🚀 go-pugleaf

In article <40712@sdcc12.ucsd.edu> vwass@sdcc13.ucsd.edu (Vincent Wass) writes:
> I am having trouble with a computer interface project.  I'm trying
> to decode addresses on the bus for my own plug-in wire-wrap card.
> So I used a 74ls688 compare chip to
> decode address lines A2 to A9, and then I wired A0 and A1 to an 8255
> chip.  I've wired the address to be 300hex.  I've also tried 280hex,
> but I had the same problem [, disk-related errors].
 
Here are some things to watch out for:

a) keep all wires to the card edge fingers (ie the bus) *really* short.
   The best way to ensure this is to place the sockets for the bus
   buffers right along the closest row of holes to the fingers.
   Do not exceed two inches.  Really.

b) buffer *all* signals with a single LS TTL load.  The IBM "spec" says two
   LS loads, but some buses are longer and some cards cheat.  A fully loaded
   system may become overloaded when you add your board.

   This also means you should buffer the lines going to the 8255; stick
   in an LS244 and ground the /enable (always on) and buffer A0, A1,
   /RD, /WR, etc.  A MOS device like the 8255 represents a capacitive load
   that slows the risetime of those bus signals.

c) use a utility to map out what i/o devices are in your system.  0x300
   and 0x280 are likely OK, but some older PCs and XTs used 0x300 for the
   time-of-day clock.

Have fun!

--
"The path of my life is strewn with cowpats from the devils' own herd."
                                          -- Edmund Blackadder II
bmw@isgtec.com   [ ...!uunet.ca!isgtec!bmw ]  Bruce Walker