C64 phase 2 of repairs

[jms2]

Back in 2016 and 2017, I posted about my misadventures trying to fix a breadbox C64: viewtopic.php?f=45&t=9452. By the end of the thread, I had started to muse about the possibility of a dodgy power supply, which is a common issue on those machines.

I found that the Commodore PSU was putting out 5.7V, which I have come to appreciate is too much. To address this problem, I made a "Y" shaped splitter which fed in the 9V from the old Commodore PSU with 5V from a modern Samsung 5V USB charger (which was all that I had to hand at the time). Apparently the 9V is less sensitive to over-voltage so it's OK to use the old PSU for that.

This setup wasn't successful - although the charger was rated at 5V, 1.55A, I found that it's output was closer to 6V (whether under load or not). At first I thought that possibly that I'd demanded too much current from it (the C64 apparently needs 1.5A). So I bought a power supply which has selectable voltages of 4.5V, 5V, 6V etc. The rating is 2.4A at 5V, and the supply did claim to be regulated. It's CE marked and cost about £12, and so I thought it ought to be good enough.

At first this PSU delivered 5.1V onto the main board (as measured at the RAM chips), so I thought my voltage problems were solved. However, after only a few minutes operation I have found that the voltage is back to around 5.7V again!

What is particularly puzzling to me is that the PSU seems to have been permanently affected by it's encounter with the C64. Regardless of whether it is powering the C64 or unplugged from the machine, I find that all the marked voltage settings are delivering more than they did initially - "4.5V" actually gives over 5V, "5V" gives 6V, "6V" gives 7V etc. It definitely didn't do this when I first got it!

Is it plausible that the C64 could have had this effect - and if so, any ideas of what I could do to fix it? I haven't renewed the big electrolytic capacitors on the board yet (I think... it's been such a long time since I last had time to dust off the machine!) so could a fault on these pull too much current and somehow damage a PSU?

[1024MAK]

Any half decent regulated PSU should have protection against over current.

If you have a 'scope and some test loads (6V motorcycle filament lamps are useful) along with your meter, you can test the PSU to see if it is regulating it's output.

Putting to one side, overloads (over current) and short circuits, the only devices that normally kill or damage PSUs are inductive loads like motors, relays and such like. I would not normally expect semiconductors to damage a PSU. With one exception. chips that require multiple power supply rails, e.g. The SID sound chip, 4116 (and equivalent) DRAMs etc. These can develop internal faults and feed the higher voltage rail to pins that should never get such a voltage.

Mark

[jms2]

Thanks Mark, I'll see if I can get a bulb off my dad.

I'll also try taking the SID out to see if that makes a difference to the voltages.

[topcat96]

It's not the C64 destroying the PSU but the other way around. It's a known problem with the original power supplies output exceeding the norm of 5v as old age creeps in. Commodore in their infinite wisdom, made sure the power supplies were non-repairable by filling the case with epoxy. It's this that causes excessive heat with in turn, cooks the electronics causing the problem.

Ray Carlsen has created properly regulated power supplies and 'save' units that negate this and keeps everything working properly. You can purchase here: http://personalpages.tds.net/~rcarlsen/

I have one of his M64/128-2D units myself and can not fault them in any way.

[jms2]

It's not the C64 destroying the PSU but the other way around.

Thanks, but I know about this problem - I think my C64 has been affected and that's why I have tried to rig up my own supply in the form of a cheaper alternative to the Ray Carlsen solution. I got the idea from this website.

I'm going to test the regulation of my PSU using a 6V bulb as Mark has suggested, but the only other avenue I can think of is to change the big electrolytic capacitors on the C64. I have not done this yet, and I reckon it could be necessary.

[topcat96]

After a quick scan of your previous thread concerning this ... Does the breadbin switch on? LED light up? Get any form of tv signal in its current state?

http://personalpages.tds.net/~rcarlsen/ ... c64-ic.txt Quick guide to chips and problems associated with them.

EDIT: Can't speel fore tofey

[jms2]

After a quick scan of your previous thread concerning this ... Does the breadbin switch on? LED light up? Get any form of tv signal in its current state?

