Project VIC

[chysn]

I recently purchased a non-functional VIC-20. I knew it wasn't working, but it was super-cheap. By "super-cheap," I mean that I can sell the keyboard for more than I paid for the VIC-20, if it comes to that, and have a good supply of parts for my two functional VICs. The idea here was to give myself a project, to gain some experience with diagnostic tools and--hopefully--to get another VIC-20 into the world.

The machine is one of the "newer" VIC-20s: an NTSC CR board with a C64 keyboard (grey function keys). The power light comes on, but there's no video. There's no RF interference on an AM radio when it's powered on. Power is getting to the cassette port pins, so my logic probe can be powered.

So far, I've used my logic probe to check the power pins on the chips. They all are getting high voltage where they're supposed to. However, all pins on all chips are static: they're either high or low or (in cases that seem appropriate to me, based on diagrams) not connected. Nothing is alternating between high and low.

This suggested to me that the clock isn't running. So I looked up the 6560 pin that should carry the 1MHz clock, and this was pin 35. Pin 35 is always high. The 6502's clock in looks like pin 37. This pin is always low. Why doesn't it match? Because the clock output goes to UB4 pin 11 and gets inverted at pin 12 before going to the 6502. This high-to-low inversion checked out at the 7406 chip.

All right, so I think that the 6560 is bad in this machine. I think that it's bad because it's getting power, but the clock isn't running. The 6560 also gets hot to the touch. That is, if I keep my finger on it for 30 seconds, my finger gets slightly uncomfortable. But this is my first attempt at a VIC-20 repair, so I am inexperienced. Can anyone see any problems with my analysis? Could the 6560 be okay and still not tick the clock?

[Mike]

I dug maybe a level deeper here and learned that the 6560 does not generate a clock signal by itself. The clock looks like it comes from a circuit built around UB9 (7402). It looks like the final output of this circuit is UB9 pin 13. In this VIC, that pin is always high. The inputs for that NOR gate (pins 11 and 12) are both low. So that gate checks out. But pins 1, 5, 6, 8, and 9 of UB9 seem unconnected.

What I'm thinking is that maybe it's not the VIC chip, after all. It could be one or more of the capacitors, which look like the beating hearts connecting the NOR gates together. If I can get that little oscillator circuit running again, maybe the 6502 will start cycling.

[...]

I moved the OT part of this post and OT follow-ups into the Off Topics section.

[Mike]

If you want to check the function of the oscillator circuit, a simple logic probe isn't going to tell you much. More likely, it might even stop the oscillator from working, because it adds additional load to the circuit!

With the CR boards, two gates of UB9 form the oscillator circuit, with Y1 as resonant quartz. You'd need a high-ohmic, dividing logic tester (say 2^20:1) and probe either pin 13 or pin 10 of UB9. Testing these pins does not load the actual oscillator circuit and 2^20 roughly divides by 1000000, so you then see a fast blinking change of levels. If you don't get that, UB9 and probably Y1 need to be replaced.

You should also check the supply voltages. The +5V, when measured with a multimeter should be within +/-5% (DC setting), its ripple (multimeter at AC) should be less than 100 mV. If there's excessive ripple on the supply, the VIC chip is known to cease video signal generation.

[chysn]

If you want to check the function of the oscillator circuit, a simple logic probe isn't going to tell you much. More likely, it might even stop the oscillator from working, because it adds additional load to the circuit!

Since this seems to be an equipment-specific question (does my logic probe impose enough load to stop the oscillator from working?), I opened up my functioning VIC-20 (with the exact same PCB) and checked pin 13 of UB9. In this machine, the logic probe beeps along happily on that pin, indicating regular high/low changes of the oscillator circuit.

You should also check the supply voltages. The +5V, when measured with a multimeter should be within +/-5% (DC setting), its ripple (multimeter at AC) should be less than 100 mV. If there's excessive ripple on the supply, the VIC chip is known to cease video signal generation.

This is probably supposed to be the very first step of any diagnostic, right?

I'm using the same power supply that I use for my working CR, and it's a modern one, so I wouldn't expect power supply issues. Could faulty hardware introduce ripple? If so, where should I measure this on the board?

[Mike]

Having a functioning unit to cross-check is always a good idea ...

Since this seems to be an equipment-specific question (does my logic probe impose enough load to stop the oscillator from working?), I opened up my functioning VIC-20 (with the exact same PCB) and checked pin 13 of UB9. In this machine, the logic probe beeps along happily on that pin, indicating regular high/low changes of the oscillator circuit.

... now, just for curiousity, what happens with the working unit if you attach the logic probe to pins 1 or 4 of UB9? (Note: if the screen goes blank, remove the probe and switch off the VIC-20 immediately, as you've just put the VIC chip into an undefined state, see below).

