Posted by saint on 19, 2000 at 6:06 PM:
In Reply to: yes i know... posted by dave on 18, 2000 at 11:49 AM:
From the Rasterlist mailing list (none of this is mine)...
----------------
As far as general repair guides go I will discuss a bench set up,
leaving cabinet wireing out of it. As always power, reseating and
reflowing connctors are assumed before we get into the chip level repairs.
Pre repair: Inspect the board. I mean really inspect the board.
Look at it for a while. I picked up a dead space dual board cheap, and on
pre-inspection found a small solder splat on the traces. Peeled it off
and the board worked!
here is what we are looking at:
1) missing parts, backward parts, corrosion on chips. Signs of
arching (had this on my blaster set)
2) damaged traces, long pins bent over shorting traces out.
3) Previous repairs. Typically these can break traces under the
socket or the chip that was fixed. I would like to find the guy who keeps
on using a curling iron as a soldering iron, cause I think I've repaired
may boards he has looked at.
As far as general repair guides go, I can only speak of williams and
PAC's since those are the only two I work on. When I get a dead board to
work on, I first hook it up and read the "Signs of life". Let's say the
board has no video output. Power is good, and roms are all checksummes
(Although I don't usually do this right off the bat, I tend to have a
known good set of roms to use, easy given I limit the style of boards I
repair)
1) Check to see that the CPU has a clock. If not, fix it.
2) I need some video output to give me hints, so I trace back from
sync and fix that if it is missing or bad.
3) Check and attempt to fix RGB signals, not always possible if the
ram is not being strobed properly.
Now I have a board giving me some video clues. I now want the self
test's to run, so I can replace any faulty ram.
4) To get the self test to run, I start looking around the CPU.
Clock signal must be present, IRQ must not be held in the wrong state for
that CPU, halt must be okay, and rest must not be toggling. Fixes for IRQ
and halt are generally simple, but reset (watchdog) can be more difficult.
It always helps to have a known good CPU chip as well. I like to keep a
spare working board of the style I repair so I can test chips in it, or
borrow known working ones from it.
5) If watchdog is Toggling, I trace it back, and 50% of the time, I
find a fault in the watchdog circuitry. 7474 or counter problems.
74LS161's are my nemisis in these cases. Bad sockets under the ROMs or
the ROMS's themselves can be an issue here.
6) Once watchdog has settled down, (not gone away) I can generally
see self test's run and point to bad ram. I try Known good ram, and if
the problems persist, I hit the write enables and any select, decoding
circuitry, noting that some elements are used to get ram data to the bus,
and that is important, and other circuits get it to the screen.
7) Fixed those up, and the game will generally run. Poor graphics
during attract means we have to look at the circuits responsible for
getting data out to the screen, cause the CPU is not wachdogging anymore,
so the bus data is fine. Sometimes this can be separate ram/roms,
sometimes it is shared.
8) then I look at CP inputs, to ensure thay are okay, and the last
tidbit is the sound. Sound is fairly unique between systems, so you are
on your own. The only hint there is to do the finger test. Put your
finger on the underside of the amplifier and if you get a hum, you have a
good amp. If not, look at the amp as the issue. This is one of the few
places a scope comes in handy, but sound circuites are simple enought that
it is not absolutley needed.
On the outside chance you are dealing with a bus problem
(tranceivers, buffers) you will have an extremely tough time, as well as
an intermitant trace. Sometimes I get lucky on this style of problem,
sometimes I just want to snap the board in half across my knee, or sell it
to Rick :-)
Thats all for now
Mike