Development of a Heat Switch for Use 20-50K 941530

If a cryocooler connected to an instrument fails there is a parasitic load on the instrument due to conduction down the cold finger. In normal operation the cryocooler provides enough cooling to absorb this parasitic load in addition to the cooling required for the instrument. If multiple coolers are used on an instrument it is helpful if the failed cooler can be thermally disconnected to reduce the thermal load on the remaining coolers. Also, it may be a requirement that a “spare” redundant cooler be thermally connected to the instrument to provide cooling in place of a failed cooler. For both of these operations some type of thermal switch is required. The design and development of such a switch is described in this paper.

When two metal surfaces are pressed together the contact between them is on the tips of irregularities on their surfaces; these asperities are plastically deformed as the load increases so that the area of contact is proportional to the load. Surface finish of the contacts is important as true metal to metal contact has to be established to provide good thermal contact.

Conventional designs rely on clamping two or more surfaces together. The approach taken in our design was to employ rotating cams that “wipe” the surface to ensure true metal to metal contact. The contacting surfaces were gold-plated copper. A stepper motor with a 0.036N m torque was used to rotate the cams and Belleville washers were used to provide the clamping force. A reduction ratio of 100 was used with the stepper motor to provide the torque required to operate the switch.

The requirement was that the switch should exceed, when closed, a conductance of .05 W/K at 20K and .2 W/K at 50K. The best results were achieved when the joint was made at, or near, room temperature. In conventional heat switches making the joint at low temperatures can degrade the performance from that of a joint made at room temperature by up to a factor of 100. In the device described in this paper this factor was only 1.3 at 50 K. This clearly demonstrates the effectiveness of using a wiping action over the conventional approach. The conductance of the switch did not follow the expected dependence on force. The conductance between plain clamped contacts has been found to vary as the F^0.75 where F is the clamping force. In tests on this switch the conductance was proportional to F^0.49 at room temperature rising to F^0.77 at 100 K. This was not due to any change in the spring constant of the compression springs.

The “off” conductance of the switch was less than 3×10^-4 W/K. This would give a heat leak below 60mW. This is less than 20% of the cooling power of the two-stage Stirling cycle cooler that has been developed at the Rutherford Appleton Laboratory for ESA (300 mW at 30K). This very low off conductance was achieved by using approximately 200 thin steel washers as part of the support structure while in the “off” state. The multiplicity of thermal interfaces provides a low thermal conductance link.

The total mass of the switch was 618g which includes the stepper motor (195g). The power required during a switching operation at room temperature was 1.9W during one switching operation of duration 100 s. The switch passed qualification level vibration testing.