Re: incorrect C64 wikipedia page?

Den Sat, 10 Nov 2018 15:27:15 +0200 skrev Marko Mäkelä
<msmakela@gmail.com>:
> On Sat, Nov 10, 2018 at 03:48:00AM +0100, Mia Magnusson wrote:
> >This page states that there were only NTSC C64's made with the Kernal
> >that clears color ram to white no matter what cursor color you had:
> >https://www.atarimagazines.com/compute/issue50/286_1_Commodore_64_ROM_Generations.php
> >
> >But that can't be true.
> 
> The -01 KERNAL definitely did that. I think that they accidentally 
> copied it from the Vic-20 KERNAL, which always clears the screen
> memory with color 1 (white). On the Vic-20, the screen is white with
> cyan frame, and the text is dark blue (color 6). On the C64, the
> screen is dark blue with light blue background, and the text is light
> blue (color 14).

Or the original plan were to use the VIC 20 colors, just like for
example on the SX 64 and on the default on the Commodore made test
cart, but then at a late stage someone decided that the C64 should have
the same colors as the Atari 8-bit computers (which might had been true
for the early NTSC versions but surely not for the early PAL versions
which did more look like some kind of bluetoned magenta or lilac color).

> >I'm 100% sure that I were disappointed when the Swedish C64's were 
> >changed to Swedish characters and Swedish keyboard layout, and the 
> >color memory started to be cleared with the cursor color.

Opps, did write that thing wrong. The screen were cleared with the
background color. And the previous early PAL version (this was in like
mid-late -83) did clear the color ram to white.

> That would be matching the -03 KERNAL behaviour. Did the Swedish C64 
> KERNAL change anything else than the maps for the keyboard layout? I 
> think that the upgrade consisted of keycaps (or stickers for them),
> and KERNAL and character generator, probably on EPROM.

Afaik there were only the keyboard layout and the char rom that did
change. As I remember it, whatever kind of chips used did fit directly
in the sockets and didn't use any adapter. The key caps were available
both as stickers and as full key caps, and which version you got were
rather random. This was when Handic/Datatronic were the importer. I've
seen pictures of C64s from the time Commodore had their own company in
Sweden (for the whole of Scandinavia) and it seems like they then did
use 2764 eproms with an ugly kind of adapter that were just soldered
onto the eprom and then the whole thing were soldered on to the
motherboard. (This was of course for the Kernal. Afaik a 2732 fits
directly as a char rom). Btw the swedish layout did omit some petscii
graphics characters, that must had been some kind of mistake.

> The -02 KERNAL would clear the screen with the current background
> colour (53281) instead of using a constant. The -03 changed that to
> the current text colour (646).
> 
> I documented the differences between the Commodore 64 firmware
> versions that were known to me back in 2003:
> http://www.zimmers.net/anonftp/pub/cbm/firmware/computers/c64/revisions.txt
> 
> >I'm 100% sure that the C64's I've used back in the days were PAL
> >machines, as TV's that can handle both NTSC and PAL were super rare
> >in the 80's, especially in the early 80's.
> 
> Starting with the -02 KERNAL, PAL/NTSC is detected and the timer 
> interrupt is programmed accordingly. The -01 KERNAL was apparently
> for NTSC only, and some timing tables were computed for 1 MHz clock 
> frequency, instead of the correct frequency of 14318181/14 Hz or 
> 17734472/18 Hz.

Are we 100% sure that there were never any "PAL only" Kernal or that
PAL C64s were never shipped with the "NTSC only" Kernal?

There are "silver label" C64s which are PAL, so PAL were available
early on.

> Maybe the original plan was to use dotclock = 8MHz and divide that by
> 8 to generate the bus clock frequency. The 6567R56A has 64 cycles per 
> line, while the 6567R8 and later have 65. While 1MHz/64 would be the 
> correct line frequency for most PAL variants, NTSC-M uses a higher
> line frequency of 15734 Hz (which matches 14318181/14/65).

Or they just forgot to do it right from the beginning? :)

At the time they surely didn't have automatic software that handled
dependencies like that, so those who wrote the original Kernal had no
way to automatically flag that the Kernal needed a change if the
frequencies would change.

-- 
(\_/) Copy the bunny to your mails to help
(O.o) him achieve world domination.
(> <) Come join the dark side.
/_|_\ We have cookies.