The Orin Nano 4GB SOM is connected to the custom carrier board and camera sensor. The MIPI data rate is set to 2.5 GB/s with a dual-lane configuration. The serdes_pix_clk_hz is calculated and updated to the device tree.
During the camera sensor initialization, the deskew calibration signal is sent to the NVIDIA host and waits for the fsync signal. The v4l2-ctl command is executed, and the subdev driver toggles the fsync signal to receive frames.
For the first time, the camera sensor is sending the deskew calibration signal and frames are captured. From the next iteration, the camera sensor didn’t send the deskew calibration signal to the NVidia host, and frames are not captured with a timeout error.
Does the RCE firmware require the deskew calibration signal every time from the camera sensor before starting the frame capture.
On the Jetson Orin Nano NVIDIA target board, the capture.sh shell script is executed to capture and save the 9 frames to a binary file. When the camera sensor sends the deskew calibration signal, the frames are captured.
When the capture.sh is executed for the second time, the camera sensor don’t send the deskew calibration signal to the NVIDIA host and fails to capture with a timeout error signal.
I understood that a deskew signal is not required in-between the individual frame. But I want to know whether it’s required before the capture of every first frame.
From the below two sequences, can you please confirm the valid one for the RCE firmware
Camera Deskew signal → caputure.sh → Deskew signal → capture.sh
Camera Deskew signal → captures.sh → Without deskew signal → Capture.sh