3 The dilution cryostat

In order to have a system noise as close as possible to the photon noise, we decided to cool the detectors to 0.1 K (see Subsect. 4.3). The development of a 0.1 K cooling system fully compatible with balloon-borne and satellite environments has been pursued at the Centre de Recherches sur les Très Basses Températures (CRTBT) in Grenoble (Benoit et al. 1994a; Benoit & Pujol 1994). The compactness and ease of use of this system render it very attractive even for ground-based photometers. Conversely, the Diabolo photometer provides a good testbed for this refrigerator before it is used on space missions. Figure 6 shows the layout of the dilution cryostat. This new refrigerator (Benoit et al. 1994b; Sirbi et al. 1996) is the first prototype of a concept that has become the baseline for the ESA PLANCK mission (formerly COBRAS/SAMBA). Its principle is based on the cooling power provided at low temperature by the dilution of ³He into ⁴He. The system does not use gravity. Instead, the fluids are forced into room temperature capillaries which, after going through a liquid nitrogen trap, are thermalised by the various shields in the cryostat down to the plate at (pumped lHe) 1.8 K. The two Helium isotopes come from high pressure storage vessels (see Fig. 6) through flow controllers. Typical flow rates are 3 $\mu$moles of ³He per second and 16 $\mu$moles of ⁴He per second. The cooling at the low temperature plate is produced by mixing the two isotopes. The available power is small (only few hundred nanoWatts). Therefore the cold plate is mechanically supported by Kevlar cords and shielded electrical wires (for the bolometers) are thermalised on the heat exchanger (capillaries of 200 and 40  $\,\mu {\rm m}\,$diameter). The output mixture flows back through the heat exchanger in a third capillary which is thermally tied to the two input capillaries. The output gas is stored in a low pressure container for later recycling through purification (it will be thrown away in space in case of a satellite version). The dilution fridge was continuously running during the campaign (i.e. for three and a half weeks), keeping the bolometers at the useful temperature of about 0.1 K, except during the main cryostat helium refilling, which required heating-up the cold plate temperature to 4 K. The absolute temperature of the 0.1 K stage is measured with a Matsushita carbon resistance ( $1000~\Omega$ at 0.1 K) in a AC low power bridge.