Re: vic-ii pal colors

On 6/7/20 4:50 PM, Michiel Boland wrote:
> Hi. I noticed that if you capture analog C64 video with the comb filter turned 
> off there is quite a difference in color saturation and/or hue between even and 
> odd lines.
> 
> I thought that the colors were generated by adding a weighted sum of sine and 
> cosine signals, but inverting one of the inputs (which is done in odd lines in 
> PAL systems) should not have any effect of the overall amplitute, unless I am 
> very much mistaken.

So I did a little more thinking, this is what I have come up with. Note that I 
don't claim to have any knowledge on chip design, or electronics in general :). 
This is just a guess.

Take the color blue for instance, this is generated on the chip by combining an 
anolog 'sine' wave with an inverted cosine, a bit like this (ascii art)

    sin
     |
    +++
    | | R1
    +++
     |
     +-- output
     |
    +++
    | |
    | | R2
    | |
    +++
     |
    -cos

where R1 and R2 are resistors, and R2 is significantly larger than R1. Now since 
R2 is larger it also has a larger parasitic capacitance than R1, so the -cos is 
phase shifted to the right, relative to the sin signal as it is mixed into the 
output. The effect is that the amplitude is too high during even lines, and too 
low during odd lines (when the -cos line is inverted.) Also the phase of the 
output will be off by some amount.

This would also explain why the even-odd difference is greatest for colors at 
the 'edge of the quadrants' (blue, red, yellow, cyan) and less pronounced for 
the other colors, (purple, orange, brown, green) since these have less 
difference between resistors. In fact the resistors are equal for the color 
burst, so the color burst has the same amplitude on odd and even lines.

Any thoughts?

Cheers
Michiel