Repairing my CBM 3032 - Request for help

  • Greetings,


    I'v been a couple of days working my CBM 3032 repairs. I started with this junk board:



    And left it in this state:



    There are a couple sockets more, but the output does not change. I have configured the jumpers as the technical manual states and replaced every EPROM with its original BASIC 2 from Commodore (NOS).



    The 6502 has been tested and is working, her only PIA has been replaced (missing pins) and the second one has been added. Some components like the '154 or the 555 have also been replaced because they were prey of rust.


    After testing a few times the result is this one.


    I'm also having no DOUT at the U256:



    And CAS fixed at high-level.



    I've tested some ICs regarding CAS and seem to be working... so I don't know what could be the issue.


    About the video circuit I'm sure there is a failure too, because those "junk" values should fill the screen completely but they do only occupy 2 1/2 lines in the middle of the screen.

    I am not sure about what should check now. There's no expertise in any system of the PET/CBM line in Spain as their sales were poor (and that's being generous). There are only a few of them, most in universities or in hands of speculators, some of them being offered for no less than 800€ or even in the range of 2k-3k€. For this reason I thought this is the best place to come for some help with this system.


    Just some background... This unit is nicknamed "The International" not without a good reason. Its case, PSU and monitor came from UK in a single block. Its BASIC 2, Kernal and Editor ROMs came from the same source. The keyboard came from Germany in a three units pack (all of them incomplete but I was able to build a working and complete one from pieces from the two 2001N, the other was one from a B model). The PIAs came from the UK too, but from a different source. After three years we found its board, which came from Italy but had memories manufactured at DDR. The character ROM was from the only Andorran PET, which is working but has some strange video problems. In other words it came from all around Europe.


    Regards,

    Jaume

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • I have a scope, it's half-working but that should be enough. I'm using Zimmer's schematics for this board (he really did a good job gathering all that information).


    The 555 in the reset circuit has been replaced as was mostly eaten with rust. About the clock, my probe detected pulses at the output. But I guess I should check them. If they weren't the responsibles for the failure, wouldn't you know what should I look for?


    Thank you very much for your answer.

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • 1. Have you already attached a display unit (monitor)? Any picture on the screen?


    2. As far as I can see, all of the RAM ICs are socketed, right? Did you check them one by one?


    BTW ... dual spring-contact for the IC-sockets would be much more better than the one's on your board ... they are not reliable and are often causing contact problems.

  • 1. Have you already attached a display unit (monitor)? Any picture on the screen?


    2. As far as I can see, all of the RAM ICs are socketed, right? Did you check them one by one?


    BTW ... dual spring-contact for the IC-sockets would be much more better than the one's on your board ... they are not reliable and are often causing contact problems.

    1. Yes. I even uploaded the video at Youtube and linked it in the thread in the first post, but I link it again here -> (https://youtu.be/9vnJ7ftoyx4).

    2. Yes. Every U256 is socketed. In fact, I placed every socked on this board, even replaced the old ones -if the ICs were rusty, I couldn't expect the sockets to be better-.


    I was told just the contrary, that precision sockets were better. Does that mean that I must desolder every one of them and resolder again? :pinch:


    Thank you very much

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • Precision sockets are very OK.

    I think deleted_01_21 meant the original ones which quality was a common sore.


    The bad thing is that you need to understand the schematics and the chips data sheets to check every important component complex.

    There are probably good books on this.

    Optimum would be if you can find the service manual; it would give you an easy step-by-step guide to find the problem area.


    The monitor looks okay to me. I guess it is free-running (not synched), as it does not get a valid signal.


    The dot after turning off is an old commodore flaw that provides for ugly burn-in holes in the screen middle.

    There are fixes for that available in the web.

  • Precision sockets are very OK.

    I think deleted_01_21 meant the original ones which quality was a common sore.

    I feel relieved now...

    The bad thing is that you need to understand the schematics and the chips data sheets to check every important component complex.

    There are probably good books on this.

