Re: Address bus visualization with LEDs

On Fri, Jan 22, 2016 at 03:19:11PM +0000, Steve Gray wrote:
>Hmm, I wonder if a singe line of 64 LEDs would work better... one for 
>each K

The 4 by 16 display was wired as 8 by 8, if I remember correctly, so it 
used two 3-to-8 decoders for driving the rows and columns.

Likewise, my 8 by 32 display was wired as 16 by 16 (two adjacent 
physical columns were one logical column). The solder side of the board 
looked pretty funny. Too bad I did not take a picture before I gave it 
to Ruud.

I would say that a matrix layout is definitely easier than a linear 
layout. With a matrix, you can easily tell if something happens at $7fxx 
or at $80xx, for example, because it would be a completely different row 
(or column), even though it is an adjacent column (or row).

>then maybe you could add a small capacitor to each LED to make them 
>stay on a little longer? That will mean more decoder chips but might 
>give a better result.

Like I wrote, I think that modern super-bright LEDs are simpler. And the 
capacitor would ruin the reliability of the instrument, if you wanted to 
use it for debugging or profiling.

BTW, back then, Andreas Boose wrote some clever code generator that 
would show a "Knight Rider" pattern on his display. :) The trick was 
that you had to JMP around in the memory, to avoid stray memory accesses 
that would ruin the display. No stack (no subroutines, no interrupts), 
no writes to RAM while it is running. Just use the registers.

	Marko

       Message was sent through the cbm-hackers mailing list