MIPI Errors explanation

  • Hello, we haven't encountered the "ERROR: csi fifo overflow" issue ourselves, so I can only help troubleshoot based on experience.

    In general, an "ERROR: csi size err" usually indicates an unstable connection or an excessively high data rate. In such cases, it's important to check whether the PCB layout and ribbon cable connections are stable, and to consider adjusting the relevant clock settings.

    However, since you're not seeing that particular error, you can probably rule out hardware layout or clock configuration issues for now. It's more likely a software configuration problem. I recommend first verifying whether you can successfully capture raw images using v4l2-ctl.
  • Hello,

    actually, we believe it is some sort of combination of hw issues. A bit of low-end camera module and not the best cable connecting the board. However, while fixing that issue we figured out we can avoid it, by closing and opening camera (cap.open(0) in opencv) everytime we need to take a photo. We need a photo every few minutes so that is actually quite a big deal when it comes to power usage. But after every few images the board completely freezes. The logs stop and I need to manually reset the board to gain access to it again. I think while closing the isp stream something majorly crashes under the hood which stops the whole board. The last entry I can see is "start stopping the isp" or something similar, which indicates that the ISP flow should stop, but doesn't 🫤

    I'm now searching for a solution on how to close it properly so that we don't hit this bug.
  • Hello

    As an update to this topic, we're still searching for a solution to the original problem when the MIPI will drop images. The current idea is to lower the clock settings for the camera and the communication on the sensor to stabilize the system since we only do individual images (not the full stream). We are not really sure which setting to change to correctly lower the clock and stabilize the transport of the data. We are still encountering the csi overflow issue so we want to stabilize the transport as low as possible. With shielded cable the issue does seem to be less frequent but it still happens. Can you point us in a direction how to stabilize this communication between the RV1106 and the camera SC3336?
  • We also have no development experience for this usage scenario. If the shielding line can improve the situation, it might be caused by the layout.
    If there are no requirements for frame rate, you can try changing the driver of sc3336 to use the 25fps mode, reducing the speed of the PHY to gain stability.