PDA

View Full Version : RAM repair for a C64



CommodoreKid
May 1st, 2009, 05:57 PM
Awhile back, after I KO'd one of my C64's (when I only had two), I attempted repair on the other one which had scrambled graphics at the time. I transplanted the graphics chip from the fried one into the scrambled one (mmmmm.... Commodore breakfast) So, in the process, I was able to get working graphics on the scrambled one, making it no longer scrambled, but revealing a new problem.

It would spit a message back at me about how it was "out of memory in 0". I could type stuff on screen, but it would only spit more errors at me or lock up completely. From there I looked up the problem online, it seems that I have an issue with one of two things: the RAM chips (mT4264-20 chips) or the RAM logic chips (7708 chips or something else... can't remember)

So anyway, I wanted to know if y'alls agree that this is the problem, and that the RAM or surrounding chips needs replacing. If so, any similar stories, how to fix this, or where to get parts would be appreciated --anything will help. Thanks.

tezza
May 1st, 2009, 06:13 PM
So anyway, I wanted to know if y'alls agree that this is the problem, and that the RAM or surrounding chips needs replacing. If so, any similar stories, how to fix this, or where to get parts would be appreciated --anything will help. Thanks.

Sounds like RAM to me. You did say anything would help so here is my adventure with an SX64 showing a similar problem (http://www.classic-computers.org.nz/blog/2008-10-18-Restoring-a-Commodore-Executive-SX-64.htm).

Tez

NathanAllan
May 1st, 2009, 06:42 PM
While you're at it, may as well upgrae it to 128K if you can fidn the chips to do it with.

channelmaniac
May 2nd, 2009, 03:44 AM
You have a URL for that upgrade? Google likes to return all links to the Commodore 128. Grrrrrr....

And on the memory problem, you have bad RAM or a bad PLA chip. Swap the PLA chip (906114-01 chip) between the 2 computers as a quick test.

I doubt it's a problem with the 7708/74LS257 chips, but it could be on an outside chance. RAM or PLA are far more likely.

CommodoreKid
May 2nd, 2009, 10:44 AM
While you're at it, may as well upgrae it to 128K if you can fidn the chips to do it with.
Um, if you know what I need to do that, let me know. It sounds like what my friend told me his father did back in the day to his C64



And on the memory problem, you have bad RAM or a bad PLA chip. Swap the PLA chip (906114-01 chip) between the 2 computers as a quick test.

Hmmm... My spare board doesnt have one of those easily accessable (its not in an easy-pry socket like the one Im fixing). I'll crank open my working one to borrow one of those... I will let y'alls know if that works or not.

CommodoreKid
May 2nd, 2009, 11:15 AM
Well, guess what... my good C64 also has its 906114-01 soldered in place... damn it all. So, that basically means that I either have to de-solder the one out of the scrapped one, or purchase one (fat chance!). So, Ill bust out the de-soldering braid, and get to work.

Something tells me that I will have to put it on some kinda platform, based on my track record with de-soldering larger IC's. Oh boy, will THIS be fun...

I'll keep y'alls posted.

dave_m
May 2nd, 2009, 12:34 PM
Take care not to zap your chips with static electricity (electrostatic discharge). When a printed circuit board is removed from the system, it is no longer connected to earth ground via the green wire on the power plug. However the chips on the board are generally protected in that they are connected to the ground plane of the board which presents a large surface area to dissapate static. The problem is handling the chips when they are not connected to the board.

If possible discharge static from yourself by touching something such as a metal chassis of something plugged into the wall, before handling the chips. Keep the chips in pink ESD foam or pink bags if possible when stored.

Things that create thousands of volts of static are shoes on carpet, and high speed air (heat guns and vacuum solder suckers).

The part you are removing is a bipolar IC which is less sensitive to ESD than NMOS and CMOS parts of the 80's, but it would pay to be careful.

tezza
May 2nd, 2009, 01:01 PM
Well, guess what... my good C64 also has its 906114-01 soldered in place... damn it all. So, that basically means that I either have to de-solder the one out of the scrapped one, or purchase one (fat chance!). So, Ill bust out the de-soldering braid, and get to work.

Something tells me that I will have to put it on some kinda platform, based on my track record with de-soldering larger IC's. Oh boy, will THIS be fun...


