the correct polarization, two mirrors are used on the LO plate. A small relay consisting of flat and off-axis paraboloid mirrors outside the cryostat and a further small paraboloid mirror inside the cryostat carries the beam from the horn on the LO plate

to the LO coupler. The adjustment to bring the LO beam into alignment is accomplished using pusher screws to tilt and shift the two mirrors. The LO coupler is machined from aluminium. The stop is a 12 mm hole in an absorbing plate which is used to clean up the beam (Fig. 11). Diagnostics are provided by observing the pumped I/V curves on the mixers to obtain as uniform a distribution of LO power as possible. An image of the I/V and P(IF)/V curves displayed by the HARP control software during operation is shown in Fig. 12.

2.3.2

Radial probe mixers

The radial-probe SIS mixers were designed and developed at the Cavendish Astrophysics Group (Leech 2000; Withington et al. 2001) and fabricated in the Kavli Institute of Nanoscience at Delft. The design, shown in Fig. 13, gives extremely broadband operation with similar characteristics for each device. A single-sided radial probe couples power from the rectangular waveguide to the SIS junction. These radial probes have been shown, using scale model experiments and modelling, to give a good impedance match to typical SIS junctions (Leech 2000; Kooi et al. 2001). The capacitance of the SIS junction is tuned out using an inductive microstrip stub end loaded with a radial stub to present a short circuit at the end of the microstrip.

HARP/ACSIS on the JCMT

7

Figure 12. The HARP control system I/V and P(IF)/V curves display.

Figure 14. Photograph of the HARP horn near the throat, demonstrating the excellent machining quality. The grooves on the horn have a width of 0.1±0.01 mm.

magnetic fields. If these vortices are close to the SIS junction, this can result in a change of the magnetic field applied to it. The magnetic field is then no longer at the value needed for supercurrent suppression and Cooper pairs begin to tunnel, leading to increased noise and decreased stability in the mixer. With large arrays, the movement of trapped flux is potentially a major problem, and one of the reasons why mixers require different amounts of LO power to achieve optimum performance. In HARP, the mixers feature arrays of tiny holes in the niobium (Nb) film near the SIS junction to prevent the movement of trapped flux (Leech 2000). If trapped flux is still proving problematic, small resistive heaters can be used to briefly raise the array above the critical temperature for Nb, destroying the trapped flux vortices. As a routine start up procedure, the telescope operator runs the `all heaters' command (top left of the image in Fig. 12), which briefly warms the array. While this process is occurring, the changes to the I/V and P(IF)/V curves can be monitored in real time. For more localized problems, individual heaters can be used on 4 of the detectors at a time. The control software screen shown in Fig. 12 is thus an extremely useful diagnostic tool, and can be used to check the status and stability of the HARP mixers throughout an observing run.

The imaging receptor is based on the horn-reflector antenna each mixer consists of a corrugated waveguide horn with a reflector mirror at its aperture. The feed-horn, waveguide, device and IF slots which form the mixer were manufactured as split aluminium blocks, then gold-sputtered. This gives blocks that are lightweight, low thermal mass, free from waveguide flanges and straight-forward to machine. A photograph of one of the finished HARP horns is shown in Fig. 14. One problem that can arise is the unwanted movement of trapped magnetic flux near the SIS junctions. Typically the unwanted Josephson tunnelling of Cooper pairs is suppressed using a coil to apply a specific strength of magnetic field across the SIS junction. Unfortunately, quantized vortices of magnetic flux, usually trapped by defects in the superconducting film near the SIS junction, can move around under the influence of external electro-

2.3.3

IF system

The IF system for each receptor consists of a bias tee, isolator and cold IF amplifier connected to the 4 K stage, and two warm IF amplifiers separated by a bandpass filter and a fixed level setting attenuator mounted outside of the cryostat (Fig. 9). The co-axial interconnecting cable from the 4 K stage to the wall of the cryostat is made of stainless steel to minimize the heat load on the 4 K stage. An IF frequency of 5 GHz, with a minimum bandwidth of 1.6 GHz, is used. The overall mid-band gain is a minimum of 57 dB, with a noise temperature 7 K.

8

J.V. Buckle, R. E. Hills, H. Smith et al.
ent IF filters, the philosophy is to have only two hardware options, and ensure that these cover most astronomical requirements. Each DCM can therefore switch between a wideband (1 GHz) and a narrowband (250 MHz) mode (see Table 1). In practice, the edges of the DCM filtering limits the bandwidth of each DCM to 930 MHz and 220 MHz respectively these are the worst case, the -3 dB power points of the DCMs. After anti-alias filtering, the total power (TP) signal is measured and the RF signal is then fed through an automatic level control (ALC) circuit. The ALC is used to maintain constant TP into the samplers, in order to ensure the most stable system bandpass (and the flattest baselines), and the optimal digitization threshold. The TP signal is recorded and re-applied to the data stream in the ACSIS real time data reduction. The DCMs use MMIC architecture and miniaturized components, and plug into a standard VME backplane. A second section of the backplane includes SMA "blind-mate" connectors for the RF analogue signals, allowing for simple field replacement of the whole unit. The IF systems are mounted within temperature-controlled racks, using a liquid glycol cooler system and external heat exchanger. On board each DCM unit are temperature probes and a heater for internal temperature control. In this way temperatures within the DCMs are maintained to better than