On Wed, 8 Oct 1997, Jeff Kaneko wrote:
> Now 80 column mode dosen't work at all. Damn. 40 is ok, but it kinda
Hmmm... Now, I've never seen this card, but that suggests one of 2 things
to me - either bit-rot in the control EPROM (I hope you've dumped it...)
or video memory failure. If this thing uses 2114's for video memory,
replace the set. I've had more problems than enough with that excuse for a
RAM chip :-). I've had them fail in Tandy M3's, CBM 8050 drives, Vectexes
(Vectrices?), etc.
> defeats the purpose of having the display card: I can get 40 columns
> on a C= 64. Damn. I haven't been hanging around much in Commodore
> circles lately, but this is the only VideoPAK/20 I've ever seen. The
> only other Data-20 product I've ever seen was a VideoPAK/64 and that
> was in 1988! I don't even have a schematic for the thing.
Now's the time to learn the gentle art of reverse engineering...
Seriously tracing out schematics is not hard (which is why companies who
refuse to supply them 'as it makes it harder for our competitors to copy'
are being plain stupid). IMHO it is easier, quicker, and plain less hassle
to draw out any schematic of < 400 chips than to battle with the
manufacturer.
How to go about it :
0) Get a good continuity checker. Make sure it responds quickly (HP
suggest < 1ms...), and that it's not fooled by diodes, etc. Designs for
these have been published in electronics magazines. A beeper type of
output is very useful.
1) Look at the board. Make a list of all the IC's and get data on them.
Obviously there's no point in looking up all the TTL functions if you have
the TI bible on the shelf. But make sure you have all the others. Custom
chips are a pain, and I'll cover those later.
2) Now list all the IC's _and sections_ on the board in component-number
order. You'll end up with a list like :
IC1 a b c d 74LS00
IC2 a b c 74LS10
IC3 6845
IC4 a b 74F74
IC5 Z80A CPU
IC6 a b c d e f 74LS04
etc.
3) Start drawing. Start with the largest/most complicated IC - like a CPU.
There's essentially only one way to use a CPU, and once you've identified
it, you probably have the data and address buses found as well. A thing
like a NAND gate. Trace out, finding things like memory arrays, address
decoders, buffers, etc. As you draw each section of a chip, cross it off
the list. That way, you know what you've drawn, and you don't waste time
doing the same bit twice.
4) If there's no CPU, start from some other reference point. A bus
connector, perhaps. Or a large video chip. Or _anything_ that identifies a
lot of signals.
5) You'll end up with the 'glue' logic. You basically have to work through
this, which is a pain, but often you can identify well-known sections,
particularly if you've seen a lot of schematics. Experience counts for a
lot here.
Custom chips : These are a PAIN!. Often you can identify most of the pins
by what they're connected to - CPU bus signals, RAM bus signals, memory
select lines, etc. Make a pinout table, and fill it in.
Interconnected custom chips : Panic. A logic analyser may help you
distiguish data from enable from clock, but you will spend a long time
figuring it all out. And of course the device has to be working for you to
do this.
Which brings me on to my next point. The time to start thinking about
maintenance/repair is when the device is still working. That's when you
can take dumps of the ROMs (and know they're good), can record important
waveforms (and know they're correct), etc. After it's failed you have a
lot less to go on.
> Jeff
-tony
Received on Wed Oct 08 1997 - 12:46:52 BST
This archive was generated by hypermail 2.3.0
: Fri Oct 10 2014 - 23:30:36 BST