This post is related to a previous post by a colleague: https://ez.analog.com/thread/47992 but we have gathered more information.
I am developing a system which is responsible for digitising and processing a PAL analogue video signal and am experiencing chrominance distortion in the top 26 video lines.
The video source being decoded is based upon composite PAL, but is a slightly non-standard PAL video source. The source video is a time division multiplexed video stream of multiple sources (1x colour, 4x monochrome). The PAL colour burst is sourced from the colour input and replicated in all frames – even those of a monochrome source. The time division multiplexing is frame based, thus the odd and even fields are always derived from the same source. The unit performing the video multiplexing does so in the analogue domain, and is a legacy piece of equipment we are unable to modify.
When digitised, the frame from the colour source exhibits the chrominance distortion. The luminance information looks to have been decoded correctly, but the colour information is incorrect.
Zooming in on a captured image it seems the distortion only occurs in one of the fields as alternate lines are correctly decoded.
If I change the time division multiplexing such that there are consecutive frames output from the same colour source then the chrominance distortion is still present. Therefore this issue occurs on every frame where colour information is present (for reference monochrome frames decode fine), and does not seem to be an artifact of the time division multiplexing between multiple video sources.
This result has been replicated using an ADV7182 evaluation board, which was initialised using the CVBS AIN1 PAL video script. The same chrominance distortion is observed on the video output.
When the ADV7188 is connected directly to a reference PAL video source such as a pattern generator, there is no sign of the chrominance distortion.
I have made some initial measurements of the PAL signal we are receiving and compared them to the PAL specification, the results of which are summarised below:
Horizontal sync pulse width
Colour burst start after falling edge of horizontal sync pulse
Colour burst duration
Start of active video after falling edge of horizontal sync pulse
Odd field start and end of active video
Even field start and end of active video
If there are any other useful measurements, or scope traces required then let me know.
Could the issue be that the colour burst appears slightly earlier than standard and are there any registers that may affect where the decoder looks for the burst?
Any ideas or recommendations are also greatly appreciated.