PAL signal vsync variants?

[Artlav]

I'm working on generating a PAL video signal, and came across a certain ambiguity between the theory and the practice.

Specifically, the descriptions of the PAL signal, i.e. http://martin.hinner.info/vga/pal.html , specify that the vertical sync should be a sequence of 6 half-line short signals, then 5 half-line long signals, and 5 more short half-line signals.

However, in practice implementations for CPLD and MC only do 3 full-line long signals at the start of the frame.
And it seems to work generally fine on the modern TVs i have.

So, the question is - is that simple VSYNC version a proper solution?
What problems could there be from it, compared to the full one?
Some of the cheaper TVs have had problems locking on to the colour when i tried the simple version - can it be the cause? (color generation might also not have been up to spec).

 

[betwixt]

I think you are referring to "equalizing pulses" which should be present to meet full PAL specification but in most instances are unnecessary. In the days of direct oscillator locking they were imortant but for a long time almost all TVs have used PLL systems with relatively long time constants so they are far less critical of the signal.

As far as color generation, again almost all systems generate ident by simply switching between two phases on alternate lines. The true PAL spec says they should advance 90 degrees per frame so it takes 4 rather than 2 frames to return to the original phase.

You might be interested in this article: http://www.atv-projects.com/Testcard_Generator.html and it's two follow ups.

Brian.

 

[Artlav]

I think you are referring to "equalizing pulses" which should be present to meet full PAL specification but in most instances are unnecessary.

These too, but i'm mostly interested in the 5 half-line long pulses specified vs 3 full-line pulses that work

 

[betwixt]

I suspect that 3 full vs. 2.5 half has something to do with whether the picture is interlaced or not. To interlace the picture there has to be an extra half line delay on one field so perhaps that difference is to offset the scan. In theory there is an additional phase shift in the color burst as each field starts but I wuld think it extremely unlikely that would cause color problems. It's far more likely if short-cuts have been taken with the sync timing and the pre and post burst delays are wrong so that the full 11 cycles of subcarrier are not being gated properly. As the burst gate in the TV is 'opened' by timing it from the edge of the sync pulse, it's quite critical that they are inserted in the right place, particularly after the vertical sync interval where many encoders omit to add burst at all.

Brian.

 

[schmitt trigger]

The way you are describing it, also understand that those are the equalizing pulses, as betwixt explained.

 

[Artlav]

Thank you for clarifications.
I've gradually got the timing to full spec, with no luck on the colour, only to find out that the colour issue was due to the (borrowed) analog part that i assumed to be right.
Basically, on a real PAL signal, the colour burst is about a third of the sync-blank distance, while mine was nearly invisible on the scope - some TVs do lock on to it, others can't detect it.