ESP32 Custom PCB Occasionally Resets When Switching Higher Current Loads

[mariawilliams1719-sudo]

Hi everyone,
I'm working on an ESP32-based control system that monitors a few sensors and switches several DC loads through MOSFETs. The prototype worked well on development hardware, but after moving to a custom PCB I'm occasionally seeing unexpected ESP32 resets when loads are switched on or off.
The design currently includes:
ESP32-WROOM
MOSFET-controlled DC loads
Temperature and voltage monitoring
External 12V power supply
Custom PCB
A few observations:
The resets are intermittent rather than consistent.
Sensor readings appear normal before the reset occurs.
The issue becomes more noticeable when multiple loads switch at nearly the same time.
The firmware itself appears stable during normal operation.
A few questions:
Are additional decoupling capacitors or TVS protection commonly required in these types of designs?
Could EMI or ground bounce from the switching loads be causing the ESP32 to reset?
What PCB layout practices help reduce noise coupling into the MCU?
How aggressively should power and signal grounds be separated?I'm also evaluating locking Molex connectors for the power and sensor connections, so I'd appreciate any advice before finalizing the design.
Thanks!

[Josverl]

Hi,

Do you know what reset reason is reported at boot after reset?

Capture the serial boot log after a reset, or at least report the restart reason in a log file. machine reset_cause()
The ESP32 reset cause is often very informative.

If it is a brownout, there is something happening to the MCU power, if it is a reset. It may be noise triggering the !EN.

As the radio on the esp32 can be a bit power hungry at times, it would be useful to determine if resets are more/less frequent with the WLAN off, or under load.

As I'm not an electronics guy, the only other thing I can recommend is to get the scope out, and monitor the power and enable lines

[sosi-deadeye]

Maybe it's a brown out. Are you able to see the debug output via UART?

[peterhinch]

There are a number of possible causes. It would help if you told us something about the load you are switching in terms of voltage and current, and whether the load is purely resistive or has an inductive or capacitive component.

A possible cause is resistive voltage drops in the Gnd connections. It's important to consider the current path in Gnd when the load is on, to ensure that voltage drops do not affect the logic levels. You can also get transient effects, especially if the load has inductive or capacitive components. Decoupling can help here. Inductive loads can require overshoot protection.

As @Josverl says, you need to get a scope on the job to find out exactly what happens when the load is switched on, and when it is switched off.

[DavesCodeMusings]

This is going way back to my school days, so take it as you will. Many times strange circuit behavior can be solved with proper bypass capacitors across a chip's power supply leads. A 10uF electrolytic in parallel with a 100nF ceramic from Vcc to ground usually does the trick. Maybe you already have this on your custom board. I only mention it, because 9 times out of 10, it solved the "weird problems" with circuits. You might even go larger on the electrolytic.

[GitHubsSilverBullet]

First thing which comes to mind is the reset pin itself. Is it using only the internal pull-up? That'd be quite sensitive to outside disturbance. An external 1kΩ pull-up and a 100nF to ground might help a lot.

[peterhinch]

@mariawilliams1719-sudo I suggest searching on "Star grounding" for advice on PCB layout.