Image Map Image Map
Page 1 of 6 12345 ... LastLast
Results 1 to 10 of 55

Thread: 5150 motherboard no boot diagnosis

  1. #1
    Join Date
    Jan 2019
    Location
    New Hampshire
    Posts
    153

    Default 5150 motherboard no boot diagnosis

    I've been working on diagnosing this motherboard for the past few weeks, but I haven't yet been able to determine the cause of the fault.

    This is a 5150 16-64k board. Banks 1-3 are populated with known good chips. Bank 0 is still soldered in the board. Basically, the board does the no boot "dead board" thing.

    I obtained a landmark rom and I'm getting two different results.

    DSC00407.jpg

    This first screenshot is the result I normally get. This fault occurs between the interrupt controller test (passed) and the start of the "hot interrupt" test. It enters into a high-low repeating beep and does not continue past this point.


    DSC00408.jpg
    In this second example, I've inserted a certain video card .. it is a clone card w/ selectable mono or cga. When I use this particular card, the diagnostic continues normally until a later point. Notice the on screen corruption beginning w/ the mono memory test. It also did not detect the mono ram? It then gives the memory error at 00800 bit 6 = 2048, so why does the 16k test pass initially, but then fail at this address later on?

    I've probed this board quite a bit with an oscilloscope and I haven't really found anything wrong ... lines are pulsing, I've checked the clock frequencies and the timer outputs, etc. I've checked all the address and data lines and haven't found anything seemingly stuck high or low.

    I should mention that a few of the ICs on the board have some minor rust and corrosion on the legs, which I've cleaned. I've given these chips a lot of scrutiny with the probe and these chips appear to be functioning despite the appearance of surface corrosion. (U5-U19 chips nearest the card slots, bus txvrs and such, but they appear to be working)

    Given the two different results, I'm not sure if the problem is really the ram or not. It could be that this board has more than one fault. Before I go replacing chips on this board, I figured I'd make a post about this scenario to get some more opinions about what the cause may be.

    Anyone ever see this sort of an problem with a 5150 board before?

    Any chance that a ram error at the given address could cause the fault the first condition?

    Maybe the interrupt controller? (but then why does it pass in the second condition?)

    Hmm, I love a good mystery

  2. #2
    Join Date
    May 2006
    Location
    Melbourne, Australia
    Posts
    6,228

    Default

    Quote Originally Posted by chris_nh View Post
    ... lines are pulsing
    Of course, that does not cater for all failure modes. For example, gate #1 of chip U14 in the circuit at [here] is faulty, and its output (pin 3) still alternating.

    Quote Originally Posted by chris_nh View Post
    It then gives the memory error at 00800 bit 6 = 2048, so why does the 16k test pass initially, but then fail at this address later on?
    The Supersoft/Landmark diagnostic is known to sometimes display misleading information relating to RAM, particularly when it is run on a 16KB-64KB motherboard.
    Maybe the "0800" is not really address 0800.

    Have you tried Ruud's diagnostic ROM (at [here]) to see if it reports a RAM fault ?

    Quote Originally Posted by chris_nh View Post
    Anyone ever see this sort of an problem with a 5150 board before?
    I haven't.

  3. #3
    Join Date
    Jan 2019
    Location
    New Hampshire
    Posts
    153

    Default

    I would love to try Ruud's diagnostic, but unfortunately I don't have any rom burning gear yet.

  4. #4
    Join Date
    Jan 2019
    Location
    New Hampshire
    Posts
    153

    Default

    Also, thanks modem7 for making me realize I haven't really looked at the logic gates yet. I've checked all the transceivers and latches and such, but I haven't really looked at the logic gate ICs. That'll be next...

  5. #5
    Join Date
    Jan 2019
    Location
    New Hampshire
    Posts
    153

    Default

    Update: I tried piggy-backing the bus chips and when I placed an LS244 on U16, the behavior changed .. the #2 result above became the same as #1 result, but it wasn't consistent between boots and had a few other intermittent weird behaviors. However, it is hard to tell if the piggy backed chips are making good contact on the rusty legs of the chip beneath, but seeing that I caused different outcome, I went ahead and replaced U16 (actually replaced LS244 U15-U17). Unfortunately, the results are the same, but it got me thinking that it is likely a problem on one of the bus lines and maybe a logical neighbor of U16. From what I can determine in the logic diagram, I see U14, U27 and U66 as possibilities. I plan to keep testing this board until it is fixed or I replace every damn chip on it lol.

    Also, I went ahead and ordered some ROMs and a programmer, so I should be able to try Ruud's diagnostic in a few days. I'm curious to see what result that produces.

