I am currently working on the design of a focal plane assembly to be implemented on a small satellite (4 EMCCD detectors) to be operated at -80°C (which is not really cryogenic and likely still considered boiling hot for most of you :-). As it is passively cooled, the assembly will experience about -120°C during times when the heat loads of implemented or neighboring subunits are not present. Due to a high readout frequency of the detectors, an immediate pre-amplification of the signal is required directly at the EMCCD output. The major design problem I am currently facing is that the related proximity electronic module will experience temperatures on the same order.
I have developed a preliminary athermal FPA design concept which uses a machinable ceramic structure that provides an excellent CTE match to our AlN detector package. However, my major concern is the reliability of the proximity electronic modules (as they likely also going to experience a large number of cycles), how to manufacture them and also how to mount them without introducing critical thermally-induced stresses. The substrates are quite large (we think about L-shaped boards of 60x60mm, 15-20mm wide) and likely not well matched to the structure material below (we currently favor Al2O3 thickfilm hybrids, CTE difference about factor 2 to the basic structure, can provide details on request). My concern is now the sensitivity of ceramics to tensile stresses. Apart from the reliability of interconnections and packaging of the mounted electronic parts on the substrate, I fear damage of the ceramic hybrids.
I would be glad if I could exchange some ideas or information on
-Stress-free / quasi-elastic, but vibration insensitive mounting of the electronic substrates on the cold base plate structure which enables a compensation of the CTE mismatch between both parts, but allows bonding and achievement of required stiffness for vibrational load testing – how do you mount or integrate circuit boards or substrates at extreme temperatures? (putting the circuit boards into a warm environment is unfortunately not an option for us)
– Reliable proximity module design which survives these extreme temperatures and quite some amount of thermal cycling, which is in contrast to ground-based instrument design (less cycles, and if something brakes it can be repaired, not the case on a satellite…)
Any recommendations, hints or literature references on this issues would be greatly appreciated.
F.e. glueing the substrates on the basic structure appears to be no option as the shear stress due to the CTE mismatch in the epoxy layer will likely exceed a critical limit, so the substrate is put at risk of delamination from the support structure. On the other hand, bolts induce critical stresses…
3 Comments
Roy Blake
August 23, 2012We are not the experts in this situation but I can tell you that stycast epoxy is a very good adhesive. Stycast was trialled and used to hold down Al pylons to support the wiring to the JPL detectors on SPIRE, clearly met all of the EQ requirements including launch forces and now
happily operating at L2.
I understand the concern with the CTE mismatch; is the substrate mass or the temperature range significant?
Worth evaluating Emerson & Cuming stycast 2850 with catalyst 9 or 11. Whilst you will not see it in the spec sheet we have used it extensively to 100mK, PLANCK HFI 100mK plate also at L2 performing the ‘al sky survey’.
Anyhow, I hope this generates some lateral thinking.
Michael Rappaport
August 23, 2012I was using a 40 X 40 mm^2 2-stage thermoelectric cooler (TEC) in a UHV system. The manufacturer recommended (warned?) that grease be used as a thermal contact agent to allow for differential contraction. Grease being anathema in UHV, I used epoxy instead. At about -30 C, the TEC became open circuit. I presumed that the TEC was defective, but the same
thing happened a second time at about the same temperature. The manufacturer was right: either use grease or several small TCEs. In general, it isn’t the relative differential contraction that breaks things, but rather the absolute contraction, so a small TCE will tolerate thermal stresses that a large TCE won’t.
There are very low vapor pressure greases. The classic cryogenic grease is Apiezon N. Dupont Krytox has even lower vapor pressure, but I’m not aware of data on its thermal conductivity at low temperatures. All greases will freeze, but before they become rock hard, they allow
differential thermal contraction. Whether they allow enough for your application, you’ll have to check.
When I clamp two large pieces together that must remain aligned, I use the following system. Imagine four clamping screws in a rectangular array. One of the screws is tightened hard. Under the heads of the two closest screws, I use Belleville spring washers. The nuts are all-metal
and self-locking; they’re also known as “prevailing torque nuts.” The nuts are tightened until there is adequate clamping pressure; one piece can slide under the Belleville washers . Because the nut is self-locking, it can be left at any position without coming loose. The openings for these two screws in one of the two pieces are slots parallel to the sides of the rectangular array.
During contraction, for example, one piece can slide relative to the other, but it returns
to its initial position upon warming. The fourth hole can be round, but it must be big enough to allow the contraction; it does not need to be a slot. The slots are necessary if precision alignment is required; if not, all openings can be round.
Good luck!
Karsten Schindler
August 23, 2012Michael,
Thank you very much for your answer to my inquiry and your time! I totally agree with you that the absolute contraction, which increases the larger the contracting part becomes, is the major
problem. Smaller units can compensate stresses large likely can not.
Given this fact, I consider our circuit boards as too large. Unfortunately we can not split up our board in two or three smaller ones. The board encloses a CCD which needs to be connected by ~ 40 bond wires. Dividing the PCB would only require more interconnections, which makes the design too complex and introduces new points which can lead to failure. I was in fact thinking about slot-formed holes which enable the PCB to slide. I have seen this principle already, however not in spacecraft engineering yet. However our electrical engineers vetoed that the
board is not sufficiently fixed for bonding if we clamp it with Belleville spring washer. We want to use ultrasonic wedge bonding, and as soon as the board is able to vibrate just a little, the ultrasonic energy from the bonding process is dissipated which can result in a erroneous bond.
Another problem we face is qualification of the assembly. The focal plane must work over 2 years without any possibility of maintenance (which is f.e. possible in a groundbased UHV, if something breaks)… not to mention the number of thermal cycles we are going to face.
Still, your email gives me confidence that I should follow this idea. I was looking for some “magic idea” used by professionals, but I guess sometimes the simplest ideas are the best. I hope we can test it on our EQM and see how it works.