Re: Non-working Commodore PET 3016
 

Well many vintage RAM's don't often seem to need much for them to fail (especially multi-supply, but also some single-rail DRAM's in Spectums etc) - Maybe getting a bit too warm for long periods, or just down to original manufacturing technology (like quite a few Commodore custom IC's in C64's and earlier / later ones)

Although I'd normally suspect a power supply surge (which is what I'd presumed had taken out that many IC's) - As with the rather-low sourcing currents of LS TTL IC's, I wouldn't expect them to fail even if their output is connected to ground (either accidentally, or via a conflict with another one driving low).
- However, shorting an LS TTL output to 5V, when it is trying to sink current, could well damage one.
I don't recall ever accidentally blowing up any TTL, despite often doing bad things on running circuits like connecting to whilst powered and shorting things. And very rarely had to change one, when fixing various computers etc. over many years (unlike DRAM's and the odd processor / ULA / ROM etc.)

Hopefully PET's don't suffer from Regulators going short
- that it seems occurred on some C64 PSU's and has had people adding over-voltage crowbar circuits onto their main boards.

Re: Non-working Commodore PET 3016
 

I'm glad e never got there because sourcing them looked like it was going to be difficult/time consuming/expensive.

I'll keep popping the odd update on this thread for those who are interested about any progress. the keyboard is out, so that's my first place to start.

Colin.

Re: Non-working Commodore PET 3016
 

Thank you - much appreciated. All I did was ask others for help and they've all been so generous with their time.

Colin.

Re: Non-working Commodore PET 3016
 

It passes - couple of screen shots for those who are interested.

Colin.

Re: Non-working Commodore PET 3016
 

About that: You mentioned that some keys do work, and by that I trust you mean that when you press one of those keys the correct character appears. When you press the other keys, nothing happens?

If you want to check the keyboard scanning hardware independently of the keyboard itself you can unplug the keyboard and 'type' characters by joining keyboard row and column pins together with a jumper lead. We can probably find more detail on this if you need it, because in John Earland's original PET thread his original problem was a fault on the key scanning hardware (not the keyboard) so we accumulated quite a bit of info about the keyboard connector in that thread.

Also, in the Daver2 test screen you've just posted do you see that line 'Kbd' followed by a row of 00s? I think when you press keys those numbers should change.

I would say that the keyboard is the next essential repair because once that's working you can write small BASIC programs to test other aspects of the machine such as the IEEE interface, which can be tested using a few PEEKs and POKEs from BASIC.

Re: Non-working Commodore PET 3016
 

I think the keyboard needed a clean.

Colin.

Re: Non-working Commodore PET 3016
 

You probably could test the keyboard switches for continuity while it is unplugged, squirt a little Servisol Super 10 down the plungers of any dodgy ones, give them the "track and field" treatment and leave the keyboard assembly in a still-warm but unlit oven to drive off the last of the solvent vapours.

If no keys work at all on the computer, despite testing OK, suspect UC9 (74LS145) first.

Re: Non-working Commodore PET 3016
 

If I recall correctly the keyboard has a PCB with interleaved finger tracks on, which are bridged by plungers with conductive rubber pads on. Dust can easily stop the pads from making good contact. Cleaning them and the PCB very gently with isopropyl alcohol is the usually recommended procedure.

Re: Non-working Commodore PET 3016
 

What Colin actually said:

It is possible for there to be a key row or key column out due to a hardware fault but by the sound of it he has so many out that this is going to be a 'button thing' rather than a 'circuit thing'. And did you see the state of that keyboard? :o

Re: Non-working Commodore PET 3016
 

This video by the 8-bit Guy shows a PET with problems including keyboard problems and includes an excellent description of how the keyboard works - starting at 2:41 in the video.

https://www.youtube.com/watch?v=uHbhH7ISL_Y

Re: Non-working Commodore PET 3016
 

OK - all keys, springs etc removed and thoroughly cleaned and dried. Keyboard PCB cleaned with 99& IPA (not the beer).

All back together and it doesn't half look nice (the PET keyboard could withstand a nuclear war I reckon). Most keys work and the ones that don't will get a layer of conductive paint in the next day or two on the pads that press onto the keyboard PCB.

However, most of the keys seem to give me the letter to the right. For example, 6 gives me *, J gives me K.

Where do I start with keyboard scanning?

Colin.

Re: Non-working Commodore PET 3016
 

Have a look at the 8-bit Guy video I mentioned in #1510 - 2.41 on, then come back.

Re: Non-working Commodore PET 3016
 

Yes, I recall removing the keyboard's PCB with it's matrix of (gold-plated rather than a bit resistive conductive-carbon?) (with loads of tiny screws holding it) on a 3000/4000? series I used to have to fix the keyboard's non working keys.
And I just did what I do on remote controls units, cleaning the PCB with alcohol cleaning solution as well as the conductive rubber parts on the keys - Did you try cleaning those as well?
If all keys had worked when it was last working, then cleaning should be OK without needing to add extra conductivity to the rubber pads.
- Watching that Youtube video, I see he only checks the keys on a DMM (first on its continuity beep range, which often needs < 200R to sound), and after cleaning them he then actually measured resistance of around 1k on each
But he doesn't seem to check if that is actually low-enough for the computer to read them OK, before he silver paints them.
(I reckon it may well be, as many remote keypads also used carbon PCB pads / tracks, that were bit resistive)
So it depends on what pull-down / up resistor values are fitted on the 'output side' lines.


Getting incorrect key-press characters returned does seem a little unusual - Hopefully not one of the PIA / VIA's.
But a study of it's matrix schematic may help in working out what might be happening.
And I'm wondering if you may have the connector accidentally plugged in one-position out to one side?
Or maybe plugged in the wrong way round? (Doesn't look like it has a polarity key)

Re: Non-working Commodore PET 3016
 

Our posts crossed - thanks; I will take a look.

Colin.

Re: Non-working Commodore PET 3016
 

I cleaned all the conductive pads on the bottom of the keys with IPA as well. There is a key on the keyboard ribbon, so it's not that.

My plan is to watch the video and see what information I can take from that, and then to go back to basics on the mechanicals of the keyboard by stripping it down and see what happens when I get a signal going directly across the gold-plated traces one by one.

I'll be back.

Colin.

Re: Non-working Commodore PET 3016
 

It had looked like on his video that the main PCB just had a row of complete pins, so looked like it was possible to mis-align it. But maybe they changed the design a bit (seems there were several keyboard, with different connections)

If all the keys that don't work are in one row / column, then it could well be an issue with the scanning not being correct (all shifted to right) rather than the keys themselves.
(Anyone who's had the joys of Spectrum & ZX81 keyboard membranes, will know how easily the silver-paint on the ribbon tails fractured, loosing a line of 5 keys or one from each of the 8 lines)

Re: Non-working Commodore PET 3016
 

If all the keys are offset one to the right then -

Check you haven't just plugged the keyboard connector in one pin position out (is that even possible?)

Otherwise, the most likely culprit is the 74LS145, UC9.

Let us know when you're ready for some more checks.

Re: Non-working Commodore PET 3016
 

Well, colour me shocked. I would never have expected PET keyboards to have been made using anything but discrete switches.

Anyway, if you are getting keypresses misinterpreted consistently as neighbouring keys, that is almost certainly an indication of a multi-way connector being displaced by one pin (the keyboard is scanned as a matrix of switches; one of the "row" lines at a time is pulled low, and any key depressed in that row pulls its "column" line low. Note that rows and columns here probably do not correspond exactly with the rows and columns of the keyboard); so check all wiring carefully.

Re: Non-working Commodore PET 3016
 

Yes, I had suggested that, as it looked to me like it was possible to have connector plugged-in incorrectly, from that video. But Colin reckoned his was keyed and was correct.

Just looked on the main board's schematics, and see they have 10k Pull-ups to +5V and 6520 has Low-level input voltage spec of < 0.8V.
So as long as the resistance of the pressed-keys is < 2k they should be OK.
- On that video, he got about 1k, after cleaning, But he thought they should activate DMM's continuity check (< 200R?) and hadn't tried it on the computer before first (now that they'd all been cleaned), before he decided to silver-paint them all. So it seems he'd been expecting that carbon-coated / impregnated conductive-rubber should have similar resistance to metal!

Agreed that the 74LS145 (a bit better having that fail than the rarer 6520) could well be responsible for some odd scanning problems.
Especially as it's another TTL one - And no doubt one he hasn't got a spare of!

Re: Non-working Commodore PET 3016
 

Commodore were rather keen on this approach - also used on the VIC20 & C64 etc. But not sure what the (literally from a calculator) 'Chicklet' original was like (But do have one, that no doubt needs some,work, when I manage to dig out an old PET I've got with that on it to fix one day)

Colin's assured us the connector is fitted all OK.
But the keyboard scanning on these is a bit more involved with a 74LS145 4bit BCD to Decimal decoder from the 6520 PIA to the keyboard's scan lines. So knowing how many TTL IC's had failed, that could well be a prime suspect to do some 'scoping on.