Image Map Image Map
Page 1 of 2 12 LastLast
Results 1 to 10 of 14

Thread: IBM PC-Retro Kit IBM 5150 SuperSoft 8237A Failed

  1. #1
    Join Date
    Jan 2017
    Location
    Zürich, Switzerland
    Posts
    17

    Default IBM PC-Retro Kit IBM 5150 SuperSoft 8237A Failed

    Hello

    I just completed the PC-Retro Kit a few hours ago. Now I tried to start it (Just see if it passes the BIOS POST because I have now floppy controller and no XT-Keyboard at the moment).
    The board starts and shows the bios screen followed by a short beep, then it types that there is no floppy in Drive A:.
    I then switched U33 to the Siupersoft POST Diagnostic to be sure everithing is right. But it displays me a whole lot of errors.
    It fails the
    • 8237A DMA test
    • the keyboard controller,
    • (as expected the floppy controller)
    • system memory at adress 32322 (with all failing bits except the 0th)
    • slow refresh to A0000

    Im not entirely sure if it is ok if it fails the memory tests, because in the manual the super soft screenshot shows these tests as failed too.

    I have an CGA card installed and 64k (one bank) of memory.
    I tried to refit the 8237A two times but with no success.

    Thanks for your help.

    Thomas

  2. #2
    Join Date
    May 2006
    Location
    Melbourne, Australia
    Posts
    6,531

    Default

    Quote Originally Posted by ThommyThomaso View Post
    Hello
    Welcome to these forums.

    Quote Originally Posted by ThommyThomaso View Post
    ... the keyboard controller,
    Note that the KEYBOARD CONTROLLER test does not test the keyboard controller. Why? Because the IBM 5150 (which the PC-Retro Kit is a clone of) does not have a keyboard controller chip. The KEYBOARD CONTROLLER test is a test of the keyboard interface circuitry and the keyboard.

    Quote Originally Posted by ThommyThomaso View Post
    ... and no XT-Keyboard at the moment)
    In the IBM 5150, the KEYBOARD CONTROLLER test is known to fail if there is no XT-class keyboard attached. So you need to wait until you get an XT-class keyboard.

    Quote Originally Posted by ThommyThomaso View Post
    I have an CGA card installed and 64k (one bank) of memory.
    Quote Originally Posted by ThommyThomaso View Post
    ...system memory at adress 32322 (with all failing bits except the 0th)
    ...slow refresh to A0000
    Note that in the IBM 5150 (which the PC-Retro Kit is a clone of), the two RAM switches on motherboard SW1 do not enable/disable RAM banks. Those switches inform the BIOS as to how many banks are populated.

    The Supersoft/Landmark ROM ignores the switches. It expects all 4 banks of RAM to be populated. That is probably why the Supersoft/Landmark ROM indicates a RAM error, but the Anonymous BIOS does not.

    Populate all 4 banks of RAM, then see if the Supersoft/Landmark ROM indicates a RAM error.

    Quote Originally Posted by ThommyThomaso View Post
    ...8237A DMA test
    Quote Originally Posted by ThommyThomaso View Post
    I tried to refit the 8237A two times but with no success.
    A faulty 8237A is a possibility.

    The PC-Retro Kit has over 2000 solder joints. I think that you should get out a magnifying glass and check all of the joints very carefully.

  3. #3
    Join Date
    Jan 2017
    Location
    Zürich, Switzerland
    Posts
    17

    Default

    I have ordered the parts to build the AT2XTKB adapter, so I should have a Keyboard by the end of the week.
    I already thought that the Supersoft ROM might check the Keyboard itself. And therefore report a failure.
    I have ordered full 256kb ram so I will later populate all 4 banks and try the Supersoft test again and see what will happen. I was a bit frightened that the DMA failure could also cause that the RAM refresh cirquitry will fail and therefore all the DRAM errors occure. But on the other hand if the RAM refresh does not work properly the Anonymous BIOS would report that, wouldn't it?

