Microcontroller resetting only at high temperatures - Thermal issue or decouplin

[James Anderson]

Hi everyone,

I'm currently troubleshooting an issue with a batch of custom control boards we recently assembled.

Everything works perfectly during bench testing at room temperature (around 25°C). However, once we put the board into the sealed enclosure and the internal temperature climbs to around 60°C - 65°C, the main MCU randomly resets.

We checked the soldering, and there are no dry joints. The decoupling capacitors (0.1uF and 10uF) are placed as close to the VDD pins as possible. We are suspecting either the LDO voltage regulator is thermal-throttling and causing a transient voltage drop, or it's an internal oscillator drift due to the heat.

Has anyone experienced similar thermal-induced reset issues with MCUs in actual applications? Any specific test points or diagnostic steps you would recommend to isolate this?

[PCB.Wiz]

What are the MCU specs ?
Also keep in mind that MCU speed decreases as they heat, so it can then get erratic if it is outside the Voltage/Temp/MHz envelope.
Cheap/small capacitors also change significantly with temperature.

[voltsandjolts]

Most LDO's won't thermally limit until their die temperature is over 125C at least. So unless you have high current situation with hot LDO at 25C ambient, then I doubt its that.

Anyway, I don't have a crystal ball and you've provided near zero info.
Which mcu?
Where's the schematic?
What's the reset circuit like? High impedance with leaky fet for external reset control?
Did you try measuring/logging Vcc at high temperature?

[mtwieg]

Any specific test points or diagnostic steps you would recommend to isolate this?

Most MCUs can tell you the reason for their last reset (POR, BOD, NMI, external reset, etc). Look there first.
Monitoring the supply voltage(s) for the MCU during testing would also be wise.

If your MCU relies on an external crystal or oscillator, that may be the weak point as well.

[Jeroen3]

Well, time to get the hot-air out then.
Go around selectively heating parts of the circuit to see when it resets.

[voltsandjolts]

Don't waste your time folks.
The OP has started two other threads with similar "help me debug" theme and never bothered replying to folks messages.
5 posts, 3 of which are new topics.