C128 & C128DCR 80 column picture distortion (or no picture) 02-22-99.

           C128 & C128DCR 80 column picture distortion (or no picture)

This problem and solution requires a rather lengthy explanation.
Through a series of lucky "accidents", a bit of intuition (plus user
input :-) and half a day of digging, I have found the cause of this
fault. It would only show up with the combination of a C128DCR and a
1084S monitor. Since I don't have a 1084S, I tried all the other
monitors I have with my three DCR's. 
     One of my DCR computers hooked up to my CBM 2002 (made for the
Amiga, just like the 1084S is) produced exactly the symptoms described!
At low background brightness levels, the picture was almost normal with
just a bit of horizontal bending in some of the characters. As the
background (or character) level was increased, the picture would start
tearing horizontally. Higher levels would blank out the picture
     Since only one of my 128DCR computers would produce the fault, I
decided to look at the signal waveforms going to the monitor on my
oscilloscope. The horizontal sync signal was reduced from a normal 3.2
volts peak to peak down to less than 1.5 volts, and it was contaminated
with "garbage" from the video. The vertical sync signal was normal. I
tried another RGB cable (one I had made up) and the distortion in H
sync went away. But if the other cable were faulty somehow, why did it
work with the other computers? It also worked fine with other monitors
and the "faulty" 128DCR. What was going on here?
     I pulled out the schematics and compared the RGB output circuits
of a C128 (that never showed a problem) with the 128DCR (only one of
which showed a problem). The circuits use the same "buffer" chip
between the RGB video controller chip and the monitor outputs. Could
the chip be marginal? I swapped it out and, although the waveform
looked only slightly better with the replacement chip, the monitor now
works just fine!
     Summery: This "problem" has many factors that worked together
perfectly to drive a technician (me) and a few Commodore users stark
raving mad. 1) Different monitors vary on how much of a "load" they put
on the signal and sync outputs of a computer. 2) Some monitors are more
sensitive than others about the "quality" of the synchronizing signals
they get. 3) A computer is designed to drive a "normal" load such as a
monitor within a certain set of specs. 4) Cables are not all created
equal!!! In this case, the cable I was using was a 6 foot
straight-through (pin for pin) extention with a 9 pin D connector on
each end. It is fully shielded inside. The one I made was not shielded,
was shorter (about 3 feet long) and worked without distorting the H
sync signal at all. I must assume that the internal capacitance of the
shielded cable was too much for the computer output to drive. In
addition, that particular monitor is sensitive to H sync distortion...
my other monitors (1084, 1902A) are not. My guess is that the 1084S and
the 2002 share the same sensitivity when it comes to marginal H sync.
As for the one 128DCR that showed the fault... after I changed the
driver chip (U23, 74LS244), that computer would drive the "marginal"  
monitor with the "bad" cable. The chip was not bad, but in this
particular application, it may not have been good enough. This is the
kind of fault that gives technicians and users alike prematurely grey
hair (too late in my case). I would expect to see this same problem in
the plastic C128 as well. The chip is the same, but the board location
is U24. In both models, the chip is next to the big RGB interface chip
inside the tin can in the middle of the motherboard. Fortunately, there
is plenty of room around the chip to do the replacement "operation".

My recommendation for users that suffer from this problem: the easiest
thing to do is change the cable. Go for the shortest one you can find
and try one that's unshielded if you can. Digital signals are hard to
"contaminate" from external sources. The next best thing... go for a
replacement chip. They are cheap... only a few bucks at places like
Jameco Electronics (www.jameco.com) in the USA. It's a generic TTL chip
and should be available worldwide. Cut the old one out by snipping each
pin close to the IC body. Solder the new one in by tack-soldering each
pin on top of the old one's pin stubs. That way, you never have to
remove anything from the motherboard. Desoldering is a pain, and you
can do damage if you lift a trace or get the board too hot. Don't feel
too bad if you do accidently lift a trace... I did when I was making
this investigation. Just make sure you fix it before you install the 
new chip or socket.

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