  6. #6
    Join Date
    Jan 2019
    Location
    New Hampshire
    Posts
    153

    Default

    OK, finally able to run Ruud's diagnostic and the results are very interesting:

    DSC00417.jpgDSC00418.jpgDSC00419.jpg

    So, I have no clue what's going on. The landmark rom does not have the screen corruption that this does. Ruud must be doing something very different to write the screen than the other rom. Can anyone translate these results into english?

    Thanks

  7. #7

    Default

    Quote Originally Posted by chris_nh View Post
    Ruud must be doing something very different to write the screen than the other rom.
    I write the text directly into the screen memory, as does Landmark. Why? Immediately after the start there is no RAM available. No RAM means no interrupt vectors and no stack. So (AFAIK) the only thing left to display something on the screen is to write it directly to the screen memory. Because we don't know what type of display is used, we write to both the screen memory of the CGA and MDA display.

    Although the layout looks a lot the same, most code to write the result is written starting from scratch. This is certainly the case for the part that writes the text to the screen.

    I really wonder why in the first picture there is no text until the check of the 8259 controller. My first impression was that this was the first part that uses the stack but that is not the case; the RAM test self already uses the stack. If my TextToScreen routine is faulty; why does it work now? And indeed why does Landmark work from start-up and my diagROM doesn't? I had a quick look at both sources but for the moment I have no idea. But see further on.

    The text 8259 shows up as xruy: hey, bit 6 is zero instead of 1! But the why show 8087 and other numbers up fine? That brings me to piggy-backing ICs. I don't like that and never will use it. If a output pin is broken and is always is in tri-state, piggy-backing will work fine. But if the output pin is fried in some way, it doesn't only distort results but also can fry the piggy-back IC. I have the feeling that a buffer playing weird, otherwise I cannot explain the results for the moment. My advise is: if you suspect a buffer, replace it. And when you use a socket a future replacement is much simpler.

    If a buffer is to be blamed to disturb things from time to time, it is a possible explanation for the reason my text showed up only after a certain time. AFAIK the the test didn't run a second run because of the zero after the keyboard test. Or does the second picture display the second run? Assuming that the video card is OK, I blame a buffer toward the ISA slots. Hmmm, just pops up: if this buffer is to be blamed, how could the video card then be configured ???
    My escape: at that moment it worked fine? FYI: I only configure the video cards once, immediately after the start-up.

    So my two cents for the moment....
    With kind regards / met vriendelijke groet, Ruud Baltissen

    www.baltissen.org

  8. #8
    Join Date
    Jan 2019
    Location
    New Hampshire
    Posts
    153

    Default

    Thank you so much Ruud! I knew it would be possible to narrow down a failing bit, but I hadn't yet been able to decipher all the clues to get there. Thanks also for the explanations of how the diagnostic is doing things. I have already replaced 3 of the LS373 on the board, but there are also a pair of LS245 with corroded legs that are sitting right next to the chips I replaced. Those will be the next targets. Yes, I also prefer to install sockets when replacing chips.

    I ran your diagnostic on a known good board, just to prove to myself that it wasn't a bad rom. It works great on a good board

    Thanks for your help. I have a lot of time and resources into this board already, and it is going to be great once I get this thing working again.

  9. #9
    Join Date
    Jan 2019
    Location
    New Hampshire
    Posts
    153

    Default

    Update:

    So I replaced U13 & U14 (LS245) ... no change in the behaviour of the board, except Ruud's rom now only boots up about 10% of the time. The landmark rom still works just fine. I've gotten Ruud's diagnostic to boot up twice, and one time had the same screen corruption, but the other time had less corruption and the words appeared normal.

    So, I replaced U13 because it has line A5 (bit 6) and U14 because they both were corroded anyway. (Of course, I also accidentally cut a trace and had to add a bodge wire, all for not fixing the problem)

    Should I be looking at D5, not A5? There is also a 373 latch at U7 which is carrying A5 and I haven't replaced that one yet. I'm just so lost on this thing and I'm desperate to get it going because this one board is holding up a lot of other things. I guess I"ll keep on indiscriminately replacing parts and cutting traces and adding bodge wires until the universe decides this board has suffered enough.

  10. #10
    Join Date
    Jan 2019
    Location
    New Hampshire
    Posts
    153

    Default

    Does anyone have a pinout for the resistor networks? RN1, RN2 and RN4??

    RN1 has some corrosion and I would like to test it. I tested RN4 as it seems each pin is 30ohm straight across. RN1 I can't figure out and get too many different values to tell if it's good or bad.

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •