Hello, Guest the thread was called18k times and contains 283 replays

last post from Tobias at the

Helligkeiten der C64 PAL-Palette - zum 100. Mal

  • Ok - I'll try to find a time slot to do that some time soon. Wondering about the odd/even lines though. Since there is no interlace, half of the lines is going to be "empty" anyway so you mean the odd/even of the progressive raster, right?

    Yes. odd/even as it can get for the C64. ;)

    Das einfachste:


    https://nc.silverdr.com/s/TXTnsQ6SZ4ptLGL


    All lines 75Ω loaded.

    Wow. That was pretty fast. Thanks alot for the luminance measurements.:thumbsup:
    To me, meanwhile it looks as if the absolute voltage levels are not necessarily linked to each VIC-II-revision.


    Robert 'Bob' Yannes:

    "Since all of this was based on selecting different resistor values and resistance varied from chip lot to chip lot, there was variation from one Commodore 64 to another. It wasn't as bad as it could have been though, since all of the Chrominance selection was based on resistor ratios, which could be kept constant even if the actual resistor values varied. Luminance was more of a problem. A trimmer resistor should really have been used to pull up the output. This would have allowed the Luminance to be adjusted for consistency from unit to unit, however Commodore didn't care enough about consistency to bother with adjusting each unit."https://www.pepto.de/projects/colorvic/2001/


    So it really might be just a chip(and board?)-to-chip variation. Just in case you have a second C64 with the same VIC-II installed from factory... ;)

  • I hacked together something that lets you fiddle with the resistors and then get the effective color values at http://www.michiel.boland.org/lu.html

    This is based on 6569R5.

    The colors may be off, I'm still figuring those out.

    Cool. :)
    Just a comment: contrast and saturation sliders allow to go negative which makes some funny stuff. ;)

    I am also preparing a kind of Excel summary sheet which I can share with you.
    The ones that I did so far and that I am using so far are completely messed up meanwhile and only understandable by myself if at all.:rolleyes::D

  • That's the current status of the color angles.
    The lines are the assumed "ideal" target angles.
    The dots are the mixed color angles of odd/even lines. Produced by the angle-dependant phase (and also amplitude) errors of odd/even lines.

    So basically we need 4 PALettes:

    1. the ideal one ("hand-picked" palettes like CoCo or mine seem to "see" the ideal color hues sometimes alarmingly close to the ideal despite the phase error thing, might be some correction in TV? or just coincidence?, some VICs have smaller errors?)
    2. the mixed one

    3. the odd one

    4. the even one


    2 (and 1?) can be used as the general C64 PAL palette.

    3 and 4 can be used to simulate the PAL delay line mixing of equi-luminant colors where the mix color wil be slightly different depending on which color is used in which line.

  • Yes. odd/even as it can get for the C64

    I see. This will affect chroma only. There seem to be no measurable difference on the luminance levels between any of the adjacent lines.

    That was pretty fast. Thanks alot for the luminance measurements

    You're surely welcome! I still have to measure the chroma / angles though. This may take a bit longer.

    Just in case you have a second C64 with the same VIC-II installed from factory... ;)

    Yes - I kept in mind to check the same chips in other boards / assys so I may try that too

    if I am referring to your measurements, how do you want me to call you/them?

    I am not particularly precious about that :-) IOW "silverdr" is fine

  • I see. This will affect chroma only. There seem to be no measurable difference on the luminance levels between any of the adjacent lines.

    I can confirm this from the measurements of mathop. Just the color angle and amplitude "swing".

    You're surely welcome! I still have to measure the chroma / angles though. This may take a bit longer.

    Yes - I kept in mind to check the same chips in other boards / assys so I may try that too

    Many thanks. No rush here. Quality takes time. :D

  • Managed to get hold of an old C64C with the short board. Haven't done any detailed measurements yet. On my capture card the picture looks very bright, compared to 6561. It also has rather pronounced jailbars and checkerboard patterns. The color signal appears to be better behaved. The peak-to-peak voltages of all chroma signals, including color burst, are more or less the same, but there are still phase differences on even and odd lines.

  • Managed to get hold of an old C64C with the short board. Haven't done any detailed measurements yet. On my capture card the picture looks very bright, compared to 6561. It also has rather pronounced jailbars and checkerboard patterns. The color signal appears to be better behaved. The peak-to-peak voltages of all chroma signals, including color burst, are more or less the same, but there are still phase differences on even and odd lines.

    You mean 6569R5 vs. 8565R2?

  • I finishied the first draft of the PAL UV color model, based on watching many palettes. The even/odd model was inspired by mathop's measurements.
    As mentioned, the color model has 4 palettes.
    Just add Y to the UV-model and make the YUV->RGB conversion (WITHOUT any gamma!) and that's pretty much it.

  • Since there is no interlace, half of the lines is going to be "empty" anyway so you mean the odd/even of the progressive raster, right?

    Yeah, 50 "even" fields per frame. After 17 years in broadcast I didn't know this was even possible until I went back to the C64 and started to figure out why I was unable to do video captures of the C64 with decent capture hardware. You'll see that even the Teranex won't help you further, only the S&W standards converters can be made to output a stable picture with the Vidiplex button, omitting every second field, and adding to the judder this causes, a picture jump occurs every 8 seconds or so.

