Re: Non-working Commodore PET 3016
Quote:
http://www.zimmers.net/anonftp/pub/c...N/320349-1.gif Is that right? Not sure it can be as there's nothing plugged into J4. |
Re: Non-working Commodore PET 3016
#196 looks like an original nmos 6502 to me.
RDY held high and the 1MHz clock on pin 37, 39 and 3. Reset on pin 40 should be low for a short time and then stays high. Then you should have activity on SYNC pin 7 and on the address lines. I think its still a good idea to check each of those pins again using the scope. Another thought is that even if the 6502 is not running, there should probably still be some output visible on the display, but would show garbage. Again that probably needs someone with experience with the pet to confirm. |
Re: Non-working Commodore PET 3016
Quote:
The resistor in question is R14, 1K, can you just visually inspect that resistor and also (with power off), use your old meter to measure the resistance of that resistor in-circuit? Also check that you have continuity (zero ohms) between the 6502 pin 2 and one end of that resistor? |
Re: Non-working Commodore PET 3016
fyi I have a copy of "The PET revealed" by Nick Hampshire which has all the circuit diagrams in Appendix A. I should have looked at that before...
|
Re: Non-working Commodore PET 3016
Quote:
|
Re: Non-working Commodore PET 3016
Do I take it that RDY is an input only? It just has one 1K pullup resistor and the rest of the track heads off to one of the connections on the expansion connector and apparently no-where else.
The track between R14 and the CPU pin 2 is quite long and on the top side so it will be worth looking to see if it, or the connection between CPU pin 2 and that track, are open circuit. Next thing would be to see if pin 2 has a low resistance to 0V with power off and / or whether the voltage on that socket pin comes up to +5V with the CPU removed. SYNC goes nowhere except to the expansion connector. |
Re: Non-working Commodore PET 3016
Ready is similar to nwait on the 8060 or z80, but only extends memory read cycles on the original nmos 6502. Colin verified its held high so shouldn’t be stopping the 6502 from running.
Checking sync output was just another possible check if the 6502 was doing something, should be high during instruction fetch, but Colin shows no activity and stays high. Reset is high, so shouldn’t be stopping the 6502 from running. It would be good to verify that it is low for a short time after power up, just in case the 6502 is in some unknown state. Would also be worth checking the circuit that drives reset, see if there is an easy way to force reset low without risk of damage by grounding the output of the gate driving reset. Seems colin verified clock input and output from the 6502, so it should be running and accessing memory but seems like it isn’t so starting to suspect the 6502. Still worth looking over the signals with a scope to confirm what we think the multimeter is telling us. Might still be worth looking at the video timing circuit to see if there are any problems there too. |
Re: Non-working Commodore PET 3016
I must have been tired last night, read the SYNC and RDY voltages the wrong way around so Colin, ignore #203 and #206.
Agree with Mark that if you can use your scope to verify that there is / is not activity on at least pins 9, 10, 11 of the CPU we can go forward from there. If you had a spare CPU we would ask you just to try that in, but I don't think you do so we'll proceed by other means. |
Re: Non-working Commodore PET 3016
Whilst I get on with the testing (which may take me a day or two to get used to the scope) it seems it might be a good idea to buy a spare 6052 to rule that out.
I've googled but there seem to be many 6502 variants. Do either of you know somewhere that sells the 'right' 6502? Thanks. |
Re: Non-working Commodore PET 3016
I have a saying, 'Always try the easy / cheap thing first'. If you can wait for a day or two until you are a bit more familiar with your scope, you may be able to rule out the need to buy another 6502 and save some money. If you just like the idea of having a spare, by all means go ahead anyway.
|
Re: Non-working Commodore PET 3016
post #203
Continuity confirmed between the left end of R14 and Pin2 of the 6502 Resistance measured with the old meter set at 2K - 0.945 |
Re: Non-working Commodore PET 3016
That's useful to have confirmed but I did say earlier that it was not necessary to make the measurements in #203 and #206 - a bit of brain-fade on my part, I thought you had measured a low voltage on RDY, not the high voltage you actually measured.
Focus now on using your scope to look first at the 1MHz output on UG5 pin 7 (which we are sure is there, so it's a good practice target) and then move to 6502 pins 9, 10, 11 (at least) to see if you have any activity on those pins. |
Re: Non-working Commodore PET 3016
OK - link to the first scope tests attached. I've put them into a doc so you can see the screenshots.
