Re: [Fwd: Diagnostic cartridges]

From: Nicolas Welte (
Date: 2002-01-31 22:13:46

Hi Nick, wrote:
> I used the SuperGames cart as it takes 27xxx EPROMS and had the 74175 latch
> (and have spares). The upper ROM only has RESET/IRQ/NMI vectors, all
> directed to $9FE0
> The patch at $9FE0 (in the low ROM), just pokes the latch to change memory
> config, then JMP(8000).

I built a hardware on my own now, which uses the concept I described in my
previous message. I based this on a dual 8k EXROM/ROML board for a 27128
EPROM, with a latch made from a 74LS00 to switch off the EPROM by a write to
I/O2. I rewired the LS00 to do what I needed: two NAND gates form the
RS-flipflop (/RES asserts /GAME, any access to I/O2 asserts /EXROM), the other
two gates are used to form a combined chip select from /ROMH and /ROML. This
way I could put all your code into a single 8k bank and therefore install two
different diagnostic carts in one EPROM. I'll expand this to the third
cartridge from 64'er magazine soon, and maybe install a (yet to be written)
more robust zeropage and stack RAM test in the fourth bank of a 27256.

This is one thing I observed now: While the two diagnostic programs start up
in any case now, it is impossible to diagnose every RAM failure with them. A
missing or open circuit RAM chip messes up the screen so much that it's
impossible to read the diagnostic message :-/ The program from GO64 magazine
also reports a RAM failure in my C64 where there is none, and the
Commodore/Handic program reports half of the RAM chips as bad if only one of
them has a defect in the zeropage (which in turn is detected okay by the GO64
program). I decided to add the 64'er mag program to the cartridge now, too,
and it's even worse than the other two. With a missing RAM chip or a damage in
the zeropage, it doesn't even start up but stays there with a black screen.

I'm thinking about adding an extra 8bit latch to the cartridge that can be
written through I/O1 and then drives eight LEDs that indicate the status of
each RAM chip. This way even without any screen output RAM failures can be
detected. The only problem is to write a ($0000-$01ff) RAM test that does not
rely on a single byte of RAM. 

> I also added LEDs (via a 7404) for some additional diagnostics: RED for +5V,
> YELLOW for RESET high (released), GREEN for ROML (verify that cart is being
> started/accessed, PLA works), dual colour for Phi2 CLOCK (so as to verify it
> actually toggles).

I have an extra LED board for this, it decodes the adress lines to 4k blocks
with one LED each, and displays IRQ,RES,NMI, GAME, EXROM.


       Message was sent through the cbm-hackers mailing list

Archive generated by hypermail 2.1.1.