[N8VEM: 17279] SBC-v2 hangup issue resolved: RESET signal going to 82C55 and 16C550

[r***@public.gmane.org]

Hi all,

First i'd like to thank Andrew Lynch and all contributing developers for this exciting retro-project and all their energy they put into it.

When I saw the n8vem project page, I decided that I needed the full retro blast and wire-wrapped my SBC-v2 on perf-board (leaving out the ECB-Bus part and RTC).

Thanks to Dan Werners excellent firmware package, I got my SBC-v2 copy up and running in less than a week. Even the SD-Card support via 82C55 PPI worked from scratch.

However I experienced during Xmodem file transfers (to RAM-Disk and SD-Card) some reproducible hang-ups.
I validated with the logic-analyzer that all my address-selection logic was working correctly. During the hang-up, the CPU seemed still to be running and addressing the 16C550.

After some longer head-scratching it was obvious that on my wire-wrap build, the 16C550 and 82C55 influenced each other mutually via the RESET signal. I employed the same logic as the original SBC-v2 schematic (/RESET and RESET generated via 74LS14 gates).

The hangup during Xmodem transfer was due to the 16C550 being reset in the middle of the transfer and having no correct initialization after that.
I solved the problem by giving each chip (16C550 and 82C55) it's own 74LS14 gate on their RESET pin.
I suspect that my chinese made Wingshing WS82C55AG-5P chip might have been contributing to the problem. The 16C550 is a TL16C550CN from TI.

As I've seen other people on this newsgroup going though several manufacturers of 82C55 before finding a stable chip,
my suggestion would be to give a dedicated 74LS14 gate to each 16C550 and 82C55 and check if this helps.
I agree that on my wire-wrap build this was easier to do than on a PCB.

Kind regards,

Yves

[Borut]

Yves,

You might be also experiencing glitches in power supply. When doing the
wire wrapping,
i learned to take extra care that my power and ground lines don't 'daisy
chain' too much and to run power lines from
a single point. You also must have adequate capacitors on power supply
lines close to the vcc and gnd pins of each individual chip,
and at the central point.

lp, Bo/

Post by r***@public.gmane.org
Hi all,
First i'd like to thank Andrew Lynch and all contributing developers for
this exciting retro-project and all their energy they put into it.
When I saw the n8vem project page, I decided that I needed the full retro
blast and wire-wrapped my SBC-v2 on perf-board (leaving out the ECB-Bus
part and RTC).
Thanks to Dan Werners excellent firmware package, I got my SBC-v2 copy up
and running in less than a week. Even the SD-Card support via 82C55 PPI
worked from scratch.
However I experienced during Xmodem file transfers (to RAM-Disk and
SD-Card) some reproducible hang-ups.
I validated with the logic-analyzer that all my address-selection logic
was working correctly. During the hang-up, the CPU seemed still to be
running and addressing the 16C550.
After some longer head-scratching it was obvious that on my wire-wrap
build, the 16C550 and 82C55 influenced each other mutually via the RESET
signal. I employed the same logic as the original SBC-v2 schematic (/RESET
and RESET generated via 74LS14 gates).
The hangup during Xmodem transfer was due to the 16C550 being reset in the
middle of the transfer and having no correct initialization after that.
I solved the problem by giving each chip (16C550 and 82C55) it's own
74LS14 gate on their RESET pin.
I suspect that my chinese made Wingshing WS82C55AG-5P chip might have been
contributing to the problem. The 16C550 is a TL16C550CN from TI.
As I've seen other people on this newsgroup going though several
manufacturers of 82C55 before finding a stable chip,
my suggestion would be to give a dedicated 74LS14 gate to each 16C550 and
82C55 and check if this helps.
I agree that on my wire-wrap build this was easier to do than on a PCB.
Kind regards,
Yves