Commodore PET 2001-32N with few signs of life

[rosshamilton29]

Hello everyone, I'm a first time poster here although I've been reading other PET repair threads on this forum weeks.

I visited the Centre for Computing History in Cambridge ( highly recommended - https://www.computinghistory.org.uk/ ) and had a great time playing around with an old Pet 3016 ( 20250308_135539.jpg below ), it made me very nostalgic so I hunted around on Facebook MarketPlace and eBay for one. I particularly wanted one of the ones with a "proper" keyboard and the pyramid shape - this was what I first started serious programming on back around 1980 when one arrived in the class in my final year of primary school.

I found one soon after - but unsurprisingly it appears to be completely dead. So I started trying to bring it to life...

I started by pulling out the motherboard and giving it a clean up with Isopropanol Alcohol 99.99% - not very thorough, to be honest, I was impatient! I might have a second go at this soon.

I spent a few days reading everything that I could find about PET repair. This forum was one of the best places, as was the blog and Patreon pages by Dave Curran (Tynemouth Software - http://blog.tynemouthsoftware.co.uk/ ). He also has a Discord where I'm documenting this repair log and getting tips from him plus others. I've also bought a couple of things from him to help - his composite video out board, and his ROM/RAM replacement board.

While on a roll with buying things, I picked up a cheap Chinese DSO138 digital oscilloscope (well, 3 of them, for daft reasons involving AliExpress descriptions). I also picked up a nice Duratool D006872 desolder station, and a vintage Gould OS4000 oscilloscope too. I like projects such as this one, that give me excuses to pick up tools that I've wanted for a while

I refitted the board, disconnected the monitor and keyboard, and started testing. On power on you get a hum from the transformer, and the 5v and -5v rails on the board were good, but there was no 12v - it measured a dead short. Dave suggested replacing all the the tantalum caps, so I did that and the 12v sprang back to life. But still nothing on screen. Received wisdom is that it should generate a screen signal even with the CPU. PIAs and ROMs pulled, so I did that but still nothing.

I think that I may have headed off in the wrong direction now - I started scoping the 6502 lines, checking that the clock was present (it is, I think, see scope trace). I haven't used a scope since the late 1980s in physics lessons at school, so I'm on a bit of a catch-up learning curve!

I'm thinking that I perhaps ought to work out the display problem - forgetting about the CRT for now and concentrating on the video out to the composite monitor. But I'm not sure where to start. Any tips? I have printed out all the relevant schematics from here: https://www.zimmers.net/anonftp/pub/cbm/schematics/computers/pet/ ... but I'm not experienced enough with this stuff to really know what to do next. Any tips welcome!

[Mark1960]

Check that the reset input on the 6502 is low for a short time and then high, and take a look at the address lines out of the 6502 to see if there is any activity.

Check the clock input to the shift registers and counters on sheet 6 master timing.

[TonyDuell]

These early PETs don't use the 6845 CRT Controller chip. Rather the video circuitry is built from hardwired TTL chips, so as you've been told it should work even without the CPU (6502), ROMs, PIAs, etc.

[Later PETs used the 6845. This has to have numbers written into its internal registers for it to do the right things and thus for the video side to work, meaning the CPU, ROMs, etc have to be functional].

I'd also check the horizontal and vertical drive signals (on the user port and/or monitor wiring connector on the mainboard). If those are doing something then most of the video counter timing chain must be active.

One of the most useful bits of test equipment at this stage is a simple logic probe. It'll tell you if a signal is stuck high or low or if it's changing state. (The ultimate logic proble I think was the HP LogicDart but that was very expensive new and not too many people bought them. Do not buy a second-hand one without the probes -- there's a lot of electronics in the moulded connectors, they are not just bits of wire, and the instrument is useless without them).

[SiriusHardware]

hi rosshamilton29, welcome to the forum.

