The internal calibration device was designed to provide an internal flat-field source, and a reproducible, but not absolute, calibration reference. Calibration of the SW channel required source temperatures up to 350 K which are difficult to fit in the low thermal dissipation allocation of ISO. The solution is a small resistor mounted on a thin kapton film. The low thermal conductivity of the kapton prevents heat losses towards the ISOCAM base plate and ensures that almost all the energy is radiated. This emitter feeds through a small input hole an integrating sphere which is mounted on the selection wheel. In the calibration mode, the output hole of the integrating sphere takes the place of the Fabry mirror. The brightness uniformity is better than 1% in the unvignetted circular 3 arcmin. field of view of ISO. Two spheres are included in the system, one for each channel, with entrance holes adapted to the flux required for each array. The temperature of the emitter can vary in the range 150 K to 350 K.
This design has two limitations. On the optical side, the beam aperture of the ICD is defined only by the aperture stop at the filter location. It is different from the ISO telescope f/15 beam. Nevertheless, a good flat-field can be obtained with the 1.5 arcsec. and the 3 arcsec lenses. The second limitation is due to the presence of some low level emission bands of the kapton support which is heated by conduction. They can be seen at long wavelength in the spectrum of our calibration device. This would preclude the use of this system as an absolute flux calibration reference. However, it is of little importance for ISOCAM, since we will use it only as a relative reference standard. During the integration and calibration phases of the flight model, on a time scale of one and a half years, the reproducibility of the calibration system has been better than 10%, in all the optical configurations of ISOCAM.
The ICD can be used in flight for flat-fielding the array. The 1.5'' will be used, irrespective of the lens used during the observations. The standard signal level of 200 ADU is adequate to get a good flat-field accuracy, while keeping the detectors well inside the good part of their characteristics; the integration time will be 2s. The whole procedure will take 5 minutes, and provide a flat-field accuracy of 1% for the LW channel and 3% for the SW channel.
A large current pulse is sent to the IR emitters at the beginning of a calibration sequence to speed up the heating of the emitter before setting the current to its stable required value. To avoid saturation of the detectors during this heating period, the filter wheel is set on a blind position. Therefore, it is not possible to make a calibration of the arrays without going for a short time to the dark position.