Annu. Rev. Astron. Astrophys. 1992. 30: 653-703 Copyright © 1993 by Annual Reviews. All rights reserved

5.2 New Instruments

In the light of the considerations discussed in the previous section, it appears that instrumental limitations are not likely to be the chief obstacle to progress in this field, and that far more serious problems are posed by the various forms of astronomical foreground signal discussed in Section 4. The need to eliminate foreground emission has therefore become a major driver in instrumental design. Spectral information is the key, and this introduces considerable complexity in both the instrumentation and analysis.

The two most important choices that must be made in the design of instruments for microwave background radiation observations are observing frequency and angular scale. Given the gestation period of space missions, we will limit our speculations about the future to ground- and balloon-based measurements. Balloon flights covering intermediate angular scales will undoubtedly continue with extremely sensitive bolometers. The main limitation here, for which there is no easy solution, is the limited amount of integration time that can be achieved in any one flight.

Atmospheric oxygen, even at great altitude, is opaque to frequencies near 60 GHz, so that instruments can be divided into two frequency ranges, below about 40 GHz, and above about 80 GHz. At frequencies below 40 GHz receivers will be based on transistor amplifiers. At frequencies above 80 GHz bolometers, SIS mixers, and transistor amplifiers will, very likely, all be used. In terms of the confusing backgrounds, the lower frequency range is dominated by discrete extragalactic radio sources, the Galactic synchrotron background, and free free emission, while the upper frequency range is dominated by Galactic dust emission. We have already discussed the difficulties associated with removing these confusing signals from observations. New data from COBE, in combination with other data in both frequency ranges, are likely to have a major impact on our understanding of these backgrounds, and continued studies and simulations of their removal will undoubtedly lead to new insights. It is certain that wide frequency coverage will be required for separation of the various backgrounds, but whether a single contiguous frequency range or many smaller ranges widely spaced is better, remains to be seen.

In the past, angular scales < 2' have been observed exclusively with instruments built primarily for other purposes, such as the VLA and the IRAM telescopes. These instruments are very well suited to these observations, and such observations will continue. In addition these instruments could be significantly upgraded. For example, the VLA is now limited at frequencies 8 GHz by discrete sources, but at 15 GHz and at 24 GHz it is limited by receiver noise. The noise temperature of the VLA receivers at 24 GHz is ~ 200 K. A factor of four decrease could now be achieved. The planned NRAO millimeter array would also be an important instrument for small-scale measurements. The upgraded Ryle telescope at Cambridge (Ryle 1972, Jones 1991), consisting of eight 13 m antennas, with baselines from 18 m to 4.6 km, is now dedicated to microwave background radiation measurements, primarily the Sunyaev Zel'dovich effect, with receivers at 5 GHz and 15 GHz. It cannot match the sensitivity of the VLA at the same resolution, but the availability of shorter baselines makes it better on larger angular scales.

On angular scales 10'-10° many different ground-based instruments are being planned, some with the goal of imaging the microwave background radiation, rather than sparsely sampling it or its derivatives. The motivation for making actual images is that they make it possible to study a wide range of angular scales simultaneously - since for imaging instruments the filter function F(k) (Equation 3) is approximately constant over a wide range of angular scales. Thus imaging observations, unlike sampling observations, can be optimized and analyzed without much regard to specific models. Although single or double beamed instruments used for sampling observations could be scanned to cover large sky areas, at the resolutions and sensitivities required the total integration times would be very long. The principal drawback of instruments designed specifically for making images, as opposed to those designed for sparse sampling of the autocorrelation function, is that it is much more difficult to control systematic errors, and thus to achieve high sensitivity, over a wide range of angular scales than it is over a narrow range. For a known fluctuation spectrum, W(k), imaging observations are uneconomical, because greater effective sensitivity could be achieved by concentrating integration time in well-chosen spots. Unfortunately, not only is the spectrum of microwave background radiation fluctuations unknown, but each of the confusing foregrounds has a different spectrum of fluctuations. Thus there is no substitute for imaging observations.

There are two fundamentally different methods for covering the sky at a faster rate than is afforded by single or double beamed instruments: 1. focal-plane arrays of bolometers or feeds; and 2. interferometers. The great advantage of interferometers is that, while each element of a focal-plane array looks at a single patch on the sky, each element of an interferometer looks at the whole area which is being imaged. In focal plane arrays, therefore, one has to carefully calibrate out any instrumental effects due to the individual array elements. On the other hand, the great advantage of focal plane arrays is that the signals do not have to be correlated prior to detection, and much of the complexity of correlators increases in proportion to the square of the number of array elements, so that a very large correlator is needed for an interferometric array of, say, fifty elements operating in seven frequency channels. However the number of elements in a the focal plane of a radio telescope with, say, 0°.5 resolution is severely limited, and at 30 GHz the practical limit is probably of order 10, so that interferometers appear to be the only practical choice for imaging areas of ~ 100 beam areas. Interferometry is not possible with bolometers, of course, as all phase information is lost.

There are many other important technical questions concerning the merits of the two types of instrument; however, given the enormous sensitivity of low- temperature bolometers at high frequencies, and the rapidly improving sensitivity of transistor amplifiers at low frequencies, it is likely that instruments of both types will be built to image sky regions up to about 10° across with resolution 1° and sensitivity approaching T/T = 10^-6 over the next decade.

ACKNOWLEDGMENTS

We thank S. Myers, T. Pearson, and A. Sandage for many useful comments on the manuscript, and S. Myers for providing Figure 2. This work was supported by grant #AST88-15131 from the National Science Foundation.