A number of calibration schemes are under consideration for the ALMA. Good accuracy is expected for frequencies below 300-400 GHz, whereas it will be more difficult at the higher frequencies where atmospheric extinction is high and system temperatures are correspondingly higher. At the lower frequencies where the goal of calibration accuracy of 1-2% may be achieved, small effects are important, including standing waves between lens and mirror surfaces, mixer non-linearities, determination of relative sideband gains, and polarization stability. Among the various schemes being proposed, the possible use of cold loads in the dewar will have a major effect on the overall receiver design. After some discussion, the ASAC reaffirms its recommendation made at the Leiden meeting that there should be no cold loads in the dewar. One reason is that there may be some receiver non-linearity as a result of the significant load intensities. In addition, loads in the dewar may have their accuracy spoiled by lens or window reflections. A general recommendation is that the calibration system be external to the dewar and therefore not play a major role in the design of the receiver dewar.
The problems of standing waves between mirrors are familiar but have largely been ignored in millimeter interferometry because their effects are usually small compared to the typical current calibration accuracies of 10-20%. Because these effects are not stable in time, they must be dealt with as part of the frequent receiver calibration. As an example, the common ``chopper wheel'' method is simple and attractive and might be adopted, but it will have to be supplemented with additional measurements if it is to be used to measure the effects of the standing waves. Two recently proposed ideas that should be developed further include (a) the use of a semi- transparent vane that could be calibrated directly on an astronomical source and (b) a coherent signal transmitted to each antenna of the array over optical fiber that could measure sideband ratios and bandpass shapes if it is injected at the input of the system. Another scheme, that is currently under development for the BIMA array, uses two loads weakly coupled at the secondary mirror. It has shown good reproducibility and the capability of measuring the standing wave pattern, and the mechanism also provides a natural means for the introduction of the coherent signal. Further tests are planned to determine whether it can produce absolute calibration at the 1% level over a wide bandwidth.
For the coherent signal scheme, it will be important to demonstrate that the accuracy obtained will exceed that of more conventional methods. This system is particularly useful in 3 areas:
The first 2 items require proper control of the power vs. frequency, which may be difficult. Baseline ripples apart, bandpass calibration can be achieved on astronomical sources within 1-2% accuracy in a reasonable time at millimeter wavelengths. It will be more difficult to do astronomically at sub-mm wavelengths where the potentially high signal/noise of the coherent scheme gives it an advantage. Measuring the sideband gain ratio is easy at mm wavelengths, but time consuming in the sub-mm region unless the specification is relaxed to about 5% accuracy. Here again, if the coherent scheme can be made accurate to the 1-2% level, its inherent high signal/noise will be an advantage.
Beyond the radiometer calibrations discussed at the meeting, absolute calibration of standard astronomical sources is an important issue. The ultimate limitation of absolute calibration accuracy is the insufficient knowledge of astronomical source strengths, particularly planets. A worthy goal of the array will be to define the relative strength of all such calibration sources, in addition to providing radiometer calibration within 1-2%. If such measurements could be obtained for several planets, atmospheric and surface modeling should be able to provide a consistent absolute scale for each to at least 5% and hopefully better than that. Our main conclusions are that there should be no cold loads in the dewar, and that the other current ideas, particularly the semi-transparent vane and the use of coherent signals, should be developed further. If the scheme using two loads weakly coupled at the secondary mirror is demonstrated to achieve the required absolute accuracy over a wide bandwidth, then it should also be a contender for the final choice.