    I'm still a bit confused that the BIOS won't notice a fault in the 8237A. I mean accourding to the technical manual the DMA Controller gets checked on POST. Could it be that the 8237A failure reported by the Supersoft could be a result of a missing Floppy controller? I have an WD Floppy/Winchester controller but it seems as it won't work. The BIOS will just show a flashing cursor if I plug the card in. So I try to test the mainboard with only a graphics card installed.

    I already checked all the joints twice while soldering, but I certainly will recheck them all with a magnifing glass later. Do you have an idea where I could find a spare 8237A? I mean if really mine would be faulty I would have to find a replacement.

    Thanks for your help.

  4. #4
    Join Date
    May 2006
    Location
    Melbourne, Australia
    Posts
    6,531

    Default

    Quote Originally Posted by ThommyThomaso View Post
    I'm still a bit confused that the BIOS won't notice a fault in the 8237A. I mean accourding to the technical manual the DMA Controller gets checked on POST.
    You should not read "checked" as "100% checked".
    Checks done by the POST are usually very basic, both to speed the boot process, and because of limited space on the ROM.
    These are only confidence checks.
    The POST in different BIOS' will check various bits to varying degree - it is up to the BIOS author.

    For thorough hardware testing of an IBM 5150, IBM supply diagnostic floppies.

    If the 8237A was 'dead', of if the 8237A (or supporting circuitry) was such that the 8237A was 'unreachable' by the Anonymous BIOS, then yes, you would expect the Anonymous BIOS to 'complain' about such a basic problem. But maybe the 8237A problem (if there is indeed one) is not basic.

    Quote Originally Posted by ThommyThomaso View Post
    Could it be that the 8237A failure reported by the Supersoft could be a result of a missing Floppy controller?
    For the IBM 5150, no floppy controller is required for the 8237A test. The test passes.

    Quote Originally Posted by ThommyThomaso View Post
    I have an WD Floppy/Winchester controller but it seems as it won't work. The BIOS will just show a flashing cursor if I plug the card in.
    * Faulty controller ?
    * Floppy controllers use DMA (i.e. 8237A) !!

    Quote Originally Posted by ThommyThomaso View Post
    But on the other hand if the RAM refresh does not work properly the Anonymous BIOS would report that, wouldn't it?
    One would expect (repeat: expect) that if RAM refresh was not working properly (or at all), that the POST in the Anonymous BIOS would report a RAM error when it did its crude RAM test.

    If you had BASIC chips, you could experiment by entering a few lines of BASIC program, then seeing if it was still present 5 minutes later.

    To be noted is that RAM refresh only uses one channel of the 8237A.

    Quote Originally Posted by ThommyThomaso View Post
    Do you have an idea where I could find a spare 8237A? I mean if really mine would be faulty I would have to find a replacement.
    Did MTM Scientific indicate that the 8237A would have been tested prior to dispatch ?

    eBay will have the 8237A (but will it be tested!)


    Some possibilities that I can think of:

    * Faulty kit assembly
    * Faulty 8237A
    * Faulty circuitry associated with the 8237A
    * False alarm by the Supersoft diagnostic ROM. Do other BIOS' indicate a DMA problem ?

    In the end, diagnosis may require the (skilled) use of test equipment such as a logic probe and logic state analyser.

  5. #5
    Join Date
    Jan 2017
    Location
    Zürich, Switzerland
    Posts
    17

    Default

    For the IBM 5150, no floppy controller is required for the 8237A test. The test passes.
    Ok thanks. I couldn't find this piece of information in the manual.

    * Faulty controller ?
    * Floppy controllers use DMA (i.e. 8237A) !!
    I have checked the controller in my 386 and there both winchester and floppy part of the controller worked. But maybe it isn't jumpered right for the 5150. I haven't found any manual yet for the controller. So I just leave the default jumper position.
    I thought about that a really faulty DMA would give some problems with the floppy controller. But as far as I understood it, the initialisation of the drives works through ordinary registers. So I'd guess the Floppy card and drive would at least get a reset, but its just dead silent. So I assume the controller is not properly jumpered.