Yes, I can switch it on and get a red LED. Before buying the dead test cart, I got a totally black screen but latterly I was able to get multicoloured jumbled text - only sometimes though.

Now, using the cartridge, I get a very clear black/white flashing on the screen. This indicates that ram chip U12 is defective (but on reading the manual it is clear that other failure modes can cause the same thing).

When I got the machine it was fitted with Samsung 4164 DRAMs. These did not seem particularly hot, but eventually (after I'd eliminated various other things) I socketed them and bought some new Hitachi 4864s. I notice that these seem to run hot (some more than others) so I'm progressively putting the Samsung items back in. These are noticeably cooler but I still get the same single-pulse white flashing of the screen.

I think I have resolved the voltage issue though. My multimeter seemed to have got knocked out of adjustment - I tried testing various things such as some AA batteries (1.7V), a 9V battery (10V) and a 12V lighting system (14V). It was apparent that a roughly 10% error was present, so I adjusted the variable resistor in the meter until all the above reported the correct voltages. The voltage in my C64 is now back to 5.1V.

http://personalpages.tds.net/~rcarlsen/ ... c64-ic.txt Quick guide to chips and problems associated with them.

I've used that (and other sources) to put together the following table, which shows what I have done so far. White background in the rightmost column means I have not taken any action so far, green means I think I have eliminated that IC.

My next planned steps are:

1) Swap out the other MOS7708 chip
2) Fit new power smoothing caps on the main board (although the 5V line does look very steady to be honest)
3) Fit some more new DRAMs

Beyond this I am not sure. I tried checking all the DRAM traces for continuity, and found a broken trace - but when I fixed it, it made no difference.

This machine is proving a very stubborn fix!

[topcat96]

You've probably already done so (but good to ask anyway) ... have you also checked that the board itself is receiving the correct voltages and that the on/off switch is properly soldered and not arcing?

[jms2]

The board does seem to be getting the right voltages. I'm not sure what to look for at the on/off switch though - there's no visible arcing, but is that a reliable guide?

The fact that I got different results on occasion could possibly indicate inconsistent behaviour at the switch, maybe?

[topcat96]

The board does seem to be getting the right voltages. I'm not sure what to look for at the on/off switch though - there's no visible arcing, but is that a reliable guide?

The fact that I got different results on occasion could possibly indicate inconsistent behaviour at the switch, maybe?

Although any arcing is not a reliable diagnosis, but it has been noted by others in the past as a symptom of a switch being well used during it's lifetime and due for replacement. Also check the solder points around it for dry joints, cracks in the solder etc ... 30+ year old solder isn't the most reliable substance in the world, as we all know.

[1024MAK]

You may want to consider buying a new multimeter. Would save you from another wild goose chase...

To test the switch, there are various things you can do.
First off, the two most common problems are dirty or damaged switch contacts and dry solder joints.
The dry solder joints are mostly caused by mechanical stress. I recommend removing all the old solder, cleaning the pads and switch legs/pins, then using new solder.

So with the power supply disconnected, using a multimeter on the low ohms range, connect the meter across one set of contacts. Then repeat for the other set of contacts. with the switch off (open), the meter should show over-range or Infinity. With the switch in the on (closed) position, you should get a resistance of less than 0.5 ohms (not including the resistance of the meter leads).

Next, connect the PSU and power up the computer. Now switch the multimeter to the appropriate voltage setting (DC or AC). With the switch in the on (closed) position, measure across the switch contacts. You will need to be on a low voltage range. You should not get any significant voltage here. If more than 100mV then then that's not good. Do the same for the other switch contact. Remembering to switch between DC to AC or vice-versa.

If the switch is found to have a contact problem, you can always by-pass the faulty contacts with a piece of suitable wire. If you do this, do it for both sets of contacts. Not just one.

Good luck

Mark

[jms2]

You may want to consider buying a new multimeter. Would save you from another wild goose chase...

Yes, fair point! I know this one is cheap (it is branded though - Hills), but I have only ever needed it to give me indicative results. eg: is something powered or not powered, connected or not connected. This is the first time I have needed to measure a voltage with reasonable accuracy.

To test the switch, there are various things you can do.

