Hi all,

I've got my hands on a defect c-64. It will start with borders ok and a screen full of scrambled characters with usually on or more "flickering" characters.
Now the fun thing is when I insert a power cartridge, freezing usually succeeds with a correct power cartridge 'freeze' screen, except for one thing.. The logo consists of 4 sprites, and sprite 3 and 4 are the same sprite 4! (So you'll see the "starry" 1985 part duplicated twice).

More fun is I can actually start the power cartridge monitor roughly one out of 4 tries. I filled memory from 1000-9FFF with AA's and 55's, saved that to disk and tried to analyze what is going on in RAM. A saved romdump of the kernel shows no troubles so ROM access seems to be ok..

Ram however is another thing.. I found errors in bits 0, 1, 5 and 6 all over the place and there seem to be some patterns.. For starters, its always only one bit failing at a certain location.

Bit 0 fails on most $XXFA, $XX87 and $XX92 locations.
Bit 1 fails on most $XXE8 locations.
Bit 5 fails on most $XX95 and $XXA2 locations.
Bit 6 fails on most $XX89 locations (and $XX2D, $XX13, $XXA1 more seldom).

Always only those addresses, those bits. Other addresses show no problems.

My very first guess before analyzing the memory dump was a faulty PLA, but this pattern failures made me think it might be multiplexers U13 or U25. But I can't make any sense in combining the failures to pinpoint the defect chip

So if some experts are around.. What would be more probable? A faulty CASRAM signal from the PLA? 4 defect RAM chips? One of the multiplexers? Or something I overlooked?

Like to have my theory of what is wrong in place before I start soldering parts in and out

I would first check the voltage going into RAM chips, then replace one of them to see if faulty bit goes away. It might be that the previous owner stubborny tried to use the computer after RAM failed, resulting in multiple chip failure.

4.98V on all ram chips between pins 8 and 16

Tomorrow i'll check if my (working) second bread-box also has one ram chip on a socket. I've read somewhere you can use the 'piggyback' method to check ram chips I think?

Argh.. Turns out my second 64 has ALL major chips socketed except for.. You guessed it, the RAM chips
Also strange that this one has no interference isolation whatsoever.. No iron "box" around the VIC circuitry, no cheap cartboard-aluminiumfoil thingie, nothing..

Now I should have some old A-500 with a RAM expansion board lying around somewhere... Believe they usually use 41256 chips that.. But i am inclined to say its not worth butchering that for a 64.. Better wait till I get my hands on some spare c-64's I guess..

Damn..

Funny problem. Keep us posted. I'm curious...

Desoldered RAM chip U21. The idea was to add a socket and try and swap chips between U21 and U23 (already on a socket).



However.. RAM chip U21 is no more. The patient died during the operation i'm afraid I did manage to solder the socket in nice and clean though!;)

I think i'm going do a little experiment. I found some 44256 chips in some old PC card. Now in theory one of those should be able to replace 4 4164 (and still have an addressline left). I seem to have 4 faulty 4164's so... Lets see what happends

So there it is.. Wires are from an old flatwire floppy-drive cable and I had to rip them apart from eachother to get something working.. (Interference?)

C-64 now starts with an almost correct start-up screen! Power cartridge logo fully and correctly present! There are still some "flickering" characters though.

For now the 44256 only replaces U21 that responds to bit 0, and guess what.. Bit 0 failures are gone from the dump!

So it seems we can conclude the problem actually is 4 defect ram-chips! I'll now proceed to connect the data lines from the 3 other defect chips to the remaining databit lines of the 44256.

Hopefully I will end up with a working C-64!

Keep you posted...

That leads me to the following question...
Would it suffice if I just cut pin 2 and 14 (din/dout) of the 3 other ram chips and reconnect one of the print leads to the 44256? Or can anyone think of a reason why I should desolder the chips completely?

Why not cut /CAS and tie it to +5V. That way output pin should always be in Z state.

(I would cut all pins and then desolder them separately, then end up the job by doing the same with the GOOD 64kx1 chips and go with two 64k*4 chips. Much neater.)

Sounds good!



If you want "neat" just replacing the 4 defect chips would be the best way to go I guess. However, for a defect c-64 I bought for 5,- i am not about to invest 4x7euro in some old chips
A 44256 found in a old soundcard should do just as well in theory so why not try for fun and... something

I confess.. I did a very dirty thing

I connected the remaining datalines of the 44256 directly to pin2 of the defect ram chips.
I seem to have a working C-64! Hurray

Well, don't forget that the RAM chips are probably broken for a reason. Could be the result of a failing PSU in the past time, or the previous owner doing something nasty like shorting the cart or user port perhaps. I'd check for hot chips. Also if the PSU came with the computer, check the output voltage. Also, it's not uncommon that the DC rectifier fails. I would take out the bad chips as they are probably failing and when they do they will certainly take more circuits down as they go.

Hmm.. I'm using the PSU that came with it and i've found no failures.. Even let it run another C64(c) for some hours (yes I know that was a risk;) and it seemed to function just fine.

Also I have a hard time believing that the result of a failing rectifier would only hurt just 4 ram chips, and not even destroy them completely. My experiences with applying a wrong voltage to a circuit is mostly that the circuit is utterly detroyed in micro-seconds. But I must say my experience with a C-64 are limited so who knows. Theory that something happend on the user or expansion port just sounds more plausible to me.

Previous owner sold it stressing explicitly he"had not tested" it. Can always try and email him if he knows what happend. Maybe he's honest enough to admit he knew it was broken now

I started out all this by checking for hot chips. Found none except for the PLA and SID that get quite hot anyway..

And yes, I will eventually remove the chips anyway. My "problem" is that I used to have allot of equipment and tools, but foolishly decided to give away or dump everything when I moved some years ago.. Gone are my oscilloscope, EEPROM burner, lots and lots of chips.. All thats left is a simple soldering station. Ohwell... Thats life i guess

But as I actually seem to enjoy this tinkering with old hardware maybe I'll use this C-64 as an experiment board and try and build some fun stuff like an logic analyzer on a PC LPT port or a simple eprom burner to see if I can create my own kernel.. Who needs a PIC programmer if you've got an old C-64?

Ehh.. Turns out i'm not using the PSU that came with it at all! It's still in the bloody box!
Note to self: Never use that one

On the possitive side, previous owner admits he knew it was defect. On the negative, he says he just got it that way and has no idea what caused this.

Not if you apply just a bit more than the chip tolerates. There's always the weakest link which breaks before any other.

And then one tends to try again just to make sure that it wasn't just a temporary problem, which is why I wrote this: