We use the DM642 video port with TI video driver to capture video from a color camera, the output from camera is NTSC analog signal, we use the analog device's ADV7180 decoder to get digital ITU-R BT.656 stream with embedded v/h sync codes feeding to DSP video input port. Our problem is we experience some video shift in the video capture or meaning that video is incorrectly locked with the wrong v or h sync position. This problem has happened a few times in our product in customer installations, but it is a problem hard to produce in our lab. The video lose lock for some reason and then stays 'mis-locked'. Has this kind of issues been seen with the DM642 video port hardware or video port driver before? We are suspicous the TI video driver may not handle this kind situation well, but we do not have enough proof. The video port should be able re-sync (or inform loss of sync) because it can interpret the H-Sync & V-Sync codes in the BT656 stream. It should be able to detect a FIFO under-run or FIFO over-run if the current video line being received is interrupted by a H-Sync code, or gets too many pixels before a H-Sync code. Likewise with V-Sync codes and the number of lines the video port is counting as received. The incoming video is 720x240 fields and we use the horizontal downscale feature of the DM642 video port to achieve 360x240 fields - unsure if this is relevant to the issue we are seeing. Another possibility might be the video port driver's EDMAs getting out of sync with the video port FIFOs?? Thanks for your assistance.
↧