Sorry it's taken me a few days to get back to you.
spyclub wrote:In your experience, how hot would something like the etherten with the PoE module, or and ethernet shield with a standard arduino get when left in a sealed box?
This really depends on a lot of things - what the box is made of, how large the box is, how thick the walls are, what the ambient temperature and airflow outside the box are, etc, etc. As Chris has pointed, thermal modelling is a full time professional job and the only way to be sure is to simulate and measure.
spyclub wrote:Assuming worst case, a linear regulator is only 66% efficient, so will I actually need more like 330mA into the system with 110mA (0.55W) lost to inefficiencies/heat? and if that is the case, how does that 0.55W loss translate to actual increase in temperature?
The efficiency of the linear regulator is because the current in and current out are the same, but the voltage drops. So if you need 220mA @ 5V (1.1W) then you're putting 220mA @ 7.5V (1.65W) in. But your overall figure of 0.55W extra heat loss is correct, though.
(Because the heat wasted is proportional to the voltage drop, this gets worse as the voltage drop goes up. If the linear regular was capable of accepting 48V directly - which it is not so don't try this - then you'd be putting in 220mA @ 48V (10.5W) and the regulator would only be 9% efficient!)
The PoE regulator is probably around 90% efficient at taking in 48V and outputting 7.5V, you can make a worst case estimate that your total power dissipation is 1.65W / 90% = 1.83W. Call it 2W to be sure.
(Remember to add on extra power requirements if you plan to run anything else from the Arduino.)
Making a test rig like Chris suggests is a great way to estimate if it will work or not. However after you've done that I still suggest bench testing a final prototype, because that way you can account for things like differences in where the heat is produced, how it dissipates inside the box, etc.
Remember that if it's sitting in your roofspace the ambient temperature may get very high, so it's worth testing in a similarly hot ambient environment if you can.
The chip on the 802.11 regulator has a thermal shutdown feature so you should in theory see the entire unit switch off if it gets too warm. This is not something you want to rely on, though.
A couple of other suggestions that might help:
- Try mounting in an aluminium box if you can get one (if you're in Australia then Jaycar sell diecast aluminium boxes), to conduct heat more effectively to the outside.
- Build some temperature monitoring into your design. Depending on the exact model microcontroller you're using, some Arduino/compatibles have an internal temperature sensor
already. It's not super accurate but would work for an "am I overheating?" monitor you can keep an eye on in summer.