As a new forum member here your posts will be held for moderation for a while but that will wear off after you have made a few posts here. It can sometimes lead to people confusingly posing questions which you've already answered in a post which wasn't visible to them at the time, but it all sorts itself out eventually.

As Tony explained these older PETs have a certain simplicity about them compared to the ones which use CRT controller ICs and the video output circuit is almost autonomous, running quite independently of the CPU system - so you were either lucky or sensible to pick one of these as your first PET to have a go at.

As we don't 'know' you yet, we don't really know how concise or alternatively, detailed, to make any suggestions we may have for you so if if you find that we are either saying things which don't mean much, or over-explaining things which are already obvious to you, please say so and we will shift up or down a gear as needed.

I see you have an adaptor PCB to convert the output from the video circuit to something more suitable for a standard video monitor - that's fine but initially, as per Tony's suggestion you just need to scope the V_Drive and H_Drive signals on the original monitor output connector. If you don't have signals there, you can put the composite adaptor PCB to one side for the time being until we find out why.

As Mark suggested, follow the clocks through the master timing circuit and see how far the signals get - again if you need us to be more specific (If you prefer us to suggest specific test points and what to look for there) please let us know.

[ortek_service]

I notice this board has (all?) white IC sockets, which if are the ones originally fitted by Commodore can often be notoriously troublesome and give (often intermittent) connection issues that may not be solved with contact-cleaner on them.
Although I didn't think Commodore usually socket-ed that many IC's
- Maybe to start-with on earlier boards? but this looks like it's got more-standard 4116 DRAM's? rather than the early Commodore(MOS) special non-standard pinout wider DIL-24pin SRAM's which are more-difficult to find replacements for (Hence the RAM-ROM replacement board as a workaround).

However, I don't think on these non-CRTC IC boards that any of the 24pin (EP)ROM's or the 40pin PIA's / CPU etc. would be required to have the video circuitry running - So should probably still produce some output with all of these removed.
If the RAM IC's are 4116's then they will need all 3 supply voltages present / may be possible that some have been damaged by lack of one of these supplies if they are not all present together.
However I believe the video RAM was separate and a couple of 2114 SRAM's - which also seem to have a bit of a reputation of being prone to failure even with their single +5V supply being OK.

[rosshamilton29]

Thanks everyone for the welcome and the tips. My experience level is a bit all over the place - I'm very happy building big things with discrete components with a soldering iron, my biggest projects so far have been related to converting old pianos to play themselves and making a sensor bar using optical devices to sit under a real piano keyboard and turn it into a MIDI device (you can see details here: https://www.youtube.com/@electricplayerpianos ). I've done most of the fabrication work, but mostly just watched over his shoulder and made suggestions while the PCB designs were put together by my friend and work colleague Dave Langley ( https://www.robotron-2084.co.uk/ ) who is already well known in the arcade/pinball scenes. He's giving me tips on this PET too, but I really want to solve this one without asking him too much, so I can go up in his estimation

I didn't get a chance to come back to this last night, I probably will tonight...

@Mark1960 - the reset circuit - that was my current plan for next things to try.

@TonyDuell - the iveo drive signals - I was coming to them next - Dave Curran showed me a couple of example scope traces so I know roughly what to look for. My oscilloscope experience though consists of a few lessons in Physics/Electronics classes 40 years ago, and playing around a few evenings in the last week! I did build that cheap Chinese scope from parts though, so I feel like an expert now, having made my own scope

@SiriusHardware - thanks for the welcome and don't worry about the level you pitch your replies. Sometimes you'll be telling me things I already know, often you won't... but for the next person who comes along, the more detail in these threads the better!

@ortek_service - in theory Dave Curran's ROM/RAM replacement board overrides all of them, except the video RAM I believe, so at the moment they're all pulled. The RAMs were missing the 12v line, but I've fixed that already by replacing the bad capacitors (I found 2 bad ones, but replaced all the large tantalum caps on the board at the same time).