    Optimum would be if you can find the service manual; it would give you an easy step-by-step guide to find the problem area.

    I'm working with a printed copy of it. That's where I found it -> http://www.zimmers.net/anonftp…ters/pet/2001N/index.html


    The monitor looks okay to me. I guess it is free-running (not synched), as it does not get a valid signal.

    The monitor is fine. It was tested and found working. Running unsynched? Good, that's really the kind of point I was seeking. The few characters displayed are seen correctly, I guess that this means that "Horz Drive" is ok and I should look for "Vert Drive". Suspects are, for now: G10 (74LS00) and H8 (74LS107); both of them in sheet 6. If found working then G8 (7474), G1 (74LS08), H5 (74LS00), I1 (74S04) and G2 (74LS20); all of them in sheet 7.


    For the testing, sync signals can be obtained at the user port easily.


    The dot after turning off is an old commodore flaw that provides for ugly burn-in holes in the screen middle.

    There are fixes for that available in the web.

    Yes, I know what it is and what it does. We call it "el punt verd de la mort" (the green dot of death) and I have other monitors that have this issue. What I didn't know is that there is a fix for them. I searched but cannot find the solution. Could you point me where could I find it, please?


    Many, many thanks.

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • I had to replace the 7474 on two of mine. In general I find them problematic on most boards. On one PDP11 CPU Board I had to replace all of them. Might be worth checking first.


    That's G8. With complement of rust. I'll check it first.


    Thank you!



    EDIT:


    I've checked the board and saw G9, which is as rusty as G8. G9 is also a '74 and also has responasbilities over video. I'm thinking again that thing of the sync and if I had sync problems the monitor should have snow, but the few characters displayed are shown correctly, without flickering or any kind of displacement. G9 controls when video is shown (or not shown) which seems to describe more the failure I see.


    Thank you again

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

    Einmal editiert, zuletzt von jlopez () aus folgendem Grund: New data

  • G8, G9 replaced, no change.

    I found a broken trace between H6 (74LS107), H9 (74LS93), G6 (74LS107) and F6 (74LS107); it is named in the scj¡hematics as /RA1. After the reconnection it continues to display the same amount of garbage characters for the same amount of time with the exception that they are now semigraphic chars. Other replacements have been H10 (74LS08), G7 (74LS10), G10 (74LS00), H2 (74LS04).


    I'm feeling a bit tired, I should take some rest and continue later.

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • Good afternoon


    This morning I was able to get a working scilloscope thanks to two of my former teachers. I've taken some measures at TP9, TP7, TP2, TP1, TP3, Phi2, R/W, CAS0, CAS1 and RAS.


    TP1


    TP3


    TP7


    TP9


    Phi2


    RW


    /CAS0


    /CAS1


    /RAS


    The scilloscope figures present at my docs are from the display, not from the computer itself. While some of them are fine, others look definitely wrong. As found with the logic probe a couple of days ago, both /CAS signals are permanently high. /RAS oscillates but seems to never go to low level either...

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • I know it is tedious, but I guess the safest way would be to check all traces for low resistance, i.e. <0.1 ohms.


    The weak RAS signal can be caused by various reasons. Some near-short to Vcc, some weak trace (with high resistance) or the LS08 is defective.


    Edit:

    Phase 2 does not look good either. An 1 MHz signal driven open collector with a 470R pullup should look different.

    Einmal editiert, zuletzt von 6502 ()

  • About /RAS, I've replaced G1 (74LS08) with a new one. Same result. I'll check the traces.

    I've checked both /CAS signals; the only suspect was B3, replaced with no change. After checking again B3, found it was working and that the CPU is completely still. It's strange because this 6502 is known to be working as it was tested in a 8032 flawlessly. RDY is high and Phase 0 pulses at 1MHz. What else could cause the processor to stop?


    Yes, Phase 2 looks weird. This capture was taken directly from the CPU.


    I know it's tedious, but I'm pleased to service it ;).


    Thank you again!

