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

5.1 Instrumental and Atmospheric Effects

Observations of the isotropy of the microwave background radiation are amongst the most difficult ever attempted in astronomy. The sensitivities at the level of tens of microkelvin that have been achieved are remarkable, given the many instrumental and atmospheric effects at the level of millikelvin to a few Kelvin. However, progress in this field demands sensitivities an order of magnitude better, i.e. at the level of a few microkelvin. It is therefore important to enquire whether these techniques have already been pushed to the limit of sensitivity that can be achieved in practice, or whether one may reasonably anticipate a further improvement in instrumental sensitivity of an order of magnitude.

The outlook for instrumentation is promising - instrumental advances are being made which permit observations with sensitivities which were not dreamt of ten years ago. In the last few years we have entered a new era - the extremely sensitive wide band bolometer observations of Meyer et al (1991a, b) provide a convincing demonstration of the future potential of this approach.

Another development which has already helped in these efforts to improve the sensitivity of the observations is the introduction of a new generation of transistor amplifiers - sometimes called HEMTs, TEGFETs or HFETs [see e.g. Das (1987), Mishra et al (1988), Chao et al (1989, 1990) and Tan et al (1991)] - both as replacements for SIS receivers, albeit at lower frequencies, and as replacements of hydrogen maser amplifiers. The full capabilities of transistor amplifiers are as yet far from being realized. The replacement of the maser amplifiers with transistor amplifiers on both the 5.5 meter and the 40 meter telescopes at the Owens Valley Radio Observatory has increased the instantaneous sensitivity by over a factor two so that the same noise levels will now be reached in less than one quarter of the observing time. This will make it much easier to track down low level systematic problems, and the full increase of a factor two in sensitivity should be directly reflected in the sensitivities achieved in the best weather. The significant effect that the introduction of transistor amplifiers has had on the South Pole observations can be seen by comparing Figures 1d and 1e. The results of Gaier et al (1992) are still preliminary, and a thorough analysis is still being carried out on these data, but it is clear that the sensitivity has been improved dramatically through the introduction of transistor amplifiers.

The impressive sensitivity already achieved by Meyer et al (1991a), see Figure 4b, indicates clearly that instrumental limits are not the problem here, but that backgrounds must be dealt with. Davies and Lasenby and their collaborators are now using transistor amplifiers and their double-switching scheme will likely enable them to push the instrumental limits well below the level of fluctuations due to the Galactic synchrotron emission, shown in Figure 5. Observations with the VLA (Fomalont et al 1988) have pushed the instrument to the limit of its capabilities, and the observed variances are only ~ 15% higher than the theoretical variances - which is typical of normal VLA operation. This exceptional performance is largely due to the automatic cancellation of most offsets and drifts in an interferometer - a point that we return to in the discussion of instruments for the future.

There is no doubt that instrumental sensitivities can be pushed at least an order of magnitude beyond present limits. However, each observational technique has its own peculiar set of problems and it will require substantial efforts to take full advantage of the new developments in instrumentation. As far as ground-based observations are concerned there are two major sources of systematic error to be overcome: atmospheric effects and ground spillover, and it is important to determine whether these are likely to impose a hard limit on further progress. The atmospheric effects can be minimized by careful selection of observing frequency and observing site. Observations from the South Pole (Meinhold & Lubin 1991, Gaier et al 1992) at 25-35 GHz have been particularly successful in overcoming the atmospheric problem.

As far as future prospects are concerned, some encouraging results have been those from the Owens Valley Radio Observatory. In the observations shown in Figure 1b and 1c no zero levels or drifts have been subtracted - these are the raw data from which only noisy data (based on internal scatter) have been excluded. We have seen (Equations 8a and 8b) that the rms sensitivity level is 10 µK. Furthermore, in the case of the RING experiment, we know that the sky average around the RING must be very close indeed to zero, regardless of the level of the true sky fluctuations, so that this result is absolutely calibrated internally. These results show, therefore, that the double switching technique on fields within a few degrees of the celestial pole can eliminate all systematic errors introduced by the instrument and the atmosphere down to the level of 10 µK. This technique therefore successfully eliminates atmospheric and ground spillover effects down to at least half an order of magnitude below present sensitivity levels. It is possible that we are on the threshold, at the level of 10 µK, of insurmountable problems from the atmosphere or ground spillover, but no hint of this has yet been seen, and it therefore appears most likely that the full factor ten improvement in sensitivity being sought will be possible with these techniques. Thus it has been demonstrated that the large effects due to ground spillover and the atmosphere can be removed to high precision in the second derivative. We are optimistic that the same can be done in the first derivative by means of interferometry and, perhaps, also by means of focal plane array observations.