Thanks - only had a couple of minutes spare so far. With contacts open I am seeing infinite resistance, and with them closed it is low, but not quite as low as half an ohm. This could be due to tarnish on the switch legs though, so I'll retest.

[jms2]

Switch seems fine. And there's more good news: I have been working through the ram chips, removing the hot Samsung chips (in order of temperature) and putting back the original Hitachi chips. When I renewed U21, the flash code changed to 2x flashes - indicating a fault at U24. Swapping U24 back to Hitachi gave me... a totally black screen. Leaving U24 empty gives me back the double flash. This is the best useability I have ever managed to get from this machine!

So I think that at least one of the hitachi rams is dud, and probably all the Samsung ones.

[jms2]

I obtained 4x new TI 4164s today. Plopping one into the vacant slot has brought the machine to life! The dead test cartridge runs and finds no problems. Even the SID is fine.

However, without the cartridge the machine still gives a black screen. I'm going to dump the BASIC and Kernel roms again.

[topcat96]

You may have a faulty character ROM. Plug in a game or utility cartridge as they usually hold their own font data and bypass the machine's. If they work okay then it's another pointer towards it

As you already have output from the dead test cartridge, try the above and see what occurs.

[jms2]

Unfortunately, although I have many tape games, I don't have any cartridges. But you are right - I have not done anything with the character rom so far.

Any kind of progress is rewarding though. I have had nothing but the black screen of death from this machine ever since getting it. Now the repair process has some direction at last.

[topcat96]

You could try loading a cassette game, preferably any one by Ocean with loading music. See if you get any screen output and sound that way.

[jms2]

I tried dumping the Kernal and BASIC roms and found that they are both OK (as before). However, to get a reliable dump of BASIC I found I needed to clean the pins. Doing this caused the machine to boot!

It still doesn't seem 100% reliable though. Sometimes when I switch it on, I just get a blank screen. However if I try a few times it works OK.

Also during the dead test, whilst the sound test produces musical notes I see from the documentation that it is also supposed to produce noise. My SID must be slightly defective because it doesn't make noise. However I will try to test from BASIC - once I've figured out how to drive a C64 that is!

EDIT: From BASIC I do seem to get noise. Lots of POKEs involved though...!
EDIT 2: The contrast is pretty minimal. Is it supposed to be like that?

[jms2]

Also.. any thoughts on what is going on here? If I change the colour to red, everything goes super-blurry. Purple is the same. Other colours are not so bad.

This is using a proper C64-SCART cable with LCD TV.

[dv8]

Also.. any thoughts on what is going on here?

It's the limitations of a composite video signal, some colour combinations will blend together like this.

It will look better if you change the background colour to black with POKE 53281,0

The C64 can do separated chroma/luma video which gives a much sharper picture if your TV has an S-Video input.

BTW, SCART doesn't always mean RGB. In this case, your cable will only be a composite signal which is no better quality than a pair of video/audio phono cables.

[jms2]

Thanks... I just googled it and found the same thing. Not what I expected, but as you say the lead may well just be composite video.

I am fighting the urge to try some games... going on holiday soon so I really ought to be doing some useful jobs instead!

[topcat96]

Good to see there's finally video output from the C64!

Here's a quick program from C64-Wiki to give the SID a small workout ... Not sure if it will produce "noise" but it's worth a try.

Code: Select all

0 REM *** C64-WIKI SOUND-DEMO ***
10 S = 54272: W = 17: ON INT(RND(TI)*4)+1 GOTO 12,13,14,15
12 W = 33: GOTO 15
13 W = 65: GOTO 15
14 W = 129
15 POKE S+24,15: POKE S+5,97: POKE S+6,200: POKE S+4,W
16 FOR X = 0 TO 255 STEP (RND(TI)*15)+1
17 POKE S,X :POKE S+1,255-X
18 FOR Y = 0 TO 33: NEXT Y,X
19 FOR X = 0 TO 200: NEXT: POKE S+24,0
20 FOR X = 0 TO 100: NEXT: GOTO 10
21 REM *** ABORT ONLY WITH RUN/STOP ! ***

[jms2]

That's the one I tried - it works, and does produce noise. Not sure why the test cartridge doesn't.