I've had to do a similar thing myself (http://www.classic-computers.org.nz/blog/2009-02-14-blowing%20up%20and%20then%20fixing%20a%20c-64.htm). i think these were probably soldered into the board from the factory. If you come across a socketed one in a unit, it's probably because someone's replaced it in the past. :)

It's a chore but take it slow, don't hurry and you'll be fine.

Of course it might be the RAM so don't be too disappointed if the repair doesn't work. C64s are very cheap. How about just getting a scruffy beaten up one that works and just replace the board?..or maybe you're like me and just enjoy the challenge. (:

Tez

Terry Yager
May 2nd, 2009, 01:38 PM
I was at the Sally yesterday, and they have(?) a C= 64 for fifteen buck$. It was just the breadbox, no P/S or any other goodies, and it had one key busted off, so I passed on it, but if someone would like it for parts, I'll go back and see if it's still there.

--T

channelmaniac
May 2nd, 2009, 01:57 PM
Not too bad, but I pay less than that for my parts machines. I'm about to clean out my source with a trade. Two Amiga CPUs (2000 and 4000) nicely loaded with accelerators and boards for the remaining C64s in his stash.

He keeps telling me it's more than will fit in the back of my F-150. :D

Now if this rain would stop so he can get them from the storage truck in his pasture. Sigh.

Terry Yager
May 2nd, 2009, 02:04 PM
Not too bad, but I pay less than that for my parts machines.

I know what ya mean. I'd have taken it if it was under five, but that's just a leetle steep for an untested, incomplete 64...

--T

CommodoreKid
May 2nd, 2009, 09:06 PM
Well, I tried that little trick of replacing the 906114-01 chip with one from my scrappedC64, and no luck. I spent nearly an hour desoldering the chip, only to plug it in with the same exact message, (even the random ! marks were in the same locations!). So, I switched back to the previous one, because its just in better condition overall.

Oh, and in the process of desoldering, I dropped the iron, and burnt a small spec into the carpet in my den/geek room. I feel really smart now.

So, based on a tip I read over in that PET 30XX repair thread, apparently "shorted" RAM chips get really hot, really fast. So, I got an idea! Let my busted C64 sit for 5 mins or so, and feel all the RAM chips. If one is hotter than the rest, I will know to replace it, right?

So, I sat it on for a short time, and the only ones that got hot were one of the kernals ROM's, the 6510, and the graphics chip...

...except for something else. BOTH of the 7708's leading up to the RAM were getting very hot, very fast. All of the RAM chips were pretty cool. The thing is, in the spot where the one being repaired has 7708's, the spare board has 74LS257A chips... Im assuming that means I cant interchange them, can I? Anyway, do you guys think that I should go ahead and replace these?

tezza
May 2nd, 2009, 09:22 PM
smart now.

So, based on a tip I read over in that PET 30XX repair thread, apparently "shorted" RAM chips get really hot, really fast. So, I got an idea! Let my busted C64 sit for 5 mins or so, and feel all the RAM chips. If one is hotter than the rest, I will know to replace it, right?

Let me know what y'alls think about this plan of action!
.

Defintely. A warm chip is normal but if there is a RAM chip much hotter than all the rest, that's probably the culprit.

A RAM IC can still be damaged and not become hot though. It all depends on how its damaged.

Tez

channelmaniac
May 3rd, 2009, 06:43 AM
Defintely. A warm chip is normal but if there is a RAM chip much hotter than all the rest, that's probably the culprit.

A RAM IC can still be damaged and not become hot though. It all depends on how its damaged.

Tez

Exactly!

You can have a bad cell in the RAM and it'll run at a normal temp. The only time they get roasting hot is if they have an internal short.

The 7708/74LS257 chips are interchangeable. I save and test all the odd duck labeled MOS chips like the 7708 just in case someone really wants to replace a chip with the MOS original version.

It's normal for those to get pretty warm, even uncomfortably warm. If they are finger blistering hot then that's a bad sign.

RJ

NathanAllan
May 3rd, 2009, 08:39 AM
I've looked and looked but can't find where I read about the ram upgrading. I sent an email to Bo Zimmers asking if he knows of a page or article. Bo is the MAN when it comes to anything Commodore related.

Nathan

CommodoreKid
June 17th, 2009, 08:35 PM
After my brief hiatus (of trips, work, and otherwise shenanigans) I have returned to try and resurrect my poor broken C64's RAM

Well, I flipped through all the electronics catalogs that I have, as well as their respective websites, and I cannot seem to find a place that sells 4264-20 memory chips to make the repair...

