nick@sron.rug.NL
Space Research Organisation Netherlands (SRON), PO Box 800, 9700 AV Groningen, The Netherlands.
The payload of FIRST, ESA's forth cornerstone mission, will consist of two direct detection instruments and one heterodyne instrument. In this paper we discuss a conceptual design for the heterodyne instrument based on the work of a group of researchers from interested European and US institutions. The instrument covers the frequency interval 480 - 1200 GHz in four sub-bands with SIS mixers to provide state of the art sensitivity. Each sub-band will have two mixers operating on orthogonal polarisations for increased sensitivity and to provide redundancy. The total IF bandwidth of at least 8 GHz will be analysed by acousto-optics spectrometers with a resolution of 1 MHz. Higher frequency resolution could be provided by a digital autocorrelation spectrometer. We have attempted to optimise the design of the instrument to give maximum performance for the envisaged astronomical observations within the constraints set by spacecraft resources and this will be discussed in the paper.
Keywords: FIRST, satellite, astronomy, SIS receiver.
ESA's 4th cornerstone mission of its Horizon 2000 science programme is the Far IR and Sub-mm Space Telescope - FIRST satellite. FIRST is an astronomy mission to allow access to the last major part of the electromagnetic spectrum still mainly unobserved, namely the wavelength interval 100 to 1000 microns. Most of this spectral region is not accessible from the ground due to strong absorption in the atmosphere.
This region of the spectrum is thought to be very important for studies of a wide range of astronomy objects, apart from the possibility of discovering new astrophysical processes. Fields of study identified as likely to benefit from observations with FIRST include
The FIRST science and mission are described in the so called Red Report [1]. A review of the detailed scientific objectives is currently in progress but this is not expected to affect the optimum instrument design significantly.
FIRST will be an observatory mission lasting 3 or 6 years depending on the type of cryogenic system employed. Its scientific payload will comprise three elements: photoconductor and bolometer direct detection instruments, and a heterodyne instrument (HET) based on SIS mixer technology. The direct detection instruments will provide low to high resolution spectroscopy (R<104) and photometry over the wavelength range 85 - 1300 microns and are described elsewhere in this Symposium. The heterodyne instrument will provide very high resolution spectroscopy (R=104 - 107) over the frequency interval 490 to 1170 GHz (250 - 625 microns).
During the feasibility studies of FIRST and in preparation for the mission selection in 1993, a Payload Working Group was formed to determine the likely requirements on spacecraft resources of the envisaged compliment of scientific instruments. For this task a model payload was devised which would fulfil the science requirements as listed in the Red Report [1]. This model payload was revised a number of times during subsequent studies to accommodate changes in the spacecraft design and the latest version is described in the Payload Description Document [2].
In the case of the heterodyne instrument, it was realised that the model payload design adopted was not optimised for the science mission. Thus during late 1995 a team of interested scientists in the US and Europe began working to define an optimal instrument within the limitations of the satellite. The group adopted the name HIFI Heterodyne Instrument for FIRST and its members are listed at the end of this paper. This work has resulted in the instrument concept which is the subject of this paper. Further work will be required to finalise the design.
In Section 2 we outline science requirements on, and the limited satellite resources available to, the instrument. In Section 3 we discuss the optimisation of the instrument design and in Section 4 we outline the characteristics of the solution.
Full frequency coverage is required from the CI line at 492 GHz to the para-H2O line at 1113 GHz. A modest increase in the frequency range would include an O2 line at 490 GHz and an HF line at 1230 GHz. Full frequency coverage is desired for completeness and to allow comparison with ground-based and airborne observations in the atmospheric windows.
Clearly, in order to make full use of the huge investment in the satellite we should aim to achieve the highest possible sensitivity. It is for this reason that SIS mixers are selected.
Single sideband receiver operation was considered essential to remove ambiguity over the identification of lines in crowded spectra. However, it appears feasible to perform the sideband separation off-line by observing the same frequency region at a number of different local oscillator frequencies [3,4].
Table 1 lists the expected minimum bandwidth and resolution requirements for a range of astronomical targets and for the range of frequencies covered by HET. For line searches one should have as many channels as possible but in all other cases no more than 1000 resolution channels are required.
Table 1. Expected FIRST bandwidth and resolution requirements for a number of astronomical targets (priv. comm. P. Encrenaz).
| Target | Bandwidth | Resolution |
|---|---|---|
| Planets | 4 GHz | 4 MHz |
| Comets | 0.2-1 GHz | 0.1-4 MHz |
| I.S.M. | 0.24 GHz | 0.1-4 MHz |
| Galaxies | 4-10 GHz | 4-10 MHz |
| High Z objects | 4-10 GHz | 4-10 MHz |
| Primordial molecules | 4-10 GHz | 100 MHz |
| Line Survey | 4-10 GHz | 410 MHz |
The spacecraft is divided into three sections: The telescope module comprising the 3 m diameter, F/9.59 Cassegrain telescope and sunshields; the payload module (PLM) with the cryogenic system containing the focal plane instruments; and the service module (SVM) containing the control electronics in a room temperature environment.