I have set the scope to Auto each time to try to get the best view of the waveforms. I guess I need to know whether they look good or awful - they're certainly not square. I can get no reading at all from pins 9,10 and 11 of the 6502. I used the resistors we pinned to UG4 for each of the tests. Was that right for the 6502 tests? https://drive.google.com/file/d/1H9z...ew?usp=sharing Thanks. |
Re: Non-working Commodore PET 3016
For scope tests you need to to have the earth side (croc clip) of the scope probe lead to a known 0V point. The resistor lead going to UG4 pin 8 pad will be OK for this but don't let the croc lead clip short to anything else (like the leg of the other resistor).
The 2.5V centre-point you made with those resistors was for the benefit of the 'frequency' feature on your new meter only. For all other voltage and scope measurements use a known 0V point as ground. I'll take a look at the images when I get home. |
Re: Non-working Commodore PET 3016
Understood. I have disconnected the resistors from each other and connected the clip to the pin 8 side of UG4 (hope that makes sense) as per your pst.
Please use this link: https://drive.google.com/file/d/1Q77...ew?usp=sharing |
Re: Non-working Commodore PET 3016
I know little about PETs but now a 'scope is available is it worth verifying that the three video signals are present (or not) on the monitor's board? If Colin is uncomfortable (understandably) about live measurements near the CRT (10kV!) this could be done at the socket end of the connections to the motherboard followed by wiring continuity checks with the PET switched off. Just a thought.
Personally I'd also clean all the plugs and sockets connecting the monitor especially knowing that the machine has lain dormant for a lengthy period. After all the main fault symptom is 'no display'. Forgive me if I've missed something but this has developed into a rather long thread. Alan |
Re: Non-working Commodore PET 3016
We've had some setbacks mainly due to an errant meter which refused to read a signal which was actually there, so we are right back at the start now but this time equipped with a scope.
What we are pursuing at the moment is just to find out whether the CPU is active or not because if it is not then there will be no point in looking at video signals. As it happens the monitor on these is so primitive that it does not have its own line oscillator stage, so as such it relies on a stream of sync pulses being output by the mainboard to excite the line output stage in the monitor. You are therefore quite right to suggest that a first line of attack would be to see if the video and sync signals are there and then if they are, move on to the monitor and if not, fall back to the motherboard. We did try that quite early on but then got mired down by problems with measurements of frequencies in particular, as Colin did not then have any other method of detecting / looking at pulse waveforms. Now that Colin has a scope, we should be able to make better progress. The other thing to note is that, as far as I know this has never worked in Colin's ownership, so it's not like it was put away working or previously seen working and now doesn't work - so literally anything is possible with this. Throw some obvious damp / moisture damage into the mix, and we have an interesting challenge. |
Re: Non-working Commodore PET 3016
Right, so looking at Colin's link in #215 I would say that in the second and third captures the scope does not have a good enough GND connection to a known 0V point on the mainboard.
The waveforms on the computer board should ideally look more like the one in the first trace which you took from the calibration / test waveform output on the scope. If the flying earth lead on your scope probe is long enough try clipping it to the lead on the right hand (-Ve) end of the big capacitor which has a cable tie around it. |
Re: Non-working Commodore PET 3016
Quote:
It would be good to get a set of scope traces for the 6502 pins that we discussed, to confirm the frequency measurements indicating the 6502 is not working, and also some practice for Colin with the scope ready for the next part of the investigation. Putting the scope captures in a doc file is a good system as it allows a brief description of what was being measure. Maybe use ctrl-print screen so only the scope window is captured. I think Colin should stay well out of the CRT High voltages, but checking at the output of the main board is probably a good next step. I’d recommend trying to find as many faults a possible before trying to buy more spares, I think there is more than just a faulty 6502. Does anyone have any recommendations on a source for nmos 6502 or any confirmation that the later 65C02 or other variants could be used in this version of the PET? |
Re: Non-working Commodore PET 3016
I'm going to delve into the loft tomorrow for other reasons but my chip museum is also there and I'll have a look to see if I have an original 6502 up there.