You should also check the supply voltages. [...]

This is probably supposed to be the very first step of any diagnostic, right?

I'm using the same power supply that I use for my working CR, and it's a modern one, so I wouldn't expect power supply issues. Could faulty hardware introduce ripple? If so, where should I measure this on the board?

A dried up stabilizing capacitor in the supply rail can do that for sure.

First check should be the +5 V supply pin of the VIC itself. This is pin 40. If the DC voltage is outside 5.0 +/- 0.25 V, this calls for trouble. If this is much lower, there might be an internal short in the chip, which corresponds to your notion the chip gets exceedingly hot after short time. With the multimeter set to AC, measured on the same pin, the ripple should remain below said 0.1 V~.

The VIC might also get that hot because it doesn't get the correct oscillation signals. That surely puts many of its internal logic into undefined state which in course leads to higher power dissipation, especially with 'half-way through' output gates. For this reason you should try to replace UB9 first, then the quartz. If the VIC is socketed, remove it while you try to get a functioning oscillator circuit.

Good luck!

[chysn]

Having a functioning unit to cross-check is always a good idea ...

It also makes me a little nervous to take it apart and start poking around at it. Well... actually, the time I'm nervous is when powering it back up after putting it together again. That's a nail-biting four seconds. But I'm gaining more confidence, and I work carefully.

... now, just for curiousity, what happens with the working unit if you attach the logic probe to pins 1 or 4 of UB9? (Note: if the screen goes blank, remove the probe and switch off the VIC-20 immediately, as you've just put the VIC chip into an undefined state, see below).

On the functioning machine, pin 1 oscillates, and pin 4 is always low. The screen did not go blank at any time.

First check should be the +5 V supply pin of the VIC itself. This is pin 40. If the DC voltage is outside 5.0 +/- 0.25 V, this calls for trouble. If this is much lower, there might be an internal short in the chip, which corresponds to your notion the chip gets exceedingly hot after short time. With the multimeter set to AC, measured on the same pin, the ripple should remain below said 0.1 V~.

At pin 40 of the VIC, it's 4.791V D/C. A/C measures 0.000mV~.

The VIC might also get that hot because it doesn't get the correct oscillation signals. That surely puts many of its internal logic into undefined state which in course leads to higher power dissipation, especially with 'half-way through' output gates. For this reason you should try to replace UB9 first, then the quartz. If the VIC is socketed, remove it while you try to get a functioning oscillator circuit.

Thank you, that's the course I'll take. If that oscillator starts ticking, I can move to whatever the next issue might be.

My very first thought upon opening this VIC up is that the 100nF caps on this thing don't look right, not a single one of them, but they seem to work okay. My method of testing them is

(1) With the power supply unplugged
(2) Set my meter to Ohms
(3) Put the meter's leads at each terminal of the capacitor
(4) Watch the meter's behavior. If it measures resistance after a second or so, and this resistance falls gradually to 0, I consider the capacitor okay.

This may or may not be the best way to test a capacitor in place.

Here's what the caps look like. They all look physically worn, with discoloration near the bottom or center. Maybe it doesn't mean anything, but they're different than I'm used to seeing.

[chysn]

I have 2 'working' VICs, on 2 pin and a CR.

The 2pin has some sort of memory problem, ie wont run games then require certain memory configurations. (doom/cheese n onion)

The CR crashes after about 20 to 45 minutes, even if its just left idling. I have a spare power supply, one seems to last better than the other.

I would love to be able to fix them myself if I had the knowledge, does anyone know of guide to repair VICs or maybe a service manual.
Maybe we could have a thread on here with know issues, something like a denial workshop manual?

The challenge is that it's hard to generalize information, given the number of things that can fail, partially fail, or cause cascading failures in other places. The information is scattered around, but these are some of the resources I've found particularly valuable ("valuable" at least in the abstract sense, as I'm still working on my dead VIC):

COMMODORE VIC20 COMPUTER - DIAGNOSTICS AND REPAIR http://personalpages.tds.net/~rcarlsen/ ... /vic20.txt

This lists the components of the VIC-20, and describes possible symptoms of each component failing completely or partially. The text has some good insights, as well.

Board Schematics https://myoldcomputer.nl/technical-info ... vic-20-cr/

This is only for the CR, which is what I'm working on. But a similar schematic for the original board is at the back of the Programmer's Reference Guide. I've printed these images out for easy reference. It usually doesn't take much Googling to figure out what abbreviations mean.

