I had an idea regarding PAL (and perhaps the Uzebox in general) the other day and thought I'd share. I'm mostly (if not only) a programmer and I'm fairly new to the AVR world (though I did manage to build a Uzebox on a breadboard
), so I'm not actually sure if any of this makes sense, but bear with me...
From what I understand, the MCU has to run at twice the clock speed required for the 4FSC input of the AD725. A timer interrupt makes PB3 pulse every other clock cycle.
If that is indeed all there is to it, why not program the CKOUT fuse, wire CLKO (PB1) directly to the AD725 4FSC input and run the MCU at half the speed it runs at now? That way, an NTSC Uzebox could run at 14.318180mhz (no overclock required) and a PAL one could run at 17.734475mhz - very doable!
When I first thought of this, I figured that halving the Uzebox speed would severely cripple its power, but then this occured to me; in order for PB3 to produce a burst every other cycle, it needs an instruction, right? Which means HALF of all instructions executed are used for pulsing and the remainder is used for processing. If one eliminates the need to pulse by using CKOUT, should that not theoretically keep the Uzebox running at the same processing speed? Instead of having half the cycles available at 28mhz, you'd have all cycles available at 14mhz.
I'm probably completely wrong about this one, but I thought I'd share it anyway. I would very much like it if someone can explain why this wouldn't work (assuming it doesn't). I'm thinking my perception of how the internals tick might be way off. I browsed through the Uzebox source, but couldn't find anything to either validate or refute my ramblings. Though that might be because I don't really read ASM very well