We all agree that there is no reason at this stage for Colin to be delving into the monitor. |
Re: Non-working Commodore PET 3016
Quote:
I agree that there's no need for Colin consider this at this stage. Alan |
Re: Non-working Commodore PET 3016
Quote:
2, 7, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 22, 23, 24, 25 Thanks. Colin. |
Re: Non-working Commodore PET 3016
Quote:
Note the switch on the probe on the scope is set to x1. It has a x10 option but I'm not really sure what that does. I was really happy enough to be able to get the scope working at all.... https://drive.google.com/file/d/1jvO...ew?usp=sharing |
Re: Non-working Commodore PET 3016
The second capture of three - from UG5 pin 7 I gather, although the text got displaced a bit - shows a reasonable 1Mhz signal, a bit rounded but it is there. There is a way you can probably get that to look more 'square' but we'll come to that shortly.
Can you now have a look on at least pins 9, 10, and 11 of the CPU to see what you get there, plus any additional pins which Mark suggested?? |
Re: Non-working Commodore PET 3016
I think pin 37, 39, 3, 2, 7, 9, 10, 11.
Should be enough to confirm 6502 is getting clock, but not running. |
Re: Non-working Commodore PET 3016
1 Attachment(s)
Quote:
Alan |
Re: Non-working Commodore PET 3016
Quote:
That comes at a cost of sensitivity of course but, usually not a problem with modern scopes and the fact their displays will auto adjust when you select x10 option will help so you don't keep misreading values forgetting to multiply them by 10. It would not hurt to just always use x10 at these signal levels but, make sure you adjust the probe with the reference signal for a nice square wave as in your setup instructions. There is usually a little screw near the BNC connector. |
Re: Non-working Commodore PET 3016
Yes, that was what I was referring to in #224 when I said the waveform could be made to look tidier. The frequency compensation trimmer is sometimes in the body of the probe rather than in the BNC plug. It would also be useful to switch channel 2 off so that channel 1 can use the whole height of the display.
I've been in the loft this morning and unfortunately could not find what I was looking for, a smallish microprocessor PCB which I know for sure has a Rockwell 6502 on it. As it's not where I was absolutely sure it was, I don't think I'll be able to find it in any reasonable time frame. I did find some NOS Z80 CTC (Counter-Timer) ICs though - anyone need one? Let's see what Colin finds on those signal lines, and hopefully we can say whether the existing 6502 is OK or not. |
Re: Non-working Commodore PET 3016
Quote:
I have had to use the chassis as earth as the earth wire on the probe is not long enough to go to C62 from the 6502. I have included the frequency that the scope has detected at the bottom of each screenshot. https://drive.google.com/file/d/1Bdt...ew?usp=sharing |
Re: Non-working Commodore PET 3016
Quote:
Sorry for the dumb question. Thanks. Colin. |
Re: Non-working Commodore PET 3016
Nothing is dumb - just do it once or if you change from x1 to x10 settings as it calibrates the probes. You would also need to do it again if you bought new probes - also try to use the same probes on each port as the trim will be different for each port. This means you match a probe to a port
|
Re: Non-working Commodore PET 3016
Did you notice that on the scope screen there is a setting that allows you to say if the probe is in X1 or X10 switch position - this will mean that the displays read correctly - in your document to ensure we are reading the values correctly can you confirm that option was set by you - if not we will just adjust the values manually when looking.
|
Re: Non-working Commodore PET 3016
I think we had better be specific. If possible Colin, try making all measurements from now on with the following parameters set:
With no signal present, adjust the vertical position of the scope trace line (the 'Y Position') so that it sits on the second faint line counting upwards from the bottom. This will be our 'zero volts'. I don't see an actual 'Y position' control such as there would be on a real scope but I think maybe you can click and drag the yellow '1' tab at the left hand side? If you can choose between 'DC' / 'AC' / 'GND' for the input mode choose 'DC'. On the probe: select x10 On the scope 'Front Panel' - select "x10" Set Channel 1 'Y' sensitivity: 1V / Div. (One Volt per division) Turn channel 2 off. With all of the above set, go to UG5 pin 7 where you have a known 1MHz signal and while looking at that, adjust the trimmer on the probe body or BNC plug for the squarest looking signal. Usually, you get a plastic trimming tool with the probe intended for that exact purpose. The setting of the 'horizontal' control will vary with the frequency of the signal you are trying to look at. For now, just get into the habit of adjusting it up or down until you can see around 5 or 10 cycles (ups and downs) of the signal you are trying to look at. I understand that the negative leg of C61 isn't always conveniently close to where you want to make measurements. With the power off, use your meter to check for continuity between that component leg and the chassis. If the resistance is less than 1 ohm then it is OK to use the chassis as the scope ground provided the point you choose on the chassis is clean and not painted - the ground lead of the probe must have good electrical contact to circuit 0V. |
Re: Non-working Commodore PET 3016
Looking back at the readings you collected, especially from pins 7, 9, 10, 11 - although the clock signal is reaching the CPU and passing through it, the CPU seems very inactive. If you can, please just verify (with a meter, with power on) that pin 40, Reset, is still at +5V.