Therefore, if anyone knows of a good reputable place to purchase said RAM, I would love to hear it. I would like to try and avoid eBay. Anyway, any suggestions of where something like this would be appreciated, as I continue looking for some on my own.

PS: thanks for the feedback about my diagnosis- it helped.

channelmaniac
June 17th, 2009, 09:39 PM
I have 'em on my site...

http://www.arcadecomponents.com

RJ

CommodoreKid
June 18th, 2009, 03:03 PM
http://www.arcadecomponents.com

Perfect! This site will come in handy for any other refurbishing projects down the line.

Which ones are compatible with with the 4264-20? (its not listed, but Im sure there are some that can be used in its place)

channelmaniac
June 18th, 2009, 07:04 PM
There are a lot of different part #s on the 64k x 1 DRAM chips used in the Commodore 64.

If there are 8 of them then it's a 4164 or compatible chip

If there are 2 of them then it's a 4464 or compatible chip

By the way, I have just a small (ahem) quantity of Commodore chips.

http://farm4.static.flickr.com/3313/3639673989_9e569cb536.jpg

RJ

CommodoreKid
July 21st, 2009, 05:13 PM
Well, I finally got back in town, and the first thing I did was install the RAM I ordered. Finished de-soldering those chips with the 40-watt, and then put the sockets in with the 30-watt. Two traces got torn up/interupted, so I had to run some wires to fix it.
http://farm3.static.flickr.com/2519/3744776364_e52c0e34ee.jpg
Doesnt that look nice? Im sure it does. But you know what DOESN'T look nice? The screen that stared back at me, upon plugging in the board for a test run. Crap.
http://farm3.static.flickr.com/2560/3744776466_8cff8fbdc4.jpg
So, any ideas of what happened, or whats wrong now would be greatly appreciated. I hope its nothing big, considering the graphics looked fine before, when the old RAM was still present. *sigh* Its always something, right?

Druid6900
July 21st, 2009, 06:33 PM
Make sure none of the pins missed the socket wipers and got bent underneath the chips when you put them in.

CommodoreKid
July 21st, 2009, 07:08 PM
I *carefully* put each new RAM chip into the sockets, and only bent the pins once (don't worry, I straightened it, and reseated it). But I checked again simply because you mentioned it. The pins are perfectly socketed.

I also tried pulling some of the ROMs to see if I could narrow it down. For obvious reasons, when the character ROM was removed, an entirely different image was present much simpler, and more colorful. No change was present when the Kernel or BASIC interpreter were pulled.

Any other suggestions?

cosam
July 21st, 2009, 11:50 PM
Well, I finally got back in town, and the first thing I did was install the RAM I ordered. Finished de-soldering those chips with the 40-watt, and then put the sockets in with the 30-watt. Two traces got torn up/interupted, so I had to run some wires to fix it.
...
So, any ideas of what happened, or whats wrong now would be greatly appreciated. I hope its nothing big, considering the graphics looked fine before, when the old RAM was still present. *sigh* Its always something, right?
I'd try beeping out all the connections to the new RAM. 30 or 40 watts is a lot of juice for those narrow PCB traces, so maybe there's one lifted of broken somewhere you didn't notice. The way these things are connected up, a break can cut a signal to a whole bunch of RAM chips. I'd even go out on a limb and say start looking around A7 ;-)

CommodoreKid
July 22nd, 2009, 09:51 AM
What do you mean by A7? I looked on the board, and I spot nothing marked with an A.

And by beeping out, Im assuming you mean using the continuity tester function on a multimeter (mine is always silent). I betcha another trace got destroyed that I missed. I will test as many as I see.

Unfortunately, 30 watts is as small as I can get, and even that has a hard time melting my solder. Thank you for the advice.

cosam
July 22nd, 2009, 10:06 AM
Sorry, yes - I mean checking for continuity. A7 (i.e. address line 7) would be pin 9 (with the notch at the top, far bottom right corner) but it's worth checking the lot to be sure.

Good luck!

CommodoreKid
July 22nd, 2009, 11:19 AM
Ah, now I understand which pin you mean. So what would I be comparing A7 to? Am I checking its continuity with something specific? Or am I looking for something happening at that pin on each of the chips?

