EVEN FOR A TECH, THE JOB ISN'T ALWAYS EASY     9-11-2014
										
     Here are some case notes from an actual repair job. One C64 pulled off the 
shelf had an unknown history. It wouldn't boot (blank screen) and so was stored 
away for when I had time to dig into it. The only obvious problems it had were
a bad power switch (sometimes didn't do anything) and a blank screen at bootup.
     The board was a common 250407. I replaced the switch and tried my DEAD TEST 
cart on it... nothing. I have a meter on my power supply to indicate how much 
current the board is drawing, but this one was within the normal range for a C64, 
namely 700 to 900mA. I felt all the chips to see if any were getting hot (shorted 
chips will often get very hot quickly). Again, no evidence of a problem there. 
The only socketed IC was the VIC, so I swapped it into a good board to verify it. 
I always try the easy stuff first, even if the symptoms lead me elsewhere. 
     The next step was to narrow the search, so I thought of which IC's fail 
most often and I replaced the PLA. Flicking the power switch this time produced
a "garbage" screen... random characters all over. That usually indicates a RAM 
problem so I plugged in the DEAD TEST cart again, and this time it flashed 8 
times. I pulled the respective chip and installed a socket. Re-running the test 
with a replacement RAM produced another series of flashes, this time different. 
I did that until all eight RAM chips were replaced. Still, the cart flashed bad 
RAM! Powering up the computer without the cart would sometimes produce a garbage 
screen, so I checked the address and data lines but they all had pulses on them. 
What now? I made sure the PLA I replaced was indeed bad by testing it in another 
board. It was dead.
     I noticed the pattern of the "garbage" on the screen was not totally random 
but appeared as a series of horizontal bars of repeating characters. That pointed 
me to the two RAM interface chips U13 and U25. They were both bad! Replacing one 
gave me an opening screen but with less than 38911 bytes free, and the other swap 
gave me a normal opening screen. Now it was time to test the RAM chips I pulled 
to see if any were actually bad. Nope. They all worked fine when re-installed. 
So, the DEAD TEST cart is not exactly foolproof (and neither am I) and this board
now has ten more sockets. As to why both RAM interface chips failed, I can only 
guess but they were both MOS 7708 chips and those Commodore MOS "clones" of 
standard TTL chips do seem to have a higher than normal failure rate. The standard 
TTL version is a 74LS257AN or 74LS257APC. Note that a 74LS257PC will not work.! 
There is something about that version of the IC that just doesn't work correctly 
in a C64 or SX64. 
     So there you have it... another working board, but it took about three hours 
and ten sockets to get there, and eight were not necessary. Not all repair jobs 
are this convoluted but the "blank screen" symptom is always a challenge. When I 
encounter more than one bad IC, I always suspect a failing power supply, but in 
this case, multiple chip failures was just bad luck. If I had found several bad 
RAM chips, that does indeed point to a bad PS because those TTL IC's are 
particularly vulnerable to higher than normal supply voltage. The black brick C64 
supply can and does do that sometimes. Even if load tested to make sure its normal 
regulated +5VDC output stays at 5 volts and doesn't creep up after it warms up,
I don't trust them at all now!
     I should mention something else here... the very next dead C64 that I worked 
on was just as crazy. After replacing the bad PLA, I had a "garbage screen" that
was not a memory chip problem. It turned out to be one of the small logic chips 
(U27), a 74LS08. Again, the Commodore version of that IC (MOS7712) failed. I am 
now suspicious of all MOS replacements for small logic chips. 

Ray Carlsen
CARLSEN ELECTRONICS... a leader in trailing-edge technology