I'm using an Arducam imx708 with m12 120deg lens (uc-a65), as the sensor is then mounted on the pcb, and it has mount holes for the lens. They are close enough for my tests though.
You are correct the lens has no home position switch, so the ony option is to drive it into the end stop on startup.
I'm still in the process of sorting drivers.
The current libcamera AF algorithm expects the lens movement to be near instanteous, which a stepper motor obviously isn't. There have been discussions on the Linux media mailing list over reporting back lens position for slower assemblies, so I'll be trying to work within the suggestions from that.
On my setup it'll be an rp2040 driving a couple of h-bridge drivers (https://amzn.eu/d/aS01Us3), with probably a couple of pwm outputs to drive servos for pan and tilt. In that regard it won't to be too far away from what I suspect arducam's board to be.
You are correct the lens has no home position switch, so the ony option is to drive it into the end stop on startup.
I'm still in the process of sorting drivers.
The current libcamera AF algorithm expects the lens movement to be near instanteous, which a stepper motor obviously isn't. There have been discussions on the Linux media mailing list over reporting back lens position for slower assemblies, so I'll be trying to work within the suggestions from that.
On my setup it'll be an rp2040 driving a couple of h-bridge drivers (https://amzn.eu/d/aS01Us3), with probably a couple of pwm outputs to drive servos for pan and tilt. In that regard it won't to be too far away from what I suspect arducam's board to be.
Statistics: Posted by 6by9 — Sun Feb 25, 2024 2:28 pm