Hi everyone. I have an Arducam 8.3MP IMX678 Manual Focus USB 3.0 Camera Module connected to an Nvidia Jetson Orin NX using the default UVC driver. The camera shows up fine in lsusb, but it never sends any frames. It was working before, then it just quit in the middle of normal operation and now I can’t capture any image frames. I also plugged it into a Mac and got the same result — detected, but no frames. Has anyone seen this?
Thanks for the details. UVC cameras like this one are usually pretty robust, so my first suspect is actually the USB cable rather than the module itself. A marginal cable can still enumerate the device (which is why lsusb sees it), but then fail once real video data starts flowing over USB 3.0. Do you have a known-good USB 3.0 cable you can swap in? If you have another working camera, try using its cable with the failing unit.
I’m a bit worried about that. These cameras are going to be deployed remotely, so I won’t be able to swap cables or make changes in the field easily. If this turns out to be a cable reliability issue, it could happen again on units I can’t reach. Right now I’m testing on a bench device (Jetson Orin NX, default UVC driver). I’ll try the new cables once my new batch of cameras arrives. To restate the symptom: the camera is detected in lsusb but sends no frames, it stopped working during normal operation, and a Mac shows the same behavior. It previously worked fine.
Understood. Every module and cable goes through a press test before leaving the factory, so I don’t think this is a systematic stability problem — it sounds more like an isolated failure on this particular unit. Once the new batch arrives, please try one of the new cables on the previously failing module and let me know whether it starts working. That should tell us if the cable was the issue or if it’s the module itself.
I got the new batch and tested the broken camera with one of the new cables — still nothing. So it looks like the module itself is the problem, not the cable. The new cameras are working fine so far. It’s a little unsettling that this unit passed the factory tests but then died during normal operation, but I’ll run some longer soak tests on the new batch and report back. To recap the original issue: detected by lsusb, no frames, stopped mid-operation, same behavior on a Mac.
Thanks for testing that. Since a fresh cable didn’t change anything and the same camera fails on two different hosts, this points to the module rather than the cable or the host. Based on what you’ve described, I’d treat it as an isolated unit failure — your new batch is working so far. I’d suggest running the new cameras through a longer stress/soak test before deployment, and if any other unit shows the ‘enumerates but no frames’ symptom, capture the dmesg output and any UVC streaming errors so we can dig deeper. For this specific module, the current conclusion is a hardware fault in the module itself.