| Electronics > Projects, Designs, and Technical Stuff |
| Soldering a heater to a TO-style metal can? |
| (1/2) > >> |
| mark03:
During this lockdown I got the idea of using one of the unused inputs on my Davis weather station for a cloud detector. A thermopile detector of the sort used for infrared thermometers, when pointed at the night sky, does a fine job of sensing clouds: a clear sky has a temperature near absolute zero, while clouds have a much higher temperature. One of the practical issues in a cloud sensor is keeping dew off of the detector. (Clearing away rain and snow is another concern, not addressed here.) To this end, it's good to heat the sensor just enough to maintain it at the ambient air temperature. The thermopile device already has a thermistor inside the can, because you have to know the device temperature to figure out the temperature of the object it's looking at. A similar thermistor facing the ground could be used to switch on a low-power heater whenever the upward-facing sensor drops below the temperature of the downward-facing sensor, with perhaps 1-2 degrees of extra safety margin. The thermopile detector comes in a TO-style metal can. I'd like to find a resistor / heating element which I could thermally couple to the thermopile can and drive at ~ 1/4 watt. I have some really thin Nichrome wire, but it would be a pain to insulate, and the resistance still isn't high enough. Ordinary SMD resistors could, I suppose, have one end soldered to the can, but it would be even nicer if there were a "resistance thingie" with an isolated metal casing which could be soldered to the can. Anyone know if something like this exists? |
| bob91343:
How about a small lamp? |
| Gyro:
Whilst a TO style can is easily solderable, I wouldn't recommend it on a thermopile hybrid detector with end window and potentially low melting point alloys used in its construction. Your best bet would be to either get a TO5 heatsink and bolt a resistor to that... or a small piece of sheet of copper or brass, shaped to clip around the package and solder a resistor to that. P.S. You used to be able to get little PTC ovens to go over TO5 packages, but that would cover up the whole thing. |
| mark03:
Hmmm, yes, a heatsink is probably safer. I don't have the thermopile yet---I was wondering why they were all out of stock, then I realized, IR thermometers!! Scored a couple on Ebay then the seller backed out because he was "out of stock." Coincidence? :rant: Anyway, some of the older sensors have TO-5 cases but most are lower profile, like the ZTP-115 which digikey calls a "TO-205AA TO-5-4 metal can variant" (whew). If the diameter is the same, it wouldn't be too hard to modify a TO-5 heatsink. This is going to be battery powered, probably by a 18650 cell, so I would like to direct as much of the heat as possible to the can where it's needed, and as little as possible to the inside of the enclosure. For that reason it would be nice to have a soldered thermal contact, at least between the resistance element and the heatsink. A lamp would be harder to target, I think. I am planning to mount the thermopile with long leads and some foam insulation between it and the PCB. |
| edpalmer42:
If you just add a heatsink and don't do anything else, the case will tend to warm/cool to ambient temperature more quickly than without the heatsink. Would that be enough to control any problems with dew? If you decide you need a little heat, I've seen heatsinks shaped like the letter 'S' that are used to keep two transistors at the same temperature. Build or buy one of those and use the other transistor as a heater. These days, most oven-controlled crystal oscillators don't use a resistive heater, they just use a transistor as the heater. Ed |
| Navigation |
| Message Index |
| Next page |