  • Right. I use Teranex (the early BMD one) chiefly for real-time deinterlacing. That's not a use-case with C64 though ;-) OTOH I have a few devices that can correctly handle the signal. And yes, older S&W (like the NRS-50) units had the "vidiplex" button but other than that one of the better handling the signal is e. g. ADVC-700. I remember paying large money for it as new in its heyday. Can be had cheaply these days if someone wants a device that behaves well with out of specs signal.

  • After 17 years in broadcast I didn't know this was even possible

    Hehe... I remember the moment I showed up in a broadcast studio the first time with a 64 many, many moons ago. And the guys connected it to their backbone and started scratching their heads. Reconnecting directly to the CRT - works. Connected to the studio backbone - black screen... Then they started running around bringing strange equipment pieces (now I know what it was ;-) and eventually saying out "shit - that's no CCIR/PAL. We need a workaround quick!".

  • Did some more detailed measurement of the color angles on various boards.

    Looks like 6561R3 and 6561R5 are more or less the same. The 6561R1 would probably also look like this, but I don't have that chip.

    The color generator was redone in the 8565 so there the angles are also significantly different.

    What's interesting is that in the 8565 the odd/even value for brown are the same, and the even angle for yellow is less than the odd angle (for all other colors it is the other way around.)

  • Did some more detailed measurement of the color angles on various boards.

    Looks like 6561R3 and 6561R5 are more or less the same. The 6561R1 would probably also look like this, but I don't have that chip.

    The color generator was redone in the 8565 so there the angles are also significantly different.

    What's interesting is that in the 8565 the odd/even value for brown are the same, and the even angle for yellow is less than the odd angle (for all other colors it is the other way around.)

    Sorry for my delayed feedback. I had some days off to find all of the easter eggs. ;)
    Wow, it keeps interesting. Thanks for the measurements! I just did a quick analysis.

    It seems that Commodore wanted to get the mixed colors closer to the "targets" for the 8565R2. 6 of the 8 mixed colors come pretty close to the target color angles. Just with the small drawback that the odd/even-deviation for blue increased to ~20°, before it was about constant ~17.5° for all colors.
    That should make a big difference for the PAL delay line even/odd equi-luminant color mixing between 6569Rxs and 8565R2.


    Any electrical obsessed volunteers here to explain which (changes to the) circuitry could have such effects? To a non-elecctrical guy.:saint:


    silverdr Happy to see your 8565R2 measurements to have further data

    mathop How about the color amplitutes? Did you measure them for 8565R2?

  • The 8565R2 color amplitudes appear to be more or less the same, regardless of color or odd/even line.

    How about PURPLE? The odd/even-difference looks similar like 6569Rxs. Just opposite "phase". Here also no color amplitude difference?

    But that could also be because the voltages are lower, so variations in amplitude tend to get lost in the noise.

    It means that the "saturation" of 8565R2 in general might be a bit less?

  • On an 8565R2 the color burst amplitude in the composite signal appears to be 0.15V - exactly what it should be by the PAL standard. Then all the colors also have this 0.15V amplitude.

    On 6561R5 the color burst starts off at about 0.15 - 0.16 V, then drops to 0.12 - 0.13 as the chip (and/or modulator?) gets hot. The color amplitudes are about 1.1 - 1.3 times that of the color burst. So in theory the 6561 should look slightly more saturated, but I'm not sure if it is noticeable.

    On 8565R2 the color amplitude appears to remain constant as things heat up.

  • On an 8565R2 the color burst amplitude in the composite signal appears to be 0.15V - exactly what it should be by the PAL standard. Then all the colors also have this 0.15V amplitude.

    It means 0.300V peak-to-peak, right? But it is just "random" that the C64 uses exactly the color burst amplitude also for the color "saturation", right? I assume it helped to save some parts/money.

    So the color "vector" length sqrt(U²+V²) should also be like 0.300 for the 8565R2?


    The color amplitudes are about 1.1 - 1.3 times that of the color burst. So in theory the 6561 should look slightly more saturated, but I'm not sure if it is noticeable.

    Factor 1.1-1.3 applies just for the 6569s or also for the 8565R2? Do you know where the factor comes from?


    On 6561R5 the color burst starts off at about 0.15 - 0.16 V, then drops to 0.12 - 0.13 as the chip (and/or modulator?) gets hot. The color amplitudes are about 1.1 - 1.3 times that of the color burst. So in theory the 6561 should look slightly more saturated, but I'm not sure if it is noticeable.

    For your composite signal measurement on the 6569R5, 0.225 seemed to be an average for the color ampitude. Swinging around +-10% depending on color angle deviation to target.

    That would be amplitudes of (0.12x2)x1.1 to (0.13x2)x1.3 for heated up chips. 0.264 to 0.338



    Brown color amplitude should be less than the others I think. I assumed a factor of 1/sqrt(2).

  • The 8565R2 color amplitudes appear to be more or less the same, regardless of color or odd/even line. But that could also be because the voltages are lower, so variations in amplitude tend to get lost in the noise. I haven't made any detailed scope measurements (the above measurements were made using a capture card.)

    Do you have the RGB values for 8565R2 odd/even from your capture card? I could have a quick check of the color amplitudes.