Latching/Interrupt based digital inputs #10

New Issue

o.winkels · 2025-07-03T13:55:09+02:00

o.winkels commented

2025-07-03 13:55:09 +02:00

Currently, the Lightgate reflections are usually missed because we only send the pin state at each sending interval. That's stupid.

Solution A, the simple but bad one:

Declare the pin as latching, sample at 1kHz and if during one sending interval we've seen a 1 at any point in time, send a 1 after the interval.

Problem: Even at 1 kHz at max vehicle speed, the reflection zone needs to be at least 2.2cm wide, which is way too close to the usable width of the cone to be a safe reading. Also, this might explain why the readings were always dodgy even with the Autobox, which also sampled at 1kHz

Solution B, the proper one:

Use Pin Interrupts (CubeMX Pin Mode GPIO_EXTI) to detect even shorter pulses.

Annoyance: Usually we'd do this through CubeMX, but we don't want 2 separate IOCs. So ideally we'd either use Pin interrupts both in the front and the back (either for those 2 pins or for all of them) or we re-implement the code gen's interrupt config using our pin configuration system.

Currently, the Lightgate reflections are usually missed because we only send the pin state at each sending interval. That's stupid. ## Solution A, the simple but bad one: Declare the pin as latching, sample at 1kHz and if during one sending interval we've seen a 1 at any point in time, send a 1 after the interval. Problem: Even at 1 kHz at max vehicle speed, the reflection zone needs to be at least 2.2cm wide, which is way too close to the usable width of the cone to be a safe reading. Also, this might explain why the readings were always dodgy even with the Autobox, which also sampled at 1kHz ## Solution B, the proper one: Use Pin Interrupts (CubeMX Pin Mode `GPIO_EXTI`) to detect even shorter pulses. Annoyance: Usually we'd do this through CubeMX, but we don't want 2 separate IOCs. So ideally we'd either use Pin interrupts both in the front and the back (either for those 2 pins or for all of them) or we re-implement the code gen's interrupt config using our pin configuration system.

o.winkels added the

bug

feature

labels 2025-07-03 13:55:12 +02:00

o.winkels commented

2025-07-03 13:57:15 +02:00

@l.gies hat Option A schon irgendwo implementiert und könnte bald ausprobiert werden