ESA is currently pursuing two alternative concepts for the spacecraft: one based on a He cryostat similar to ISO, and one based on closed-cycle mechanical coolers developed by British Aerospace. The scientific payload must be compatible with these two options. Both designs will provide an optical bench at between 10-25 K on which the focal plane instruments will be mounted. Cooling of the detectors to 4.5 K (and optionally to 1.8 K in the case of the He cryostat) will be provided via thermal straps. The total heat lift available to the three instruments is listed in Table 2.
The volume available to the three focal plane instruments is limited and the chosen division of the focal plane between the three instruments is shown in the PDD [2]. The HET can occupy a volume of about 800x400x400 mm. The adopted placement constrains the field of view of the HET to be less than 15' x 10'. and the maximum practical throw of a focal-plane chopper is about 5'.
Table 2. FIRST total cryogenic heat-lift capabilities. Note that in the case of a He cryostat the "25 K" stage will actually be at about 10 K and that cooling to 1.8 K will also be available.
| Stage | Heat-lift, mW |
|---|---|
| 25 K | 120 |
| 4.5 K | 15 |
| 1.8 K* | 5 |
In recent years there have been remarkable improvements in the sensitivity and maximum frequency of SIS mixers [5], now almost exclusively based on all-niobium junctions. These devices have much lower local oscillator power requirements than Schottky mixers, but only operate at temperatures below about 6 K. The local oscillator power requirement for SIS mixers rises to approximately 10 mW at 1 THz. Although SIS mixers can operate at 6 K, maximum sensitivity is achieved by cooling below 4 K.
The RF bandwidth of SIS mixers in the sub-mm region is currently about 100-200 GHz. Therefore, a minimum of 4 mixers will be required to cover the desired frequency range.
One of the major challenges with the HET will be the construction of the LO system to cover the full frequency range. The bandwidth of current solid-state sources will have to be increased by more than a factor of 2, although the power available is sufficient. The LO unit will be placed outside the cryostat on a radiator to remove the several Watts of heat dissipated. The LO beam will pass through a window in the cryostat wall and into the focal plane unit.
The LO frequency must be phase locked to a stable reference oscillator to achieve the narrow linewidth and high stability required for observations at 100 kHz resolution.
At least 20 dB gain is required close to the mixers to overcome the losses in the cables leading the IF signals to the SVM. In fact, due to the long cable harness, it will be necessary to have further amplification close to where the harness emerges from the cryostat.
Currently available (commercial) HEMT technology for use in the cryogenic IF amplifiers has a power dissipation of about 5 mW per 10 dB of gain and, therefore, we have assumed an IF dissipation of 10 mW per IF channel. New devices currently under development reduce this dissipation to under 4 mW.
It is clear that the cryogenic IF amplifier dissipation will be too high to be placed at 4 K and must be operated. Taking the conservative dissipation figure of 10 mW per IF channel and the limited heat-lift available it appears that a maximum of 4 operating IF channels is allowed.
Cryogenic IF amplifiers with bandwidths of 10 GHz are available. The remainder of the IF chain is not expected to limit the astronomical performance of the receiver.
There are two spectrometer technologies which appear relevant for FIRST.
The acousto-optic spectrometer (AOS) has been used at ground-based observatories for a number of years and will fly shortly on two astronomical space missions: NASA's SWAS and Sweden's ODIN satellites. The AOS provides wide-band spectrometers with about 1000 resolving channels. Current technology limits the maximum bandwidth to 2 GHz but by combining AOS modules much larger bandwidths can be covered (see Lecacheux et al, and Schieder, these Proceedings). AOS' typically have a power consumption of about 10 W and a mass of 7 kg, including IF amplifiers and readout electronics.
The second type of spectrometer we consider is the digital autocorrelation spectrometer (see Emrich et al, these Proceedings). These offer a flexibility in bandwidth and resolution not matched by other technologies and are potentially much more compact. However, at the moment the power consumption of a broadband DACS is significantly higher than for an AOS. A DACS is scheduled to fly on ODIN in 1998.
Note that it is feasible to combine several spectrometer modules using a hybrid approach to achieve a very broadband, highly redundant, and flexible spectrometer.
Mainly due to the limits on thermal conduction along IF cables, though also from constraints on volume and mass, it appears that a maximum of about 10 mixers and IF channels could be accommodated in the HET. This taken with the present bandwidth of SIS mixers implies that it is possible to have a maximum of two mixers at any given frequency.