This machine isn't released from hospital yet though. I have tried giving it a more thorough workout and various problems have come to light (probably all related to each other I suspect):

1) Whilst it always boots the dead test cartridge successfully (and it never finds any problems), it doesn't reliably boot without it. Sometimes I have to switch on, then off, then on again.
2) It refuses to load any tapes. Whilst the screen clears as it should, and the tape motor runs, nothing happens after that (I've tried playing all the way to the end of several tapes). There appears to be no way to reset the machine after this, and after switching it off it then won't boot for a while afterwards.
3) If I do get it to boot, and just leave it at the startup screen, it will sit there happily for about 5 minutes. Then something odd will happen, such as strange characters appearing.
4) At switch-on, sometimes the cassette motor starts but then immediately stops (this tends to be a good sign) - but at other times it just keeps on running.

These faults suggest some kind of issue with the 6510 and possibly an overheating problem with the VIC-II. I do have spares of both from my scattergun troubleshooting process.

[topcat96]

It looks like there's still a possible issue with the RAM, but that could also indicate a bad VIC-II (overheating?) so best swap it out for a known working one.

Also try anything that clears or updates the screen; RUN/STOP+RESTORE, SHIFT+CLR/HOME and scrolling down (or rather, causing the screen to scroll up by moving down) with the arrow keys should indicate whether it's a permanent memory issue or something that's intermittent.

[jms2]

Those screen-clearing functions do work, but I have not tried scrolling the screen; that's a good tip.

Unfortunately my tinkering time will now be zero for the next couple of weeks - but I'll be having fun doing other things so it's all good.

[topcat96]

Enjoy your break

If you are heading somewhere nice on this planet then don't forget to send us a postcard!

[jms2]

Hello everyone, I'm back now... and I have had another look at the C64. Sorry, I didn't get time to send a postcard from Cornwall but I did check the site from time to time!

So... I found a couple of dry joints on my Kernal rom socket, and reflowed them. The machine seems to boot reliably now, and I have even managed to save my own (1-line) program onto tape and reload it! Hopefully I just need to renew all the capacitors in the machine and it'll be done...

Loading commercial tapes is something I have yet to succeed at though. I have only tried two, Way of the Exploding Fist and Green Beret - chosen because they also exist on the BBC of course. Way of the Exploding Fist uses some kind of turbo loader system, but whilst it started loading ok (the system found the first file, then I pressed the C= key to load it, and saw moving coloured bands in the border), the process halted after a minute or so. At this point the border went from multicoloured to solid white.

With Green Beret the machine doesn't "find" anything on the tape.

One issue I've noticed is that once I start loading (or trying to load) something, it does not seem to be possible to reset the machine by pressing Run Stop + Restore. Instead I have to switch it off completely. Does anyone know if that is normal for the C64?

I know of two possible issues with the tape player:

1) The drive belt was perished, and so slipped under load. I have replaced it temporarily with a rubber band of roughly the right length. This doesn't slip to the same extent, but I guess it might not be perfect.
2) I have not cleaned the heads with IPA or anything else.

I think my next port of call will be to clean the heads.

[jms2]

Cleaned the heads now, and whilst the cotton bud didn't come out visibly dirty, tape loading now seems to work - at least for some games. I have not re-tried the ones that didn't work initially, but others certainly work.

[topcat96]

Cleaned the heads now, and whilst the cotton bud didn't come out visibly dirty, tape loading now seems to work - at least for some games. I have not re-tried the ones that didn't work initially, but others certainly work.

Games that use turbo loaders are notorious for needing a fully working C2N tape player ...

If you have any original game on this list then try loading it. If it crashes or the C64 resets during loading then check the head alignment after another good clean.

http://www.tzxvault.org/C64/novaload.htm

Also factor in that the games you have tried and found not to load, it might be a simple case of old age or how they were stored taking it's toll on the tapes.

[jms2]

Thanks for the tips. I think my temporary drive belt (aka, a rubber band) is not quite up to the job because on some tapes the speed drops right off. This is noticeable by the coloured loading bars on the screen getting wider and wider, and the tape counter slowing down!

I didn't find any dirt on the head but I have thought about adjusting its alignment, so I'll look at that. I don't have any of the games on that list though.