OpenMV RT1062 Legacy 5 VIN Pin USB weirdness + workaround

OpenMV Products
1

I have an issue with the power supply on the OpenMV RT1062 (Legacy). Firmware 4.5.9, IDE 4.4.4

Here’s what happens:

  • I connect the OpenMV with USB to my Macbook Pro and run code.
  • The code tells the robot to supply 5V on the VIN pin for the OpenMV. (It also puts 3v3 on the 3v3 pin).
  • I stop the script, the robot stops supplying power on Vin, but maintains 3v3.
  • I restart the script
  • The robot starts supplying 5V on the VIN pin again, and maintains 3v3.
  • The OpenMV board crashes hard: I lose USB connection, unable to reconnect. The power button on the OpenMV does not reset it anymore. Only disconnecting all wires works.

If I do the exact same scenario with a voltage meter on the OpenMV Vin pin, everything is fine. I can stop and restart at will. Since the Vmeter has a non-zero resistance, I imitated that circuit by adding a 1MOhm resistor on the OpenMV between Vin and Gnd. This also works perfectly and it is my workaround. What I don’t like is that I don’t understand the reason why this workaround works. I also feel queasy about shorting Vin all the time.

Do you have a better solution?

20241110_175946
20241110_1759461920×1920 234 KB

20241110_180045
20241110_1800451920×1920 336 KB

2

If the power button doesn’t work then that means the MCU got into a bad state hardware wise. The supplying 3.3V into the camera isn’t really the correct usecase with 5V on VIN at the same time.

Putting on my EE hat… I’m guessing you’re getting a reverse current draw backwards through the VIN pin when your regulator is turning on which is causing a short to ground glitch which is briefly causing the voltage rail to sag.

If you scope the 3.3V rail on the camera you most likely will see if dipping when you enable the 5V source. The resistor pulling the input to ground likely discharges some of the caps on the input to the ideal diode, which resets it such that it doesn’t allow reverse current initially.