Go Down

Topic: Mega 2560 Operating Temperature/Overtemp Protection.  (Read 1 time) previous topic - next topic

Rob-C

Hello all,

In the ATmega640 datasheet, it reads that the operating temperature is -40C - 85C industrial. After reading through the forums, i learned that the "overall operating temp range" of the whole mega2560 is not known, or at least that information is not otherwise available.

My question is this- does the chip have any built in overtemp protection, so that it would shut itself off automatically if it ever reached 85C, to prevent damage? Or will it just burn itself into the ground?

If it doesn't have built in overtemp protection, i'll add a little NTC thermistor to my PCB to measure the ambient temperature inside the enclosure (it's gonna be a completely sealed ip68 steel enclosure that we're custom fabricating for this project). I'll also probably add a simple conditional statement to turn on some cooling solution like a fan or peltier (tbd) when the temp goes above 40-50C.

Thanks

EDIT: Follow up question-
Can you put a little heatsink on the ATmega chip? Is one sold somewhere that would be particularly compatible? How would you go about mounting it?

hammy

If you have a sealed box a heat sink won't do anything as the heat has no where to go.
IMO you'd be better off minimising power dissipation by supplying at 5v onto the 5v pin , minimising the load on the parts ( outputs) and putting the system to sleep periodically .

Rob-C

If you have a sealed box a heat sink won't do anything as the heat has no where to go.
The box may be sealed, but the plan currently is to have a heat sink/spreader inside and outside of the enclosure, passively transferring the energy with heat pipe. Something like this -> https://imgur.com/m4ir5Fl

Quote
IMO you'd be better off minimising power dissipation by supplying at 5v onto the 5v pin , minimising the load on the parts ( outputs) and putting the system to sleep periodically .
Since writing the original post, I've done some testing and determined that the heat generated by the device itself will not be an issue.
I placed the device inside a small, *insulated* enclosure. For reference, the project is a relatively simple temperature display, including an LCD display and about a dozen 10k NTC thermistors. In the final form of this project, the script will delay for about 5 second between each measurement update cycle. However, for this test, I removed the delay entirely and let it go balls to the wall. After allowing the device to run continuously in this test setup for over 24hrs, the internal temperature of the enclosure never went more than ~5F over the external ambient temperature.

Seeing as how this device will be installed in the engine cab of a turn-of-the-century coal-fire steam locomotive, however, I am (slightly) worried about the radiant heat from the firebox heating up the enclosure.

My thought is this- I'll just install the device/enclosure in the cab with the passive heat sink/radiator/spreader i mentioned earlier, and during its first trial runs I'll keep an eye on the internal enclosure temperature. If it begins to approach, say, 50C/120F, I can add a low wattage peltier to that heat sink, relay controlled by a D/O pin.

If you have any better ideas than that, feel free to let me know!

hammy

Sounds good

If you google " Arduino sleep" there are methods to drastically reduce the current consumption of the processor - in your example it could sleep
for 5seconds and wake to take the measurement for say 100mS . That would reduce your power consumption dramatically

Go Up