Re: Strange 8255 behavior

From: Jim Brain <brain_at_jbrain.com>
Date: Fri, 15 Jun 2018 22:59:53 -0500
Message-ID: <e78be76e-d260-ef62-2251-f5f705fb7382@jbrain.com>
On 6/15/2018 10:46 PM, Mia Magnusson wrote:
> There must be some reason for a VIC 20 to have a transceiver between
> VIC+RAM and CPU+ROM+I/O+cartridge port.
I think it's because on a write cycle, the lines go to output in the 
early part of the cycle, which would mess up teh VIC-I.

THis issue bit me on the 6809 project, so it's somewhat relevant to 
current discussions.

I decided to use the TSC line (tri state control) to move the 6809 off 
the bus, so I could share the memory with the main CPU (coprocessor 
card).  What I found is that if TSC is high (which holds the address and 
data lines in tristate) up to the rise of E clock, the CPU would HALT 
for the cycle.  And, due to the setup needs of the 63c09, you really had 
to move TSC low about 80nS before E rise or the condition persisted.

IN the prototype, I got around the issue by letting the 63c09 have the 
bus for all of E | Q, which then meant I needed the latch main CPU reads 
from shared memory on the first half of the main CPU's clock cycle (when 
63c09 E was low and 63c09's Q was low) and then send onto the main cpu 
during the E=1,Q=1 quarter cycle.

Thus, for the newer design, I gave up on TSC use, and put 541's and a 
245 on the address and data lines.  That way, I can slide the cpu off 
the bus for the E=low half cycle, allowing the main CPU access to the 
memory, and not have to worry about internal TSC latching issues.

Jim

>

-- 
Jim Brain
brain@jbrain.com
www.jbrain.com
Received on 2018-06-16 07:00:04

Archive generated by hypermail 2.2.0.