Re: Software for MS-DOS 1.25

Den Sun, 5 Nov 2017 22:19:58 +0100 skrev Rhialto <rhialto@falu.nl>:
> On Sun 05 Nov 2017 at 18:41:10 +0100, Micha? Pleban wrote:
> > smf wrote:
> > 
> > > I swapped your font with the standard one in vice and it looks
> > > very IBM'y. I don't know if vice supports the 2nd font bank, but
> > > it would be nice to get that added if it doesn't.
> > 
> > It appears that it doesn't. On a real 710, pressing Esc,Z will
> > change to the second font, and Esc,Y will revert back to normal
> > font. These don't seem to work in VICE and it would be nice to add
> > them.
> 
> There should be at least partial support in VICE since it works for
> the SuperPET. The SuperPET has APL characters in the second character
> set, and it can be selected via bit 5 of the CRTC_REG_DISPSTARTH
> register (the one that sets the start of video ram address). It is
> kind of misused by Commodore: the higher address bits do other things.
> 
>         case CRTC_REG_DISPSTARTH:   /* R12  Control register */
>             /* This is actually the upper 6 video RAM address bits.
>              * But CBM decided that the two uppermost bits should be
> used
>              * for control.
>              * The usage here is from the 8032 schematics on funet.
>              *
>              * Bit 0: 1=add 256 to screen start address ( 512 for
> 80-columns)
>              * Bit 1: 1=add 512 to screen start address (1024 for
> 80-columns)
>              * Bit 2: no connection
>              * Bit 3: no connection
>              * Bit 4: invert video signal
>              * Bit 5: use top half of 4K character generator
>              * Bit 6: (no pin on the CRTC, video address is 14 bit
> only)
>              * Bit 7: (no pin on the CRTC, video address is 14 bit
> only) */
>             /* FIXME: check if the above also applies to CBM2 and all
> the other PET models */

According to the 8296 schematics, it seems like 13 of the 14 bits is
connected to the RAM muxes UA9/UB9, but the 13th line (MA12) can be
disabled with a jumper. Also a signal for reading two chars in a row,
from the PLA, is fed to the RAM muxes, and also RA4 (the signal that
goes high when displaying the 17th-32nd line of really tall chars).

On the other hand there is also a signal called PULLUP fed to the muxes,
and that signal seems to just be pulled up by a resistor, also feeding
CS0 on UC13 (6520) and CS0 on UR2 (6520). That's probably to force
screen ram being on the upper 32k half of the 64k adress space. Don't
know why all theese signals seems to be fed by 5V through a common
resistor, maybe it has something to do with how the NMOS
and TTL semiconductors work?

If I understand this correctly it would be semi-easy to make a hardware
patch for 8296 making it possible to display a 40 char screen at $400.
You all know what that could be used for :) (It might also be possible
to program the CRTC to show a 22 char display at $1E00 or $1000... :) ).


Anyhow MA13 from the CRTC on a 8296 is fed to pin 18, A11/_CS2, on the
char rom which can be either 2316 or 2332. Pin 19, A10, on the char rom
is fed from CA2 on the 6522 and that signal is called GRAPHIC so this
is what selects between the upper case + graphics or lower + upper case
char modes.

Every time I look into anything memory related on 8296, I get a feeling
that they didn't have the time to implement all that the hardware could
be made to do with some more logic :)

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

       Message was sent through the cbm-hackers mailing list