HOW TO FOOL A REPAIR TECH
               or you can't always trust your test cart.  9-6-2017

     The C64 was a PAL version which I don't often see on this side of the big
pond. It came in with a blank screen, RAM chips running very hot and its input
current nearly 3 Amps! A working board usually pulls less than one Amp. Normally
when I encounter a board damaged by a failed power supply, I swap it out rather
than trying to fix it. In this case, I didn't have another PAL board, so my only
option was to repair it. A PAL computer will run on an NTSC monitor but there
will be no color and the image appears to flicker since the screen refresh rate
is lower. I don't have a PAL monitor but this is acceptable for diagnostic
purposes. The board was a 250425.
     After changing two chips that were getting hot, I still had a blank screen
and the input current of the 5VDC source was still about 2 Amps. Rather than
trying to diagnose RAM further, I decided to swap them all out. I cut the pins
of each one near the IC then extracted the stubs and cleaned the holes. This
board now has eight new RAM chips in sockets. The boot screen then came up
normally but of course there was no color since I was running it on an NTSC
monitor. I thought I was done.
     I ran it up on my diagnostic cart and harness and to my surprise, half the
RAM tested bad along with some other failures. What? The C64 automatically tests
its RAM at boot and it was 38911 as it should be, so why was the cart test
throwing errors like that? My Dead Test cart didn't blink but did indicate two
bad RAM, and it stopped when it got to the color RAM chip. OK. I then ran the
board up on a video to HDMI convertor that accepts PAL and got a normal color
boot screen on that monitor. Back on the test cart, changing some chips like the
Kernal (-o2 subbed for -03) or the CIA IC's made the screen blank out when
running the cart. Different PLA substitutes did the same thing. Timing problems?
What was going on here?
     At this point, still trying to determine where the problem was, I resorted
to what was known in the business as the "shotgun" diagnostic approach, replacing
chips to find the bad one... more sockets and time wasted. After nearly all the
chips were replaced with the same symptoms apparent, I went back to the test cart
that indicated a bad color RAM chip and that was it! Installing that IC in my
NTSC test board showed blinking color characters, so I knew that IC was bad. Why
it should corrupt the data lines and produce RAM errors with the test cart is a
mystery. Why it worked so well in a PAL computer, likewise. The user is happy to
have his computer back and I'm glad to be rid of it but pleased I learned
something new. My stumbling block was my own unfamiliarity with a PAL C64, so I
missed some obvious clues along the way. I'll probably never see this fault
again but I'll be paying closer attention in the future... hopefully.

Ray