# How to remove heat from an IP66 enclosure?

I have a small IP66 polycarbonate enclosure with a Zirconia-based oxygen sensor which produces significant amounts of heat (the internal sense element runs at 450C using a 1.5W heating element, sensor surface reaches 85C). The enclosure also contains sensitive analog electronics to read the sensor. The residual heat from the sensor is driving the ambient temperature in the enclosure up past 40C (measured on the outside of the enclosure front panel). The oxygen sensor and another gas sensor are only rated for ambient temperatures of 50C maximum, so I'm worried about keeping those in spec. I'm also concerned about needing temperature compensation for the analog circuits.

How can I remove the excess ambient heat from the enclosure while still maintaining an IP66 rating?

• If the oxygen sensor is connected to a "remote/external" air source via a tubing, does the sensor & associated electronics have to be in a air/water tight container? If not drill holes into the container; some in the sides & some in the top
– Fred
Feb 25 '15 at 1:56
• Heat pipes do a superb job of moving thermal energy from a concentrated source to another location. If you can provide convective cooling somewhere on the package (aka a heat sink) then a heat pipe will allow you to transfer heat energy to it. | A heat is a tube etc containing some fluid at a reduced pressure such that it boils at a controlled temperature. The vapour carries substantial energy per mass and transfers the heat at the condensation point. A wick or gravity may be used for fluid return BUT you can buy commercial units in many sizes and do not have to do actual HP design yourself. Apr 29 '15 at 11:54

Considering the three methods of moving heat:

• Convection - To keep the IP66 rating of the enclosure, you can't add any holes for exhaust fans.
• Radiation - At the temperatures that you are talking about, radiation will not be removing much heat.
• Conduction - This is a viable alternative since it could work through the walls of the enclosure.

You can add fins on the outside of the enclosure and rely on heat being conducted through the walls of the enclosure to the external heatsink. This will depend on the material that the enclosure is made from. If it is metal already, this could work as-is. If it is plastic, you may be able to retrofit a metal plate into a wall and still seal it to IP66.

Once the heat is outside of the enclosure, a fan could be added to help the heatsink function.

Another option may be to have an internal heatsink and then move the heat through the walls of the enclosure via either heat pipes or a circulation closed-loop cooling system. The wall penetrations could be sealed to still meet the IP66 requirements.

Just out of curiosity, if the enclosure is sealed what will an oxigen sensor measure.

I would conduct the heat outwards via a metal structure. But cosidering the heat your sensor generates, I would reconsider the polycarbonate choice. Polycarbonate max operating temperature is about $115\,^\circ C$. A metal enclosure would conduct the heat much faster to the environment.

• It's sampling gas from an external environment chamber through an air line. A small pump pulls air from the environment chamber across the sensor. Unfortunately, for numerous operating environment reasons we can't place the sensor in the chamber directly. Feb 23 '15 at 15:19

Since your enclosure is IP66 you can pour a steady stream of water over it to absorb and carry the heat away.

Could you mount Peltier cooling units to the outside to pull heat out of the enclosure? I would think you'd want to have cooling fins inside the unit to increase the rate of heat absorption.

You could pre-cool the gas entering the enclosure, the heat would then go into re-warming up this gas.

This could be done by passing the gas through a simple finned radiator-pipe in a temperature controlled environment (basically a pipe through a fridge/freezer). This might be a good way to do it if you want to control the temperature that the gas re-enters the chamber as you could control the temperature of the pre-cooler and monitor the temperature that goes back in.

40°C is really no big deal for most electronic components, especially when those components themselves don't dissipate much heat. Electronics to amplify and buffer the output of a sensor, or convert it to digital and pass the data on, shouldn't require much power by themselves, certainly a lot less than the 1.5 W the sensor is already dissipating.

You have to check the datasheets, of course, but most ordinary electronic components work over the 0-70° range at least. For a little more money, you can get most semiconductor ICs in the "commercial" temperature range, which will usually extend well above 100°C, like 125°C.

So the short answer is that you haven't demonstrated there is a problem here to solve. Just make sure to check all the datasheet, and calculate the worst case offset and gain errors using the full temperature range you expect. But then again, that's standard practise anyway, and the datasheets will already spec things like offset voltage over the operating range of the part.

I'd stay away from electrolytic capacitors, since their lifetime goes down significantly with temperature. For a sensor amplifier and maybe digital converter, you shouldn't need them anyway. Again, I don't see a problem here.

• To be clear, the 40C number I measured is the outside of the plastic enclosure, so the ambient inside is probably worse. Unfortunately, the oxygen sensor and another gas sensor in the enclosure are rated for maximum ambient temperatures of 50C so I'm worried about keeping those in spec. Feb 24 '15 at 15:38
• @Joe: You might have mentioned this important information up front. That totally changes your question. Feb 24 '15 at 15:40
• Sorry about that, it didn't occur to me to check them because the existing off the shelf design we buy (that we're trying to replace) runs that hot with the same sensors and it hasn't been a failure point in practice. That said, the more I examine their design, the more I realize that it's made almost entirely out of red flags... Feb 24 '15 at 18:22