8-Bit Guy's The Worst VIC-20 Ever, Part 2 https://www.youtube.com/watch?v=_rjjCkcXknE&t=635s

This video goes through a troubleshooting process from beginning to end (spoiler alert, it's a bad RAM chip). Even if you have a different problem, just seeing the steps he takes is life-changing. For example, he shows how to test the 555 timer and rule that out as a problem. You'll also learn what many of the chips do, and how they're related to each other through the PCB traces, and there's key information about how to apply tools like a meter and a logic probe. 15 minutes well-spent. Part 1 is interesting, but focuses on the outside of the VIC.

[crusti]

Thank you very much Chrysn, most helpful.
You are an awesome dude with ya helpfulness.

[chysn]

I've ordered parts (oscillator, 7402, socket for 7402), which should be here Wednesday. It's times like this that I really miss Radio Shack. On Wednesday after work, I should have time to test the parts on a breadboard, and then I'll do the repair next weekend. I think the possible outcomes are

(1) The oscillator starts oscillating and everything is great. Then I have another VIC.
(2) The oscillator starts oscillating and there's another problem. Back to the lab, looking at other systems.
(3) The oscillator doesn't work, and I look for bad components within the oscillator

[chysn]

This project was at a standstill for a little while. While I was desoldering a capacitor, I dropped my desoldering tool and broke it. So I was out of the desoldering game for a bit. It was one of those little sucker cylinder things with a plastic tip. I never liked it because I had to hold the iron in one hand while trying to put the desoldering tool's tip in the right place. The thing sucked. And by "sucked," I mean that it didn't suck solder very well.

So, it was lucky that I dropped it and broke it, because what I replaced it with is awesome. It's basically a soldering iron with a suction bulb attached to it. Before risking my VIC-20 board, I practiced using this thing on a C-64 board, and it was easy. I had a 24-pin character ROM chip out in like ten minutes, without damaging the chip or the board. It's amazing what the right tools can do, and have a lot of confidence that I'll be able to get the quad NOR chip out in no time.

Speaking of "the right tools," the other thing that happened is that it was my birthday recently. My wife bought me a Weller digital soldering station. For years and years, I used a regular soldering iron. It was a good one, and I was always a soldering station skeptic. Everyone said, "Get a good soldering station," and I was all, "How could it be better than my trusty Weller iron?" But I don't say that, not any more. Being able to control the temperature is no joke. Having the thing go from off to the correct temperature in about a minute is no joke. The second I assembled a wAx cartridge with the thing, I was hooked. So fast, so precise. I use 0-ohm surface mount resistors as jumpers on the cartridge PCB, and this station makes SMD about a million times easier.

Now that I'm decked out with good stuff, all I need to find is time. I temporarily got sidetracked with a software project, but I'd like to get this VIC back on its feet.

[chysn]

So, I did the following:

• Removed the quad NOR, installed a socket, and put in a 74AHCT02
  Removed and replaced the six electrolytic capacitors

I was fully expecting to have to move onto the next diagnostic task, but no. It's alive!

Post-fix testing included loading from SD2IEC, checking sound, checking the joystick port, checking a game cartridge, checking memory expansion. Everything seems to be in working order once there was a ticking clock.

So, thanks, everyone, for your input on this. It was a lot of help. I'd like to go through this process again someday. The problem is, you see too many VICs on ebay as "Untested, buy it now for $150."

[Mike]

It's alive!

[chysn]

Well, things are promising based on this VIC working, but sometimes it looks like this on startup:

The pattern is always a bit different. A few of the pixels go on and off, but the image is otherwise stable. There's no cursor and the machine is unresponsive in this state.

When the machine starts up normally, it works perfectly, as described above. So I think the fact that this is intermittent rules out character ROM, RAM, the VIC chip, the processor, etc.

My guess is maybe a cold solder joint somewhere, possibly affecting a data line. While replacing the caps, I did find one cold joint and corrected it. But I think I'll have to give the whole board a close inspection. Now that I know there's a working VIC-20 here, I don't mind putting in additional time.

More to come!

[chysn]

After going over the board with a magnifying glass and a tactical flashlight, and finding no additional broken joints, I figured I'd have to get to checking a bunch of data lines. But, no. I pulled the 6502 and put it back in, and it started up normally. It seems to be reliable now. I'm going to keep it running, maybe overnight, and see if it's solid.

If it proves to be solid overnight, I'm going to run it as my main VIC-20 for a while, and give my other CR a bit of a break. I may establish a three-VIC semi-annual rotation, so that one doesn't bear all my usage while two sit in storage tubs waiting for a failure.

[mathom]

That's awesome. Fingers crossed that you wake up to a happy VIC!

Thank you for this thread. Reading/watching other people's troubleshooting journeys makes me want to someday try my hand at it if the right VIC-20 comes along at the right price. I do have a Tandy model 102 that needs some attention after being stored away for too many years. We'll see how that goes.