John Larkin wrote: : On Fri, 29 May 2009 13:01:35 -0700 (PDT), snipped-for-privacy@gmail.com wrote: : >> > I don't think you'll have any problems with the board. Wide range temp : >> > cycling can stress surfmount solder joints, so thru-hole parts are : >> > probably safer.
My statistics is not sufficient to really draw conlusions, but my gut feeling is that I've seen thru-hole parts break approximately as often (i.e. not often) as SMD parts. I vaguely feel that the breaking likelihood may be more closely tied with the physical size of the component. Large ones break more easily - or perhaps they are just easier to observe :)
: >The circuit board will be go through lots of thermal cycling. It will : >be used by students to measure the Johnson noise from resistors as a
OK, you cannot be completely careless then when choosing a readout amplifier with sufficiently low noise temperature, and cannot afford screwing up much with noise matching.
: >function of temperature. So I expect maximum thermal stress from the : >students.
But no fatal consequences if something breaks. I think you're safe with a quick-and-dirty PCB without taking much precautions.
: >I thought I might try some small surface mount diodes as temperature : >sensors, but I can use the glass encapsulated variety if these won?t : >work.
: Regular diodes should work fine.
I agree that they should work, but I think I encountered some funny unexpected behaviour when experimenting with both Si and Ge diodes several years ago. I forget exactly what it was (may have been below
77K anyway) and feel too lazy to find the lab notebook now. I didn't spend much time to figure it out, because I had the official Lake Shore sensors available anyway. I think there is a paper about using off-the-shelf diodes as thermometers in some back issue of the Cryogenics magazine. I'd be interested to hear about your experiences if you end up using a diode as the thermometer.
Many off-the-shelf components work at 4K and even more work at 77K. Resistor values change, so you need to measure and compensate. NP0 caps hardly change at all. X7R caps fall to 1/10 and Z5U caps freeze out completely. Inductor cores are more varied, mu collapses in most of the high-mu cores but occasionally you find a surprise core where it does not. GaAs devices typically keep functioning, but Si devices tend to freeze out below 77K, except for those enhancement-type MOSFETs whose ohmic contacts happen to be doped strongly enough (eg. 2N7000 from some manufacturers). Some Si JFETs can work at 4K when made to dissipate enough so that they internally heat up to 50-80K. A worn-out notebook where one lists which parts work and which won't work in LHe or LN belongs to every cryogenic engineers assets. I think such a notebook is what allows 'cryogenic suppliers' to ask for hefty premiums when selling parts. Unfortunately the contents of the notebook may vary when the manufacturer changes some tiny bit in their fabrication process which does not affect the room-temperature specs.
: >Since I have your attention I also need a robust heater that will : >survive repeated trips to 77K. I?m going to try some of the 5 or 10 : >Watt aluminum housed power resistors made by Dale, Ohmite. Do you : >have any advice? I?d rather not have to wrap my own out of resistance : >wire, much cheaper to buy something commercial.
Our heaters are typically deposited on-chip, so I cannot comment on durability of ordinary resistors.
Regards, Mikko