[SiriusHardware]

You're obviously pretty comfortable with circuit diagrams, identification of parts, use of your multimeter etc. So that's all good.

Regarding the reset circuit, the circuit centred on the 555 timer (bottom left corner on sheet #1 of the circuit diagrams) is designed to hold the CPU in reset for a short time immediately after power-on before allowing the CPU to run after the supplies have stabilised.

To see that working you will need to scope the RESET pin of the CPU while switching the machine on, or alternatively, you can power the machine on and trigger the reset circuit manually by momentarily shorting out C68 - every time you do that you should see the CPU RESET pin blip low, then return high again.

[SiriusHardware]

If your reset circuit looks OK, have a look for some clocks and a couple of other signals.

Try looking for:

Waveforms on pins 7 and 39 of the 6502

8MHz squarewave on UG5 (74191) pin 3
1MHz squarewave on UG5 (74191) pin 7

1MHz waveforms on all of the following pins of UH3 (74164)
Pins 3, 4, 5, 6, 11, 12, 13

1MHz waveforms on the following pins of UH2 (74S04)
Pins 10, 4, 12, 2

A waveform somewhere in the region of 15KHz to 20 KHz on UH7 (74LS107) pin 2
A waveform of around 50Hz-60Hz on UG10 (74LS00) pin 11

Waveforms on UG1 (74LS08) pins 8 and 11
Waveforms on UG7 (74LS10) pins 6 and 12

Initially you are just looking for the presence or absence of signals at these points but if you have the means to do so please also confirm the waveform frequencies at each point, and note also that all of these signals are expected to be clean, square TTL logic signals so if any of them are absent - or present but suspiciously small or distorted compared to other similar signals, that would be what we call a 'measurement of interest'.

If you find a missing or dodgy looking signal on any of these points, please report back.

[rosshamilton29]

Hi everyone!

I've had a good look at the RESET circuit and it seems to be working correctly. I get 0v for 1-2 seconds after power-on and then it jumps to a steady 5v.

I tried to take some readings on the J7 connector that goes to the monitor. I *think* the pinouts are:

Pin 1 - Video
Pin 3 - Vertical
Pin 5 - Horizontal

I've uploaded the scope images I could get at those points, but I'm not sure if I am using the scope properly! I'm going to read some documentation and watch some videos on youtube.

With being a bit rubbish at reading the scope, I pulled out my logic probe and looked for pulses at all the points that SiriusHardware suggested... a YES below means that my probe shows pulses, I can't say if they're nice square waves, or what frequency. My next aim is to to learn how to do that with my scope...

---

Waveforms on pins 7 and 39 of the 6502 - YES

8MHz squarewave on UG5 (74191) pin 3 - YES
1MHz squarewave on UG5 (74191) pin 7 - YES

1MHz waveforms on all of the following pins of UH3 (74164)
Pins 3, 4, 5, 6, 11, 12, 13 - YES

1MHz waveforms on the following pins of UH2 (74S04)
Pins 10, 4, 12, 2 - YES

A waveform somewhere in the region of 15KHz to 20 KHz on UH7 (74LS107) pin 2 - YES

A waveform of around 50Hz-60Hz on UG10 (74LS00) pin 11 - YES

Waveforms on UG1 (74LS08) pins 8 and 11 - YES
Waveforms on UG7 (74LS10) pins 6 and 12 - NO, but I did get them on 5 and 12, could the suggestion to check 6 be a mistake? I didn't try to find it in the schematics yet.

[SiriusHardware]

OK - only got to this very late (nearly two in the morning as I write this) - can you say more precisely what the frequencies on UH7 pin 2 and UG10 pin 11 are please? If we can establish for sure that the H_Drive and V_Drive signals are correct then we can have an early stab at getting something to appear on the monitor, which will be a great help for further diagnosis.

