*********** REPLY SEPARATOR *********** On 6/23/2003 at 6:41 PM Marcel Kilgus wrote:
>ZN wrote: >>>For the ROM slot? IIRC it only has 16 address lines, i.e. yes, it can >>>only have 16kb. >> That would be 14 address lines, addressing 16384 bytes i.e. 16k. >> Marcel, you are getting old ;-) > >True ;-) >Ah well, somehow I calculated $C000 + $4000 = $10000 = 16 address >lines. Haven't played with all the ROM stuff for almost a decade >now, though. So how is it in reality? 64k beginning at $C000 or >how is the port mapped? > >Marcel Well, there are actually all 16 lines, and a ROM select line (high). It's just that the 14 lower bits are fed directly to the EPROM in a standard cartridge. A third of a 74LS10 (tripple three input NAND gate) is used to provide an active low chip select by detecting A14, A15 and ROM select high. SInce ROM select is high whenever the first 64k (*) is accessed, this maps the EPROM into $C000 to $FFFF. This also privides a clue to moving the complete ROM area onto the ROM slot. In this case, a 64k (27(C)512) EPROM can be used to hold 48k of the original ROM plus an image of whatever is to go into the ROM slot (normally TK2, I suppose). All 16 address lines are then routed to the EPROM chip. The ROM select is only inverted, as EPROMs have an active low chip select input. The original ROMs MUST be removed. The big omission on the ROM slot is a 'write' signal. Clever hardware can be used to 'remember' the state of the lower 8 address lines when a read is done from an area (obviously at least 256 bytes in size) set aside - the read data is normally ignored, as the purpose was only to capture the state of the address lines and later use them as data - then this data is written somewhere by the hardware. The most notable example of this is Romdisq, although other hardware used the same trick (there was a Kempston printer interface, I think, that used this, as well as the original Miracle hard disk).
