marqs wrote:
Counting and processing of equalization pulses is filtered out Macrovision stripper so pre/post coast does not need to cover them. Serration pulses during vsync I think are ignored by vsync processing circuit as line counting and sync processing work the same way regardless whether they are present or not. For TVP coasting is only needed if hsync period is not stable during pre/post equalization phases.
marqs wrote:
By default TVP7002 cannot decode interlace signaling via TTL sync inputs the same way it can't decode even-even signaling via analog sync inputs. RGB data from the chip is still generated ok while output sync is messed up (missing or extra vsync), ...
Thanks!
So this sounds like VSync detection and processing, using "Macrovision removal", works the same on both chips.
This confirms to me that I need to focus on getting the Macrovision settings right, then fix the remaining quirks with appropriate coasting.
However, I think the handling of equalization pulses is a little different.
For example, on the TV5725, I do have to coast 3 to 4 pulses more, depending on the source.
It will eventually be stable and reliable with all sources (that I can test with), with the longer coast setting.
Some sources are fine with pre/post coast at 1, but right now I've settled for pre = 3, post = 7.
Maybe it's just that the TVP7002 counts in actual scanlines, and the TV5725 counts in pulses it sees.
Something is causing that even-even problem on the TVP. Maybe it's that scanline vs pulse counting.
(Is Chrono Cross on the PSX a reliable "even-even bug" test? If not, are there other common consoles to test?)