The monitor on this machine is which, the 9" or 12"? If the 12", and if it has the most common monitor PCB then check two resistors R752 and R753 (both should be 56R) - they are such a common failure (open circuit) that they are worth an immediate check over. If it's the 9" then none of that applies.

UG7 pin 6 - under normal circumstances I would expect to see activity there but it may be that due to current circumstances one of the RAM banks is never being selected so we'll park that and may eventually come back to it.

Regarding frequency measurement, if you don't have an actual frequency meter, your multimeter may have a frequency measurement or 'HZ' range on it, or frequency measurement may possibly be one of the functions of your Chinese digital scopes. It is possible to roughly measure frequency using a basic scope but we will wait to hear if you have any more conventional means of measuring frequency first.

[SiriusHardware]

Just looking back at the photos I'm guessing the monitor in this case is the 9" version - but if you can determine the frequencies of the signals at UH7 pin 2 and UG10 pin 11 then we may (or may not) divert onto the monitor for a while.

[ortek_service]

Originally Posted by SiriusHardware

>>
>>

Regarding frequency measurement, if you don't have an actual frequency meter, your multimeter may have a frequency measurement or 'HZ' range on it, or frequency measurement may possibly be one of the functions of your Chinese digital scopes. It is possible to roughly measure frequency using a basic scope but we will wait to hear if you have any more conventional means of measuring frequency first.

On the last of those Chinese TFT 'scope modules, the displays shows 10us in green, that would normally be 10us per division. However, there is a kind of cursors line in green spanning several divisions at the top of the screen so not clear whether that width is actually 10us.

And when using the Gould OS4000 D.S.O. (10MHz Analogue B/W but rather-less sample-rate in storage mode), these being rather-vintage lack on-screen display. So it would be useful to know what the Timebase is set to by adding this to picture filenames or stating this.
BTW, if the OP hasn't got a Manual for it, then there's a copy of this here: https://www.kevinchant.com/uploads/7/1/0/8/7108231/os4000manual.pdf

I would suggest using 20us/div for Horizontal drive measurement - Where you'd normally expect a 64us long Sawtooth waveform to scan this at UK standard 15.625kHz line rate.
And using 10ms / div for Vertical drive measurement - Where you'd normally expect a 20ms long Sawtooth waveform to scan this at UK standard 50Hz field rate.

I presume the service manual shows the expected waveforms ?

However , From what I can see, there is no vertical-drive or video. And the Horizontal drive looks very-odd, with much higher frequency bursts of pulses. So this doesn't look to be correct.

[rosshamilton29]

Yes, it's the 9" monitor in these. I'll have a better go with the scope and come back with results.

I bought Dave Curran's video to composite adaptor board and a tiny composite monitor (a reversing camera), so that I could try to get the motherboard going before I tackled the CRT, rather than both at once. I tested the tiny monitor and it worked fine on a composite signal from a DVD player... but at present it just stays blank when connected in the PET.

[rosshamilton29]

Looks like I counted the pins from the wrong end... so my 1, 3, 5 should have been 3, 5, 7. I'll double check though!

The service manual does show CRT waveforms, but as I said above, unless you think it's a mistake to do so I'd be happier trying to get the PET working with video out via the composite adaptor rather than trying to fix both the motherboard and the monitor at once...

[SiriusHardware]

Yes, sure, it's just that your initial report seemed to say that you thought you had healthy looking V_Drive and H_Drive signals so I thought it was worth seeing if those signals would wake up the monitor, or if we could work out why they were not waking it up. After Owen's less sleepy observations, I think we had better try again to re-measure those waveforms.

Let's establish a rule for when you are posting screenshots from your scopes:-

Give each image a detailed filename which includes the test point it was taken from, and the Volts/Division setting and the Time/Division settings which were in use at the time. A typical (completely made up) example might be

UG10_Pin11_1.0V_10mS.jpg

This will make it clearer to us where and with what settings each measurement was taken.

Incidentally the signals I asked you to look at on UH7 pin 10 and UG10 pin 11 are the H_Drive and V_Drive signals respectively, I just asked you to measure them at the chips rather than the connector because I anticipated that you might not know which end of the monitor-out connecter was pin 1.

[ortek_service]

Originally Posted by rosshamilton29

Looks like I counted the pins from the wrong end... so my 1, 3, 5 should have been 3, 5, 7. I'll double check though!

The service manual does show CRT waveforms, but as I said above, unless you think it's a mistake to do so I'd be happier trying to get the PET working with video out via the composite adaptor rather than trying to fix both the motherboard and the monitor at once...

I was thinking more of the waveforms of the H & V drive signals from the Mainboard to the inputs of the CRT Monitor board.
If the manual does shows these, then you won't need to have the monitor powered-up in order to check that these are present and correct, before trying to get that going.
And as you have Dave Curran's Video-output board, then you shouldn't need to worry about getting the PET's monitor going for now (Although when getting round to that, it seems it's mostly some resistors in-series with supply rails / electrolytic capacitors on these breaking down that are at fault with some commonly to blame suspects to check / prepare to replace).

[rosshamilton29]

Thanks very much for all the help!

I've worked back through the comments above, I think I've got trace photos for all of the questions. I can't upload this many at once to this forum, so I've uploaded them here: ******************************TeYESxkJjkrQt5zm9

In order they are:

1. GND reference
2. 5V reference from 4116
3. Clock at 6502_Pin37_1.0v_10uS
4. Horz Sync J7_Pin3_1.0v_10uS
5. Vert Sync J7_Pin7_1.0v_10uS
6. 6502_Pin7_1.0v_10uS
7. 6502_Pin7_1.0v_10uS
7. Video J7_Pin5_1.0v_20uS
8. Open PET reference shot
9. 6502_Pin7_1.0v_10uS
10. 6502_Pin39_1.0v_10uS
11. 6502_Pin37_1.0v_10uS
12. UG5_Pin3_1.0v_1uS
13. UG5_Pin7_1.0v_5uS
14. UH3_Pin3_v1.0_5uS
15. UH3_Pin4_v1.0_5uS
16. UH3_Pin5_v1.0_5uS
17. UH3_Pin6_v1.0_5uS
18. UH3_Pin11_v1.0_5uS
19. UH3_Pin12_v1.0_5uS
20. UH3_Pin13_v1.0_5uS
21. IPA reference shot
22. UH2_Pin10_1.0v_5uS
23. UH2_Pin4_1.0v_5uS
24. UH2_Pin12_1.0v_5uS
25. UH2_Pin2_1.0v_5uS
26. Scope reference shot
27. UH7_Pin2_1.0v_200uS
28. UG10_Pin11_1.0v_200uS
29. UG1_Pin8_1.0v_200uS
30. UG1_Pin11_1.0v_5uS - hard to stabilise
30. UG1_Pin11_1.0v_5uS - trying for better stability
32. UG7_Pin6_1.0v_1mS
33. UH7_Pin12_1.0v_200uS

I guess something is going wrong in the tail end of the video driver circuitry - I get good signals a long way through the chain but the H and V are bad by the end...

[rosshamilton29]

Hmm, it looks like Google Photos was a bad idea, you can't easily see the file names, or even keep them in the right order. I'll put them in this Google Drive folder instead.

I've replaced a few pics with better ones - I'll go back and edit my last couple of posts accordingly.

[ajgriff]

Access denied. I think you need to change the permissions at your end.

Alan

[SiriusHardware]

Uh, gents, I think this site has a policy of not allowing links to off-site photo repositories - mainly because they tend to go dead. There is a limit of 5 attached photos per forum post but no limit within reason on the number of posts.

Another workaround which also gets around the forum auto downsizing images to 800 x600 is to put the images into a .zip, but note there is also an upper limit (per post) on the size of a .zip attachment.