There are three main observing modes to consider: line survey, single frequency observations of weak sources, and mapping.
For a line survey, the observing speed is proportional to the total instantaneous bandwidth, independent of the number of operating mixers. Therefore, in this case the total spectrometer bandwidth should be maximised, while maintaining the required resolution. Here the limitation will be the bandwidth of the spectrometer or the telemetry downlink, and there is no advantage in observing both polarisations or at two frequencies. However, due to the IF system architecture it may be more practical to operate two or more channels simultaneously.
Since the targets of a line survey are likely to lie in large, dense interstellar clouds it may not be possible to use the internal focal plane chopper. Therefore, we expect to have to use position switching to obtain reference spectra.
In this case we require the maximum possible sensitivity for a point source. Therefore, we should maximise the number of mixers observing the source. Observing both polarisations is a simple way to implement two coaligned mixers observing one source.
If it is feasible to have more than two mixers operating at one frequency then it is possible to interleave them in time using a focal plane chopper (group A observes sources while group B observes reference and vice versa). This time multiplexing scheme allows 4 mixers to observe the source but only works well for sources that are smaller than the chopper throw.
If it is required to observe the source at two widely separated frequencies then a dichroic beam splitter can be used to divide the single beam from the source between two pairs of mixers giving 4 mixers observing the source simultaneously. One penalty of this scheme is the loss in the dichroic which may be about 5 %.
We assume that the sources to be mapped are larger than the HET field of view or separation between mixer beams on the sky. For mapping at a single frequency we have the same goal as above in Section 3.3, namely to maximise the number of operating mixers. However, it is not necessary that the mixer beams are coaligned since the region to be mapped is large. Conversely, two mixers in an imaging array configuration has no advantage over two mixers in a dual polarisation configuration.
We believe that the on-the-fly technique is the most efficient one to be used for mapping. In this mode the integration time should be short, of order 1 s, to minimise the effects of drifts and this has implications for the spectrometer read-out and the satellite telemetry rate. However, in this mode it is unlikely that it will be necessary to use the full spectrometer bandwidth.
Below we list the important characteristics of the HET as derived from the above considerations.
The HET will provide dual-polarisation coverage of the required frequency range. This gives a root-2 sensitivity advantage when observing a single frequency and a point source. It also naturally provides redundancy at all frequencies. To cover the full frequency range will require about 10 mixers and IF amplifier channels.
To achieve the maximum possible sensitivity a fast focal plane chopper will be included. This is a potential single point failure mechanism (though, it may be feasible to include a redundant chopper in the optics system) and hence must be designed for highest reliability and to fail safe. Such mechanisms have been developed for ISO.
A calibration unit must be included in the optics system. The simplest way to implement this is to use the chopper scanning mechanism to direct the beam onto a black body calibration source.
The HET will not have a singe-sideband filter for the following reasons. Firstly, by observing DSB a root-2 increase in sensitivity is achieved by measuring in both sidebands simultaneously when making a line survey. Furthermore, it simplifies the optics and eliminates a source of loss. Since an SSB filter most be tuned to each new observing frequency, its omission removes the need for a cryogenic actuator.
To identify which sideband a spectral feature is appearing in will require multiple observations of the same line with different LO frequencies and software deconvolution. However, experience with ground-based frequency surveys leads us to believe that this will work well.
The LO signals have to be combined with the astronomical signal and fed to the mixers. In the HET we think that this should be accomplished with a passive beam-splitter. The astronomical signal will incur some loss in the beam splitter and also requires higher LO power. However, the alternative of using a tuned frequency diplexer has been rejected to avoid having a mechanical actuator. Its removal also allows the use of a broadband IF at a low IF centre frequency.
We believe the HET should have an IF bandwidth of more than 4 GHz, and a goal of 8 GHz depending on spectrometer performance. A suitable centre frequency would be in the region of 10 GHz.
We propose to use spectrometers built from a number of AOS modules to provide the very wide bandwidths required. Each individual module would provide a bandwidth of 1-2 GHz and resolutions of 1-2 MHz. A total of 8 such modules is feasible at the present and we suggest a total spectrometer bandwidth of 16 GHz in two units as a goal.
For higher frequency resolutions we include two DACS each with a bandwidth of 1 GHz.
It is a pleasure to acknowledge the contributions of the members of the HIFI Group who are:
I also thank the FIRST project team at ESA for their assistance, in particular J. Cornelisse, C. Jewell, T. Passvogel, G. Pilbratt, H. Schaap, and J. Steinz.
This page was created using TextToHTML. TextToHTML is a free software for Macintosh and is (c) 1995,1996 by Kris Coppieters