We had a M0014 Tracking Camera and a M0107 Hires, configured correctly for camera group 1 on MIPI J7. We had done some test flights of this aircraft, Hires camera looked great, tracking camera was working perfect. The aircraft was not involved in any crashes.

While performing a stationary range test, the cameras stopped working. Running voxl-inspect-cam -a results in this error:

Voxl-inspect-services show that the voxl-camera-server is enabled but not running. Trying to restart the camera-server does nothing.

I tried all new cameras and cables, and it produced the same error.

Is something corrupted in the camera driver? Would reflashing the SDK potentially solve it?

Reconfiguring the cameras in group 2 shows the cameras working fine again. But as far as I know VIO tracking can only be set up in camera group 1. Is it possible to set up the VIO pipeline through a different camera group?

Final note, we conformal coated the Voxl2 prior to flight testing, but the MIPI ports are free from conformal coating or any other visible damage.

@Vinny It is possible that there are a few M0054-2's in the mix. Majority of the Voxl2's giving us issues at the moment are M0054-1's.

@Moderator

We have a daisy chain of M0076 to M0036 to M0025
and another chain with a M0084-1 with a M0149-1 and M0107-2 on it

Thank you for mentioning the limited cycles of insertions/removals. I will keep that in mind going forward.

@swickliff I am using Ubuntu 20.04 on both the VM and Linux PC

I'm trying to unbrick a VOXL 2 that has gone into EDL mode (Reading as QUAL_BULK). I've got the system into QDL mode and it is being read as (QDL MODE). No issues provisioning or flashing CDT.

Once I get to the Flash Flat Build stage, the system immediately sends the following failure {"message": "Flash flat build failed\n"}. I've tried on a Virtual Machine and on a dedicated Linux PC. Any thoughts as to why it may be failing?

Flash Flat Build: 2024-06-17-1.7.4-M0054-1

VOXL2: QRB 5165M

We have experience this issue with several Voxl2s configured the same way. I with 3 cameras on J7L, J7U, and J8L.

We first notice the issue when the camera streams are not coming up on our GCS.
We go into the voxl2 to investigate. Sometimes you cannot adb shell or detect the voxl2 through USB. Sometimes it allows you to adb shell but voxl-camera-server does not detect all the cameras.

Power cycling immediately typically results in the cameras still not being detected and/or the voxl2 not being detected as a USB device.
Power cycling and waiting ~5min or unplugging and re-plugging in all the cameras typically solves the issue.

Any idea why this might be the case? Is there a capacitor on the voxl2 that may take several minutes to drain fully?

The cameras and associated cables don't have any visible damage and will work on other voxl2 boards that don't have this issue.
We have had this happen to probably 5 or 6 different sets of Voxl2 and cameras.

We are at a loss for what could be causing this issue.
Any advice on how to investigate this further?

@Alex-Kushleyev , I believe we are not receiving any frames.

I disabled the hires and the same buffer error occurs.

Here is the readout from the dmesg -w command when I run voxl-camera-server in another terminal. I don't see any SOF or EOF messages:

@Alex-Kushleyev

voxl-camera-server -l detected both cameras

voxl-camera-server threw a few warnings and Error: Received "Buffer" error from camera: tracking

The whole read out is below:

We had a M0014 Tracking Camera and a M0107 Hires, configured correctly for camera group 1 on MIPI J7. We had done some test flights of this aircraft, Hires camera looked great, tracking camera was working perfect. The aircraft was not involved in any crashes.

While performing a stationary range test, the cameras stopped working. Running voxl-inspect-cam -a results in this error:

Voxl-inspect-services show that the voxl-camera-server is enabled but not running. Trying to restart the camera-server does nothing.

I tried all new cameras and cables, and it produced the same error.

Is something corrupted in the camera driver? Would reflashing the SDK potentially solve it?

Reconfiguring the cameras in group 2 shows the cameras working fine again. But as far as I know VIO tracking can only be set up in camera group 1. Is it possible to set up the VIO pipeline through a different camera group?

Final note, we conformal coated the Voxl2 prior to flight testing, but the MIPI ports are free from conformal coating or any other visible damage.

@Alex-Kushleyev It looks like I am having the same issue as Jacob. We've added a brand new Voxl2 to our drone, this one is running on SDK 1.1.3 and we cannot get it to connect to PX4. Our prior build used Voxl2s with SDK 1.1.2 and worked fine for us, so we tried downgrading the SDK but that did not work either.

We re-updated the SDK to 1.1.3 and I stopped and disabled voxl-px4 service and rebooted the Voxl2 and started px4 manually. Want me to share a text file with the output?

@Alex-Kushleyev Thanks for all the info! I'll keep those things in mind when soldering to the pads.