Date   : Wed, 29 Mar 2006 20:52:35 +0100
From   : Fragula <fragula@...>
Subject: Re: Model B NMI PITA.

Pete Turnbull wrote:

> That's not right.  I've got Beebs here that have both; it's always been
> possible to have both.  In fact, Level 1 and Level 2 file servers have
> to have both, as they're floppy-based.
Indeed. But *my* Good Old Beebs don't want to play.
I have noticed one thing though.. ICs 79 and 80 are 7438 on the map at
the back of the AUG, but my board appears to have 7437s, which are totem
 pole outputs, not open collector. However these are buffers on the disk
interface, as far as i can see, and that's working just fine.

> 
> It's possible you have an early PCB that needs some fix to use the disk
> interface properly, though offhand I can't remember what the fix would
> be (something to do with that 7438 at IC27, and S9).
Well the problematic ones, One really, the one I've owned for 20 odd
years, is an Issue 7, but its been home upgraded (everything + kitchen
sink) and has been problematic ever since fitting the Econet upgrade.
The econet has never worked right, though the disk system is working at
the mo (or was till i ripped it all off earlier this evening).

> It's also
> possible that the interrupt on the STL interface works differently from
> Acorn's, but that's less likely.
The DFDC Mk1. was different, though *probably* the IRQ stuff was the
same. The Mk2, which is what I have here, is register/software etc.
compatible, and simply plugs into the 8271 socket. Oh there is one
flying wire, but thats to allow it to bypass some drive I/O stuff that
is built into the 1770.

> The FDC interrupt is active-high and comes from pin 11 of the 8271, to
> pin 9 of a 7438, which (pin 8) drives the /NMI line.  Pin 10 of this
> 7438 is tied to +5V.  Pin 9 is connected to S9, which shorts this to
> ground, thereby preventing this gate from going low, when the FDC is
> not fitted.
i've just traced all that. (/me has eye socket bruising from squinting
on a loupe all evening) Seems to be o.k.

> 
> S2 does a similar job for the interrupt from the 68B54 ADLC, although
> that is active-low, and passed from the ADLC pin 7 through part of a
> 74LS132 (IC91 pin 13) coming out on pin 11, then into pin 2 of the same
> 7438 used for the FDC interrupt, coming out onto the /NMI line on pin
> 3.  However, pin 1 of that part of the 7438 is fed from a flip-flop
> that can be used to turn the Econet interrupts on and off.
That's something I need to look at, that flip-flop.. I think a little
substitution might be in order.. cheating i know, but.. ;->

Cheers!

M.