Problem with debugging a chicken door project

[leviathan_engineer]

@nicolas_schlafer

Hello !
First post here, so don't hesitate to tell me if it's the wrong section or anything.

I'm working on yet another automated chicken door.
I was happy of the first breadboard test phase, and soldered it into a protoboard for testing.

The project is powered by a small solar panel + controller.
It use a RTC3231 to setup the opening and closing time, a L298N for the motor, and a relay to switch off the L298N when not in use. There is a switch for auto/manual mode, with button to control the door in manual mode.

I installed the circuit in test mode, and it works as it should, for a few hours, or even a few days, but then it start to behave erratically:

• In manual mode, when i pressed a button, there is a short action but then the relay and L298N switch off directly.
• The end of course switch are not reconnected anymore
• The time of the RTC3231 gets off
• I feel that without the L298N plugged the circuit works fine but i should test it for an extended period of time to be sure of that

All these erratic behaviour are fixed if i reboot the arduino by switching off/on the power.

Another suspicious thing, at some point i disconnected the 12v power and plugged the USB to the arduino, the arduino USB port smoked directly. But it is still working. (i changed it still in case, to be sure it is not the problem).
Note: This was before adding the Schottky diode after the stepdown +5 DC output, i did not retry it, didnt want to fry another arduino

What i suspect could be to investigate:

• The 5v regulator of the L298N is overheating
• Needs some decoupling caps or other electronic components i missed (I'm a beginner)
• Some bad soldering, but i find it weird that it works fine for hours first.
• Instability in the voltage or ground loop

Ideas i had for now:

• Replace the relay by Mosfet/transistor system to cut more properly the L298N 12v and 5v, this time not using its internal voltage regulator for its logic part.

I would be very interested to have troubleshooting ideas from more experienced people around here to put me back on track !

Here is a KiCad schematic

Hey! So just based off what you said, here's a list of possible issues that might be occurring:

1. Grounding issue or ground loop – Causes erratic behavior; ensure all grounds are tied at a single point.

2. Voltage dips or unstable 5V rail – Use decoupling caps (100nF + 10µF) near ICs, especially RTC and Arduino.

3. Back-EMF from motor – Add flyback diodes across motor terminals even with L298N.

4. L298N’s 5V regulator overheating – Bypass it; power logic separately.

5. Relay switching noise – Causes microcontroller brownout/reset; add flyback diode and snubber across relay.

6. Inrush current or poor solar charging – Capacitor bank may help; verify stable power under load.

7. Bad solder joint or cold joint – Thermal cycling reveals it later.

8. RTC losing time – May be underpowered or I2C interference; decouple and isolate SDA/SCL with resistors.

9. USB smoking – Likely 5V backfeed from step-down into USB; Schottky diode fix was correct.

Check those, especially power and grounding.