I've been having some fun with a home brew parallel i/o card for 
Apple2. It's a modified parallel printer card with a flat cable that 
plugs into an IBM-PC parallel port allowing the two computers to 
exchange data directly. 
Using Borland C I wrote some preliminary software for the IBM to 
configure it as a file server for ProDos 8 but just as easily it 
could directly download programs, operating systems such as DOS 3.3, 
etc. 
So far I've been using a MacroprintPlus card (by Macrotech Computer 
Products of BC, Canada) as it was designed for parallel i/o as well 
as printing. 
However, the Grappler+ is much more widely available, especially the 
clone versions. Some time ago I posted instructions on how to modify 
a Grappler for this function and I've thought up an extension of the 
concept to make it more versatile by allowing the on-board firmware 
to be writable at run time. This is particularly handy for me because 
I use a 6502 cross-assembler on the IBM. 
Basically the idea involves using both RAM and EPROM. The operating 
system, program and firmware would all be downloadable from the IBM 
which has much more storage resources than the usual II+ or //e 
system with 143k floppy drives. The ROM would contain only minimal 
boot code, used to download whatever firmware is needed at the 
moment. 
It would be handy if the RAM were non-volatile, and Dallas 
Semiconductor makes a battery-backed 2k x 8 RAM, the DS1220. 
Modifying the Grappler+ 
----------------------- 
This card has the following hardware features: 
* socket for 2732 EPROM (more on this below) 
* 8-bit latched output port 
* output strobe (active low) 
* 8-bit input port, currently allocated as: 
     bit 0:   printer ACK- 
     bit 1-3: printer status 
     bit 4-6: DIP switches 
     bit 7:   interrupt pending 
* optional interrupt (IRQ-) generated by printer ACK- 
The provision for a 2732 (4k x 8) EPROM is the basis for this scheme 
because this chip is replaced by a 2716 (2k x 8) EPROM and a 4116 
static RAM of the same size. The high order address line, A11 
normally used to select the Grappler's firmware banks becomes the 
device select line; low for EPROM and high for RAM. 
The physical placement of the two chips is a minor consideration; The 
RAM can piggybacked on the ROM, pin for pin except CE- or it can be 
fitted somehow onto the solder side of the card. In my computer this 
card is in slot 7 so there's no problem with clearance room on the 
component side. 
A slight modification is needed to make the RAM writable. I tried 
this tonight with a 4116 in the firmware socket and it seems to work: 
1. Since this is RAM, it needs a WR- signal to accept data. I cut the 
trace to U9-21 (was A11 on the 2732 but now the WR- input for a 
4116.) and connected a jumper wire from this pin to U3-3, ie. the 
RD/WR- line. 
2. I cut the short trace from U9 pin 20 (OE-) that connects it to the 
line from U6-11 to U9-18. This pin needs to be connected to the 
invert of RD/WR-, ie. RD-. For my present purposes I don't need 
interrupts from this card, so the resources normally used for 
interrupts become available including the latch at U2A, etc., 
and RD- was derived from the NAND gate at U6A. 
When I tried this with a card in slot 1 I could write and read 
to $C100 - $C1FF with no problem. However, $C800 to $C8FF seems 
to be mapped into this same space, so the expansion ROM space 
really starts at $C900. 
The Grappler+ soft switches are somewhat peculiar. According to the 
manual the nominal designations for an i/o write are as follows 
$C080,Y    output 
$C081,Y    Select ROM bank 2 
$C082,Y    reset interrupt Req and IRQ data bit 
$C084,Y    output/interrupt on ACK 
What actually happens is this: 
1. If A0 is high then ROM bank 2 is selected. (From IC U1C). Note: 
ROM bank 1 is selected on any access to the peripheral firmware 
(ie $C700 - C7FF for slot 7, etc.) 
2. If A0 and A1 are both low then the data bus is written to the 
output latch and an output STB- is initiated. Also, the ACK 
latch at U2C is reset. 
3. If A1 is high then the interrupt latch is reset. (U1B) 
4. If A2 is high then the interrupt latch is set. (U1A 
Point 2. above is why the card also performs output on $C084,Y. 
Anyway, after all this messing about I'll be able to program an 2716 
just once with some minimal interface code and from then on 
everything else will be downloadable from the IBM in a convenient 
way. So when I show my friends a highly modifed board crawling with 
jumper wires and cables leading here and there and if anyone asks me 
what's the point, I'll just say -- hacking, that's what the Apple 2 
was all about (originally) -- and despite mega this and giga that on 
the latest Pentiums, the spirit lives on.