[Coco] CocoNIC

[RETRO Innovations]

On 8/30/2016 9:10 AM, John W. Linville wrote:
> If you have it in you to make two, I would still be interested in
> having one. I plan to be at VCFMW as well.
>
> John
Good news.  To the extent that Brett's test apps can test the 2 units I 
have assembled, they both work

Success was delayed as the initial testing seemed to work halfway. 
Brett's code to read the device ID was:

$ff4a>0
$ff4b>0
read $ff4c (should have decimal 14).

Which worked.  Then, the next test:

$ff4a>$14
$ff4b>1
$ff4c>$f0 <--- should turn on LEDs

Did not.

I spent days trying to diagnose the issue.  I verified Brett's code was 
correct by performing the same statements on 64NIC+, which indeed turned 
the light on.

I then compared the output of:

fort=0to15:?peek($ff4a+0):next

on both units.

I found the values differed in location, not in value.  A value at 
position 7 on the 64 showed up at position 11 on the CocoNIC.

Looking at the addresses:

1011
0111

Showed the issue.  A3 and A2 were swapped. I checked my wiring in the 
PCB and the CPLD and could not find the issue, but I swapped the lines 
and it started working.

This morning, I had a hunch and checked.

In the design, to make layout easier, I used a CPLD and I pass A3..A0 
through the CPLD from the Coco BUS to the NIC IC.  I checked the NIC 
addresses compared to the CPLD pin assignments, and they were correct.

I know, you're already at the answer....

This morning, I checked the pin assignments of the Coco bus to the 
CPLD.  A3 and A2 were crossed.

I was surprised the issue did not show up anywhere else, but the current 
HDL in the CPLD does not look at the lower address lines at all.  It 
simply passes them onto the NIC.

I have already packed up for the show, so I will worked to unswap and 
swap the correct lines when I get to the show Friday night. THen, it's 
all up to others to make the code come alive.

Jim

-- 
RETRO Innovations, Contemporary Gear for Classic Systems
www.go4retro.com
store.go4retro.com