The strange case of the Commodore 64 colour bands.

[stephen_usher]

OK, I managed to get onto the Lemon64 forum. My spam filter was rejecting the activation e-mails with a score of 7.5 (the OK limit is 5.0). After turning off spamd I could get the e-mail to come through.

Well, well... After being able to use the search function I discover that... These colour bands are normal for PAL C64 machines up until the C64C, when they changed the design of the VICII chip.

So, basically it's a design flaw in the PAL colour generation system.

Given that the design team for the ST were mostly head-hunted from the C64 design team you can see where some of design bugs came from.

[rubber_jonnie]

OK, I managed to get onto the Lemon64 forum. My spam filter was rejecting the activation e-mails with a score of 7.5 (the OK limit is 5.0). After turning off spamd I could get the e-mail to come through.

Well, well... After being able to use the search function I discover that... These colour bands are normal for PAL C64 machines up until the C64C, when they changed the design of the VICII chip.

So, basically it's a design flaw in the PAL colour generation system.

Given that the design team for the ST were mostly head-hunted from the C64 design team you can see where some of design bugs came from.

Oh damn, so no fixing it then?

[exxos]

You can get video filter buffer chips to clean up signals, I used them in my 2600 mod. Like I said earlier, screening is important. But probably most is down to bad PCB like some other machines we know about. More caps on top of chips, maybe even place a couple ohms in the power rail to some chips see if it helps,

[stephen_usher]

Unfortunately it’s not a “noise” issue but a fundamental signal generation issue in the VIC-II chip itself. Details on the Lemon64 board are almost non-existent though other than someone talking about the dot clock being out of sync with the colour clock. I’m guessing that there’s some sort of frequency leakage in the chip causing a beat frequency to taint the chroma signal. It only affects PAL machines.

The VIC-II chip in the PAL C64C is said to fix the problem but it’s electrically incompatible with the original chip.

[rubber_jonnie]

Unfortunately it’s not a “noise” issue but a fundamental signal generation issue in the VIC-II chip itself. Details on the Lemon64 board are almost non-existent though other than someone talking about the dot clock being out of sync with the colour clock. I’m guessing that there’s some sort of frequency leakage in the chip causing a beat frequency to taint the chroma signal. It only affects PAL machines.

The VIC-II chip in the PAL C64C is said to fix the problem but it’s electrically incompatible with the original chip.

I have a C64C and the image is perfect, so it's a shame you can't use the updated VIC-II.

[mrbombermillzy]

Sorry to hear all this.

Maybe it now time you thought about a modern day VICII replacement?

I've been keeping a very close eye on the Kawari project:



Also, Bill Herd has been doing some looking at the C64 schematics for the 40th anniversary. Maybe not too much on the VICII, but it may be good to see if he mentions something in passing about the colour bands (seeing as the 128, which was HIS design also suffered this problem; albeit on the VICIIe):

[stephen_usher]

There is only one VIC-II replacement project and that's had to go on the back burner due to no FPGAs being available, and it's not ready yet anyway.

The later VIC-II chip apparently fixed the problem, so knowing about the later machine doesn't really help. The schematic for the clock-circuit is easy to understand and the later replacement with a single chip makes no difference as it's intrinsic to the earlier PAL VIC-II chip.

[stephen_usher]

Well, as part of my preparations for the retro display at Lavecon I fixed a display I rescued from work, which used to be part of the CCT system. I have to solder back onto the PCB one of the BNC composite video connectors. (Unfortunately, even though it does 15KHz on the other inputs it can't do so on the VGA connector. Booo!)

Anyway, this display has S-Video input, making it ideal for the C64.

So, this evening I hooked it up along side my LCD TV, feeding the sound and composite video into the TV and S-Video to the monitor.

The colours on the monitor are nicer and there are far fewer "artefact colours", e.g. (look at the thumbnail of the hooded figure in the text box)

IMG_2720.jpg (376.68 KiB) Viewed 73 times

I always thought that the character should be really colourful, but as it turns out he should just be grey under the cowl.


Anyway, S-Video hasn't got rid of the banding, but it is lessened a bit:

IMG_2725.jpg (429.13 KiB) Viewed 73 times

The other thing I noted, which is important for games, is that the TV has a LOT of lag on the composite input. The monitor is a good quarter of a second ahead of the TV! In this image look at the "shooting star" passing through the letters of "CHRONICLES". On the TV it's just reached the top-right edge of the R next to the Moon but on the monitor it's already passed though the R and it's 2/3rds of the way through the H.

IMG_2723.jpg (308.77 KiB) Viewed 73 times

For a shoot-em-up game this lag would be bad for gameplay.

P.S. This monitor doesn't have sound input but there are locations for speakers and the control menu has sound options. There's a header on the board which looks like it may have been for the speaker connection and a location on the board for an input jack to be solder on, so I may try adding audio. This would be very useful. Instead of a jack I may add a twin phono connection instead.