ESP32 pico D4 SoC heat problem

[ksnboopa]

I have designed my own PICO board using PICO D4 SoC with help of the document : https://www.espressif.com/sites/default ... eet_en.pdf. Everything works perfectly fine except heat is produced when chip is using WiFi functions. Device has to be always connected to mesh, otherwise I would have used sleep functions also.

I have tested with mesh code for a week and no issues found (without plastic case covered) except heating. My plan is to put the same board behind wall switch and will have a plastic enclosure also(no air flow will be there). Only SoC chip gets heated no other components are impacted.

According to datasheet temperature recommended is from -40 C to 85 C. I am unsure how can I proceed with this.

Is it safe to put behind wall switch?

Framework I use : ESP-MDF v1.0

Circuit Diagram : https://i.stack.imgur.com/AhNaI.png

Could someone help?

[ESP_Sprite]

That is very hard to say, as we have no idea what the thermal properties of 'behind a wall switch' would be. Also, while you can technically run the SoC up to 85C, depending on the implementation you may not want it to get to that temperature if it's close to something that people can touch.

[ksnboopa]

Correct. I will measure exact temperature reaching and then let know. My concern is, in the open place itself is getting hot. Just thinking with board enclosure and will be placed behind another wall mounted switch it would be higher than now as there won’t be much air flow too...

But mesh will be connected always 24*7

People will never touch the module as it will be enclosed and placed inside safely. Heat is the only sensitive factor I’m worried about...

[username]

What you might want to try is to ensure that there is no issue with your "PICO board" is to try your code on a manufactured board that uses the PICO D4 SoC, and see if that overheats.

[PeterR]

Buy a development board & measure!
On menuconfig set CPU speed to max & then max your inputs & burn that mother!
If you have a mechanical engineer on hand then you can get a very good idea from his FEA.
I find it quicker & easier to put a resistor in a package & match the max device/circuits thermal output.
Ok, you need to think that through some as a resistor tends to be more of a point source than an ESP/CPU and especially as an ESP is typically a package with a case etc but you can learn a lot & if you are OOM safe, well happy days!
EDIT: PS when thinking temperature then cool is usually better (obviously!). Reliability & especially FLASH writes tend to drop off as u get hotter. So I would strongly recommend that you scope the problem. There is plenty of stuff you can do to reduce any problem, usually involving big chunks of metal. Clearly you need to consider/test the extremes of your operational environment & so your capacity to disipate heat. Back in the day lacking a thermal oven I used a cardboard box & 'calibrated' fan heater, its not proof but with the right attitude smooths the qualification edges