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • I don't know whether CAS matters at this stage.

    I just wildly assume that the refresh logic should take care that RAS gets pulsed at some minimum rate by the hardware, even without processor.


    Are you sure the processor starts up at all?

    Can you observe activity at its address/data/control pins after you pull !RESET low for a moment?

  • Hello,


    About CAS, it didn't matter until I found it matters. The clock circuit certainly has responsibility over /CAS0, /CAS1 and /RAS, but I realized that the processor also has control over those signals. /RAS is generated using a 74LS157, which selects certain clock signals but it is enabled (or disabled) by B2 (7425) which is fed with CPU addresses, some of them after the switches. Still, I think this gate is failing, because both the strobe and one of its inputs (pin 9, connected to A15) are high and according to the datasheet, input should be low in this case but is high. Therefore, it disables the 74LS157, also disabling /RAS.


    The CPU is running for short time, but long enough to erase the screen. When a pre-crtc CBM/PET powers up its vram is full of random contents. If working appropiately the CPU inits it to 0. Otherwise those values remain. When I power it up and the monitor is not initialized no values are shown, but if I turn it off and power it up again quickly the values appear for a few seconds and then black screen is displayed. If no ROMs are in place (or wrong combination of BASIC and KERNAL) the values remain displayed. I tested the reset by removing the 555 and using a wire from the empty expansion sockets to pin 3 of the 555 socket, making contact manually for a couple of seconds while the board is powered. Prior to the reset signal I have /RAS and one of both /CAS signals. When the signal is applied, it starts to generate that weak RAS... what's strange is that because between pin 3 of the 555 and the /reset pin of the 6502 there's a 74LS04 but the thing seems to react only with a low signal on pin 3 (or a high level on /reset). When that happens, the screen is erased, meanwhile the characters are there.


    Thank you and regards

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • /RAS is generated using a 74LS157, which selects certain clock signals but it is enabled (or disabled) by B2 (7425) which is fed with CPU addresses, some of them after the switches. Still, I think this gate is failing, because both the strobe and one of its inputs (pin 9, connected to A15) are high and according to the datasheet, input should be low in this case but is high. Therefore, it disables the 74LS157, also disabling /RAS.

    Sounds like broken pcb traces. You know, the 74 chips have internal pullup.

    Did you measure the traces' conductivity?


    Edit: When looking at the massive rust, I would not exclude the possibility of internal chip damage, too...

    Maybe the processor is working until the point where it depends on functioning RAM access to proceed with initialization...

  • I remember having revised all traces from /RAS, /CAS0 and /CAS1 up to the CPU. I imagine some ic is rot... that or rust that impedes contact.

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • Two days ago the spare parts I couldn't buy at Barcelona arrived. I still have no /RAS and /CAS, but I least I found the cause of the CPU being stopped: /irq is permanently down. Today's first test will consist in testing VIA and PIAs, which are the only components that can cause this signal to go low.

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • Having removed both PIAs (they were pulling /IRQ down) and replaced B2 (7425) and H1 (74LS74) I can proudly say that after 40 years the /RAS signal is pulsing again. I have but the side effect that the screen is not erased anymore. The two (and a half) lines are permanently on the screen. Doesn't seem to react on /RESET. Next step, fixing the video circuit maybe?


    Regards

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • Video system restored. Still has some flickering, but at least I can see all the screen full. https://youtu.be/q-gkgciBJHA


    This other thread was critical in finding the issue with the video system. Same effects and same solution.

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • After H9 (74LS93) was replaced, the flickering was gone for good. The monitor now displays a clear image without issues.



    What would you suggest as the next step?

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

  • Char ROM was pulled from an (almost) working 8032 as unfortunately this computer came without its own. The replacement was known to be working. The VRAM was tested on that other board and found to be working too. Just as a curiosity, I recently found that they were made by Tesla, soviet-era Czechoslovak manufacturer; this means that the totality of its RAM ICs were from the Eastern Block.

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol: