Re: SX clocks generation (was: 6569 luminances)

On 2013-05-03, at 02:01, Segher Boessenkool wrote:

>> Meaning the level of the luma varies from pixel to pixel (or more than one pixel - haven't counted/measured so can't say now exactly how wide the period of this wave is but looks close to two pixels) creating various patterns in place where a solid colour would be expected (at least horizontally). Various, because it seem to depend on which colour is actually selected.
> 
> There are nine vertical stripes per sixteen pixels.

Well, so I don't have to bother with perusing it ;-)

> Internally on the VIC-II, there are two colour clocks, 90 deg out
> of phase; one of-em is flipped polarity every other scan line.
> So you see the "stripes" when the non-flipped component is dominant
> for your colour, and the "checkerboard" when the flipped one is.

That would explain. So in some colours one of the clocks is "dominant" and with other colours the other, it seems. But what means "dominant" here? That both are used with different level? Those are clocks (as you wrote), not analogue signals, right? So how can they be "dominant"?

>> I am wondering if anyone knows (before I start disassembling the SX in order to experiment) whether this effect can be caused by the differences in the clocks generation methods between regular 64 and the SX or something else (like the output amp).
> 
> It looks to me like the chroma signal is imprinting on the luma
> signal; this happens somewhere at the output end, not at the
> clock gen.

Might be. The main difference I see is the clock generator circuit, which is completely different than the one found in a 64 (any version) and the output stage, which is also different due to lack of the RF modulator. Interestingly the schematics found on zimmers is only for the NTSC version and NTSC version of the SX does /not/ exhibit this behaviour... 

>  But you haven't described the clock gen construction,
> so we cannot really tell :-)

:-) Yeah, if nobody has any info on it, I'll have to reverse it by hand.. one day.. What I like about this circuit is that it is built using easily available LS chips, total cost of which is about €0.5... Sure there are passive elements in a number too and the second crystal, which add to the cost but still.

-- 
SD!
       Message was sent through the cbm-hackers mailing list