Page 1 of 2

ESP Thermal considerations

Posted: Wed Nov 21, 2018 2:31 pm
by eli_az
Hi,

So i'm using the ESP32 thing from sparkfun and I wonder what are the thermal limits on the ESP32 chip itself..
I saw on the Data-Sheet that the chip's operating temperature is 125C - is that ambient temperature or operating temperature?

Another question is if anyone know if the chip have a thermal shutdown routine to prevent over heating?
I tried measuring the internal temperature sensor with no success...

Thanks.

Re: ESP Thermal considerations

Posted: Thu Nov 22, 2018 3:29 am
by costaud
eli_az wrote:
Wed Nov 21, 2018 2:31 pm
Hi,

So i'm using the ESP32 thing from sparkfun and I wonder what are the thermal limits on the ESP32 chip itself..
I saw on the Data-Sheet that the chip's operating temperature is 125C - is that ambient temperature or operating temperature?

Another question is if anyone know if the chip have a thermal shutdown routine to prevent over heating?
I tried measuring the internal temperature sensor with no success...

Thanks.
125C is Junction Temperature.
The chip has no thermal shutdown routine to prevent over heating.

Re: ESP Thermal considerations

Posted: Thu Nov 22, 2018 6:24 am
by eli_az
costaud wrote:
Thu Nov 22, 2018 3:29 am
eli_az wrote:
Wed Nov 21, 2018 2:31 pm
Hi,

So i'm using the ESP32 thing from sparkfun and I wonder what are the thermal limits on the ESP32 chip itself..
I saw on the Data-Sheet that the chip's operating temperature is 125C - is that ambient temperature or operating temperature?

Another question is if anyone know if the chip have a thermal shutdown routine to prevent over heating?
I tried measuring the internal temperature sensor with no success...

Thanks.
125C is Junction Temperature.
The chip has no thermal shutdown routine to prevent over heating.
Hi,

Thank you for the help.

Yesterday I was transmitting over Wi-Fi and suddenly the chip turned off and refused to turn on.
I tried connecting / disconnecting from the PC repeatedly and thought that maybe the chip got too hot.
I decided to put the chip in the fridge to cool down for an hour and when I got it out it started working again.

Any thoughts about what could cause the problem? Do i have a defect board? (esp32 thing)

Is there any thermal resistance data available? (theta_junction to case)

Thanks.

Re: ESP Thermal considerations

Posted: Thu Nov 22, 2018 11:27 am
by costaud
What was the temperature on the chip when you found it not working ?
Well, it could also be something else on the board that caused the failure, I'm afraid I can't give an accurate answer based on the current information.
If this happens again, I think we can check the 3.3v power supply first, then, check the EN pin. If both of them are fine, check whether the strapping status and UART output are correct.

Re: ESP Thermal considerations

Posted: Thu Nov 22, 2018 1:13 pm
by eli_az
costaud wrote:
Thu Nov 22, 2018 11:27 am
What was the temperature on the chip when you found it not working ?
Well, it could also be something else on the board that caused the failure, I'm afraid I can't give an accurate answer based on the current information.
If this happens again, I think we can check the 3.3v power supply first, then, check the EN pin. If both of them are fine, check whether the strapping status and UART output are correct.
I was supplying 3.3v from a power supply and used a FTDI (usb-uart) cable instead of the onboard uart-usb converter.
the temp on the chip was about 100degC while it was transmitting in wifi.

Thats why I wanted to ask about the thermal resistance to be able to assume the die temperature.

Re: ESP Thermal considerations

Posted: Thu Feb 14, 2019 12:03 pm
by eli_az
I would appreciate any ideas or answers

Re: ESP Thermal considerations

Posted: Thu Feb 14, 2019 11:26 pm
by ESP_Angus
Hi eli_az,

We don't have any published thermal resistance values, sorry.

I've never heard of thermal problems with the ESP32 Thing development boards**, or any ESP32 development board for that matter (assuming reasonable ambient temps).

There are ESP32 chips that run in high temperature environments (like the inside of light bulbs) that require some thermal design considerations. But these also work (and are in shipping products).

If I had to guess, I would say if it is a thermal issue then maybe the ESP32 on your ESP32 Thing has some soldering problem - maybe the thermal pad under the chip somehow didn't get soldered fully so the chip is not sinking heat to the PCB correctly.

It's also possible to have a cold solder joint where thermal expansion of another part causes it to lose connection and then cooling it causes it to make the connection again.

A third possibility is that there's a short which is causing a lot of power consumption and heat near the chip. How much current is the ESP32 drawing from the power supply?

A fourth thing which will create excess heat from the chip is if the VCC voltage is over 3.6V (ie out of spec) but it sounds like you've checked this already.

If none of this helps, I have a few more questions:
  • When you say the chip refused to turn on, do you mean you got absolutely no output at all from the UART after a reset? Or that the chip would start running but not connect to Wi-Fi?
  • How did you measure the 100C on the chip?
  • How sure are you that cooling the chip is the factor which makes it "come good" again?
** One exception for ESP32 Thing: as it uses the much less common 26MHz crystal, if "autodetect" is set for the crystal config mode then ESP-IDF may fail to detect the crystal speed correctly depending on temperature. Something will still happen, but UART output may be garbled and the firmware won't connect to Wi-Fi. Setting the crystal speed to 26MHz in the config should work around the issue.

Re: ESP Thermal considerations

Posted: Mon Feb 18, 2019 9:17 am
by eli_az
Hello,

I checked for all the possibilities you have mentioned and it didn't help.
The esp32 consumes around 100-150mA when working but not transmitting and around 250-300mA when transmitting.
About the questions:
- By did not turn on I meant the 1st of option of unresponsiveness from UART(the com was recognized by the computer due to the FT231X component), - make flash did not work. The current dropped drastically, from 300mA in transmitting to 70mA. Idle was 7mA insted of 150.
- With a thermo-couple on the chip.
- Cooling seemed the only thing that helped.

Thanks.

Re: ESP Thermal considerations

Posted: Tue Feb 19, 2019 2:18 am
by ESP_Angus
eli_az wrote:
Mon Feb 18, 2019 9:17 am
- By did not turn on I meant the 1st of option of unresponsiveness from UART(the com was recognized by the computer due to the FT231X component), -
Just to make sure I 100% understand: you open a serial program at 115200bps, reset the ESP32, and see absolutely no output at all from the UART?

It seems maybe you have a bad board of some kind

Re: ESP Thermal considerations

Posted: Tue Feb 19, 2019 7:01 am
by eli_az
ESP_Angus wrote:
Tue Feb 19, 2019 2:18 am
eli_az wrote:
Mon Feb 18, 2019 9:17 am
- By did not turn on I meant the 1st of option of unresponsiveness from UART(the com was recognized by the computer due to the FT231X component), -
Just to make sure I 100% understand: you open a serial program at 115200bps, reset the ESP32, and see absolutely no output at all from the UART?

It seems maybe you have a bad board of some kind
Yes that was the case and it seems that cooling helped.