I am working on a board that has two TVP5150AM1 decoders. One decoding external video and one decoding the output of an Analog Devices video encoder chip. The following issue happens with both 5150's and doesn't matter what the source video is. I am using the current version of the RAM firmware update for both 5150's (v048CAA)
When feeding in NTSC video the TVP5150AM1 generates discrete vertical blanking (VBLK) that has a different length blanking on the two fields. It outputs 20 lines of blanking in the first field and 21 lines in the second field. (I'm not talking about half lines vs full lines, I'm talking about 20H time period vs 21H time period)
When feeding in a PAL signal the TVP5150AM1 generates a VBLK that by default is 23 lines long in both fields.
The PAL one is not really a problem, since the width of vertical blanking can be changed with VBLK Start and End to be the correct 25 lines that PAL is supposed to have. However the NTSC one is not as simple. No matter how you adjust the start and end of VBLK in the 5150, there is always one more line of vertical blanking in one field vs the other. NTSC defines vertical blanking as 20H in both fields.
This problem is very simple to see on an oscilloscope. Just look at the VBLK signal with the timebase set fast enough so both fields are overlaid (ie: 200uS/div on a typical scope with 10 divs). When you feed in a PAL signal you'll see that both fields have the same length VBLK. If you feed in NTSC you'll see that the end of the waveform is flickering for the width of 1H, which shows the different lengths of VBLK in the two fields.