If by each of the pin 9's has continuity to all the other pin 9's of the RAM, yes they do.

BTW, I pulled all the RAM out, and doubled checked my solder points (touched up one or two) then tested continuity on the traces with some wear on them. Based on the ones I tested (all underside ones, and at least half of the topside, mainly damaged looking ones) and they all checked out fine. I also did a test in which the RAM was completely out, and the computer gets fired up- different graphics appear, mainly with alphabetic characters, but still the same garble or colors and other graphic bits intertwined.

I will keep trying. *sigh* I will get this thing working yet.

cosam
July 22nd, 2009, 11:50 AM
Continuity between all the RAM chips would be the first thing to check. That's probably easiest done with the chips out, measuring on the sockets. Then you need to check each pin is also connected to the rest of the circuit, either by using a schematic or simply visually checking which traces go where. I've skipped that last step before and it turned out there was a break (practically invisible, of course!) between the first chip in the chain and the CPU...

Druid6900
July 22nd, 2009, 07:09 PM
The other thing I'd check would be the pins on the RAM chips you removed.

What you are checking for is a little metal cylinder around one of them to see if you pulled out a via when you were removing the chips.

CommodoreKid
July 22nd, 2009, 07:19 PM
The other thing I'd check would be the pins on the RAM chips you removed.

What you are checking for is a little metal cylinder around one of them to see if you pulled out a via when you were removing the chips.

By that, do you mean the trace popping up with the chips? I watched carefully with each one: only one chip did that, and I followed the traces, and put a little wire in to replace it (not visible in the picture, because its attached to the socket from the underside).

Alright, I will check all 128 connections... crap, where am I going to get a good schematic that is board shaped? I will start looking...

Druid6900
July 22nd, 2009, 07:24 PM
No, a via is the plated hole that goes through the board from the component side to the solder side and, if you yanked one out, you're not quite screwed, but close. The are several ways of fixing it.

The way to determine this is to check the continuity from the pad (not the pin) on one side of the board to the corresponding pad on the other side of the board for all the RAM.

CommodoreKid
July 22nd, 2009, 08:38 PM
Oh great.... that sounds like fun. But Ill do that next.

I just spent the past hour observing schematics, and checking where each pin should lead:
Pins 3-13 & 15-16 all properly connect from one to another, no matter the chip selected. 5,6,7,12,& 15 were all easy to test their continuity to some other component on the board, and they all checked out. Pin 1 is supposedly not connected, according to schematics I dug up.

Its pins 2 & 14 that are giving me trouble. Some of them have continuity to others, and some seem isolated. Also note, that 2 & 14 are connected to each other, which makes sense, if BOTH are giving trouble. U9, U10, U21, U22, & U23 all are properly connected to each other, but U11, U12, & U24 are the isolated ones. They have no contact with each other, or the other chips.

The other 5 chips have continuity at pins 2/14 all the way to the 6510 on pin 37 (D0) but those three outcasts are not connected. *sigh*

I will try checking the continuity from the pin to the pad. At least Im getting somewhere.

cosam
July 22nd, 2009, 11:46 PM
Pins 3-13 & 15-16 all properly connect from one to another, no matter the chip selected.
...
Pin 1 is supposedly not connected, according to schematics I dug up.
So far, so good ;-)


Its pins 2 & 14 that are giving me trouble. Some of them have continuity to others, and some seem isolated. Also note, that 2 & 14 are connected to each other, which makes sense, if BOTH are giving trouble.
As yes, I was forgetting each of those chips connect to a single bit of the data bus. So while 2 and 14 on each individual chip should be connected, they shouldn't be connected from one chip to the next.


U9, U10, U21, U22, & U23 all are properly connected to each other, but U11, U12, & U24 are the isolated ones. They have no contact with each other, or the other chips.

The other 5 chips have continuity at pins 2/14 all the way to the 6510 on pin 37 (D0) but those three outcasts are not connected. *sigh*
From the schematic you will see that each RAM chip has a D<number> line going to it. These should connect up with the D<same number> pin on the CPU. If you really are measuring close to zero ohms between pin 2 on different chips, you have a short somewhere. Hopefully you were just measuring the resistance of some part further down the line.

Speaking of shorts, it's worth checking you didn't inadvertently create any between adjacent pins when soldering. Thankfully that's easy to check by "walking" your test leads up each side of the socket and checking for continuity.