Date   : Wed, 14 Jul 1993 13:10:13 +1200
From   : David Andrew Sainty <David.Sainty@...>
Subject: Re: Tube connections (was RE: colour Graphics) 

Date: Tue, 13 Jul 93 12:50:06 +0200
From: stjohn@...

>>So, this seems to indicate only the low 8 locations are required by TUBE
>>hardware. Does the splitter ensure splitter switching writes don't make it
>>to the second processor? (What I'm really asking is, if a write is make
>>to a second processor above those bottom 8 locations, does it ignore it,
>>or might it panic?)
> 
> Pandora won't. I've disassembled all the importants parts of it so
> I can guarantee that in general it won't panic. If you want to make
> extra sure, see the new AUG which details the Tube I/O and protocol.

Hmmm, but does the tube ULA mask out the illegal address?

> Be careful with NMI because Pandora at least uses the NMI as a 
> sort of software DMA. After a disk I/O the beeb sends NMIs to the
> 32016 with data in the Tube ULA. This is how data is transfered
> transparently into the second processor. If you start fiddling with
> them, then expect some BIG problems!

Hmmm, is that a function across the tube? I have a feeling that is a
function on the 2nd processors onboard tube ULA, not involving the
NMI line on the tube connector....

> PS: On my board there's a big 16MHz crystal hooked up to a NAND
> gate chip... why don't you try following where the output of that
> clock goes and trapping it somewhere on the board instead of messing
> with the Tube? Knowing Acorn, there's probably a pin somewhere on 
> the board that lets you plug straight in to the system clock ;-)

Now you've hurt my feelings! :-)

In fact, the best line to take would be the character clock straight off the
6845. What I'm trying to achieve here is a solution that doesn't require
nasty connections inside the machine. I figured a plug-in solution was
far more preferable....?