    If you had BASIC chips, you could experiment by entering a few lines of BASIC program, then seeing if it was still present 5 minutes later.
    I will burn the basic chips as soon as possible. Its just I first need some adapters. I just have 27Cxx laying around.

    Did MTM Scientific indicate that the 8237A would have been tested prior to dispatch ?

    eBay will have the 8237A (but will it be tested!)
    I wrote an E-Mail to MTM and asked if they have encounterd such an error before.

    Sure. thats an issue I havent thought about.

    * Faulty circuitry associated with the 8237A
    I will check out the technical manual and try to check the solder joints and the IC's. I have the equipment to test some simpler IC's but unfortunately I'm not that well equipped here.
    I will ask some people at school if I could get access to some testing equipment.

    False alarm by the Supersoft diagnostic ROM. Do other BIOS' indicate a DMA problem ?
    I will burn the original IBM Bios to a 27Cxx and see how it wil react. Also I saw there is a whole series of Testing roms on the support CD. I will burn those and see what they will report.

    Thank you for your assistance. I really was as careful as possible when soldering the board and placeing the IC's. I really hope that in the end the board will work properly.

  6. #6
    Join Date
    May 2006
    Location
    Melbourne, Australia
    Posts
    6,531

    Default

    Quote Originally Posted by ThommyThomaso View Post
    I will burn the basic chips as soon as possible. Its just I first need some adapters. I just have 27Cxx laying around.
    The adapters that I have seen are wider than the ROM sockets.

    Quote Originally Posted by ThommyThomaso View Post
    I really was as careful as possible when soldering the board and placeing the IC's.
    There is one more possibility that came to mind. You may have been supplied with a faulty circuit board. I have not been involved in modern circuit board production and so I do not know what the likelihood of a faulty board is. I think that it would be pretty low, but a possibility nonetheless.

    Quote Originally Posted by ThommyThomaso View Post
    I really hope that in the end the board will work properly.
    All of the chips are socketed. Potentially, MTM Scientific could loan you set of working chips for diagnostic purposes.

  7. #7
    Join Date
    Jan 2017
    Location
    Zürich, Switzerland
    Posts
    17

    Default

    I would solder one adapter for all 4 chips. As far as I understood it just switches a few lines and pulls the unused adress pins high.
    I think chances are much higher that ether a chip is faulty or I really have messed up one solder joint. I mean just because I see no bad solder joint doesn't mean there is none.
    I wrote MTM Scientific but I still wait for an answer. I think that would be the easiest way to go. Switch the DMA with a confirmed working one.

  8. #8
    Join Date
    Jan 2010
    Location
    Central VA
    Posts
    4,796

    Default

    Quote Originally Posted by modem7 View Post
    There is one more possibility that came to mind. You may have been supplied with a faulty circuit board. I have not been involved in modern circuit board production and so I do not know what the likelihood of a faulty board is. I think that it would be pretty low, but a possibility nonetheless.
    I should think they would be e-testing a board that complex. Even the XT-IDE rev 3 boards get an e-test!

  9. #9
    Join Date
    May 2006
    Location
    Melbourne, Australia
    Posts
    6,531

    Default

    Quote Originally Posted by glitch View Post
    I should think they would be e-testing a board that complex. Even the XT-IDE rev 3 boards get an e-test!
    Until your post, I had never heard of an "e-test" for PCBs.
    According to the web page at [here], the "flying-probe" type of e-test is not flawless.

  10. #10
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    33,656
    Blog Entries
    18

    Default

    If the 8237A is a complete and utter washout, DRAM won't be refreshed. That should manifest as a whole pile of issues.

Tags for this Thread

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
  •