Apart from that, anything else we can go at before we have to suspect the CPU? Mark? |
Re: Non-working Commodore PET 3016
I guess we had better check for +Ve supply to the CPU since it seems so dead, so Colin, as well as re-checking the voltage on pin 40 (reset) could you also look at the voltage on pin 8 (should be +5V). Although the clock is getting through from one side to the other, that might happen even if the IC was not powered. I'm thinking possible bad socket contacts or missing supply due to damage by damp / corrosion.
|
Re: Non-working Commodore PET 3016
Quote:
For the scope settings I would suggest starting with 500ns per division and widen out to about 5us per division. If you don’t see a trace on the scope try adjusting the trigger level, ;keep it on auto might be easiest. Did the pet have more than one 5v supply? Some of the older systems needed too much current for a single regulator. |
Re: Non-working Commodore PET 3016
At it happens, there are two +5V supplies, although both were verified as being present at the regulators during early checks. The supply rails are the thickest tracks on the PCB so they are unlikely to have been corroded through, but individual pins could have corroded away where they meet the PCB surface if there has been standing moisture / dew on the PCB.
|
Re: Non-working Commodore PET 3016
I have some UMC6502A's in my stash, these are NMOS according to the details I can find. Are these likely to work in the PET ?
dc |
Re: Non-working Commodore PET 3016
Quite possibly Dave, so we'll keep you in mind if that is OK. Still trying to establish whether the original 6502 is doomed or not.
|
Re: Non-working Commodore PET 3016
No problem, PM me if required.
ATB |
Re: Non-working Commodore PET 3016
Quote:
Colin. |
Re: Non-working Commodore PET 3016
Quote:
|
Re: Non-working Commodore PET 3016
Quote:
|
Re: Non-working Commodore PET 3016
Quote:
Pin 40 gives me 5.135V |
Re: Non-working Commodore PET 3016
Quote:
Pin 8 gives me 0.81 V And since removing the 6502 to check the legs for any signs of corrosion, I'm no longer getting any frequencies from pins that I was getting frequencies from before (pins 3, 37 & 39).... I was careful and did earth myself whilst removing/checking, but this feels like a step backwards now. Despondent of Perth. |
Re: Non-working Commodore PET 3016
Don't be despondent Colin. The fact that you haven't got 5V on pin 8 means that the the 6502 doesn't have power and it won't do much without it. Could be a number of causes but the most likely is a poor connection between pin and socket (assuming that you measured the voltage at the pin itself).
Alan |
Re: Non-working Commodore PET 3016
From what I've seen on YouTube, Commodore motherboards are notorious for using really cheap IC sockets that only make contact on one side of the pin. Unseating and reseating a chip from one of these can make things worse, not better.
If you fit turned-pin replacements, it might be worth stacking an extra socket into each one just in case; it's a much messier job to remove one of these from a double-sided board if you damage it. |
Re: Non-working Commodore PET 3016
With replacing big connectors in frail boards I sometimes use cutters to break up the plastic and desolder the pins one at a time.
|
Re: Non-working Commodore PET 3016
The reading Colin has on pin 8 says there is no power to the CPU power pin, it shouldn't take too much to find out why that is. As others have said (and was mentioned somewhere way back in this thread), PETs have a reputation for bad / unreliable IC sockets.
Colin is now, probably always was quite good at reading circuit diagrams and especially good at making continuity measurements, so the next thing we need him to do is to put the CPU back in the socket and (with power off) begin looking for continuity from each CPU pin to one other thing which that CPU pin should be connected to according to the diagram. Note down any which don't appear to be connected to where they should be. It might ultimately be necessary to replace all of the sockets, but we'll cross those bridges one at a time. |
Re: Non-working Commodore PET 3016
Thanks all. I did measure the voltage at the pin, so that's a true reading from the 6502's perspective.
I've ordered some 40 pin turned-pin sockets from Cricklewood Electronics in case I need them and I'll get on with chasing the continuity tomorrow. Thanks all. |
All times are GMT +1. The time now is 1:33 pm. |
Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.