Hi,
A little more news on the Ace repair front - I've pulled up the LS367s
and found a load of tracks that were ripped up during resoldering - about
half of these have now been repaired. It looks like the surge went straight
through data line D3 - I've put the ROMs in my EPROM programmer and checked
them against the ROM images Lee Davison sent me and they don't match at all.
In every case, the output value is off by 0x04 - meaning data line D2 is
probably cooked...
I'm also beginning to suspect the video shift register - something isn't
right. Address decoding is fine, the RAMs should have survived being
de-soldered and re-soldered (IME they usually do) and I can't find any
problems with the video timing chain (I assume the fact that I'm getting a
video signal on my TV is good enough proof of this).
It looks like the PCB was laid out with the main CPU section on the left
and the video section on the right with a set of LS367s splitting them up.
This is not as cheap a design as the ZX81 (IMO), but the PCB is complete and
utter crap - the tin plating is stripping off all over the place, solder
resist is cracking and has got thin, raised strips running along its length,
kinda like the veins on a tree leaf.
I've bodged a pair of 2732 ROMs I had in my junk box to fit the sockets
and pinout. The video is reasonably stable, but it's mostly white. I haven't
got any idea why the font RAM and video RAM aren't being loaded, though. Now
if only Farnell still stocked 2532s and 2114s... Thankfully the buffers seem
to have isolated a lot of circuitry in the video circuitry, especially the
timing chain.
The only possible fault condition I can think of is that Jupiter Cantab
designed the Ace around National Semiconductor/Motorola/TI 74LS logic and
the newer Fairchild Semi logic devices I'm using are incompatible with the
existing logic... Stranger and stranger...
Later.
--
Phil.
philpem_at_dsl.pipex.com
http://www.philpem.dsl.pipex.com/
Received on Mon Dec 30 2002 - 03:22:00 GMT