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

Thread: CMOS system options don't persist on cold boot after DS12887 replacement

  1. #1

    Default CMOS system options don't persist on cold boot after DS12887 replacement

    Hello there, thanks for reading my request for assistance. I'll try to keep this as brief as possible while sharing what I believe to be all relevant info.

    The original Dallas DS12887 in my 486 died a few weeks ago. On cold boot, it would display:

    CMOS battery state low

    RUN SETUP UTILITY
    PRESS F1 to RESUME
    If I went into setup to define drive geometry and boot order, the settings would be lost upon save+exit and the subsequent reboot, as expected. I was not able to boot from C: no matter what I tried.

    I decided that I didn't really want to spend the time modifying my original RTC module, and immediately bought one of glitch's pre-modified replacement modules.

    Last Sunday, I completely disassembled the machine, removed the old module and installed the new module from glitch (foolishly I didn't install a socket first). As an aside, I also removed my BIOS chip in order to dump it, but my programmer failed to read it properly, so I reinstalled it and re-assembled the entire machine. (I mention this just in case I might have damaged the EEPROM somehow???)

    After re-assembling the system and testing, I was presented with 2 beeps and the following message on POST:

    CMOS system options not set
    CMOS display type mismatch
    This wasn't terribly unexpected, as I figured all of the registers were uninitialized and needed proper values! I went into setup and set the time, defined drive parameters, and boot order. I saved and exited, and upon the following reboot my settings appeared to have been saved and it booted from C:, yay!

    Unfortunately, I quickly discovered that upon power-off, all of the settings seem to be lost! I get two beeps and the two messages about system options + display type mismatch.

    Did I really screw up by not installing a socket? Are these RTC chips particularly sensitive to heat? Why would my system be able to save settings across reboot but then lose the values after a power-down. I'm also a bit confused by the 2 beeps; a lot of references mention this is a memory parity issue within the first 64K, but I'm a little ashamed to admit that I'm not quite sure when "POST" ends or if there are several phases and if "beep codes" only apply to the earliest phase(s). Just in case, I tried a few combinations of my four RAM modules with no change (I did not exhaust all possibilities, though).

    Things I've checked:
    • battery charge
    • battery inserted properly into battery holder
    • checked continuity between glitch's board and the DS12885 pins. everything seemed to buzz out fine according to data sheet
    • proper orientation of module
    • quality of solder joints
    • a few different RAM configurations


    I wanted to post here in case there's anyone who's run into a similar issue and it turned out there was a fix through software. Is there any use in using DEBUG to probe around in the CMOS RAM? Should I bite the bullet, disassemble the machine, add a socket, and just modify my original RTC module? Any advice is greatly appreciated! I can post photos if necessary but I fear I've already rambled on too long.

    Motherboard: Shuttle HOT-433 (UMC8880 chipset, AMI BIOS)

  2. #2

    Default

    Quote Originally Posted by matt.c View Post
    Did I really screw up by not installing a socket?
    It makes it much easier to replace any chips later on if you install a socket. The RTC will need to be replaced again at some future point if you keep the system going...

    Quote Originally Posted by matt.c View Post
    Are these RTC chips particularly sensitive to heat?
    No, since they're hermetically sealed.

    Quote Originally Posted by matt.c View Post
    Why would my system be able to save settings across reboot but then lose the values after a power-down.
    You don't completely remove power from a motherboard on a "warm-boot" (reboot/reset). You do when you completely shut down a system and power-up again.

  3. #3

    Default

    I understand the value of installing a socket, but I also didn't expect to ever need to replace this module since it has an integrated battery clip. My question isn't "what makes a socket valuable" but instead "could the temperatures experienced by a Dallas chip through hand soldering potentially damage the chip?

    >You don't completely remove power from a motherboard on a "warm-boot" (reboot/reset). You do when you completely shut down a system and power-up again.

    Yes, I understand that, but I suppose what's confusing is that these values seem to be written to the proper registers in the RTC chip (because they are then able to be read after exiting setup and rebooting). Yet upon cold boot they are lost. To me this indicates a defective module or some incompatibility with my board. I trust glitch's design, but maybe I have a bad chip or I damaged it somehow. This is what I'm trying to understand.

  4. #4
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    32,474
    Blog Entries
    18

    Default

    Just curious--was that really a DS12887 in your 486 originally, or was it a DS1287?

  5. #5

    Default

    Quote Originally Posted by Chuck(G) View Post
    Just curious--was that really a DS12887 in your 486 originally, or was it a DS1287?
    Fair question. It was really a DS12887.

  6. #6
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    32,474
    Blog Entries
    18

    Default

    It might be useful to dump the values after a cold boot if your BIOS doesn't insist on writing to the chip.

    There should be a few programs in SIMTEL20 for this, if you don't already have a utility.

  7. #7

    Default

    Quote Originally Posted by Chuck(G) View Post
    It might be useful to dump the values after a cold boot if your BIOS doesn't insist on writing to the chip.

    There should be a few programs in SIMTEL20 for this, if you don't already have a utility.
    Thanks. That's a great idea; might give me some clues. I think I should be able to find that utility, wasn't aware that something like this already existed! When I first bought this machine, I had to overwrite part of the checksum in order to clear the BIOS password.

  8. #8

    Default

    Things work fine after a warm boot so IMHO the RTC IC is fine. Because it only goes wrong after the power has been turned off, the only thing I can think off is that the RTC IC doesn't get its power from the battery somehow. I cannot see how the IC gets its power from the PCB but IMHO you should focus on that. If you measure power on the two pins going inside the IC then I'm leaning towards the theory that the IC is bad. Power is there but data is not kept, what other reason than a bad IC could cause this problem?
    With kind regards / met vriendelijke groet, Ruud Baltissen

    www.baltissen.org

  9. #9

    Default

    Ruud makes a good point about the likelihood of a bad IC. While glitchworks is known for making some good stuff, it's possible something slipped by them. We've had serious problems with "fake" RTC's (not the 12887-which we don't use) at work from various vendors not working in some of our repaired equipment. Only truly good source we found was Digi-Key (which was significantly more expensive, but worth it)...

  10. #10
    Join Date
    Jan 2010
    Location
    Central VA
    Posts
    4,663

    Default

    I replied via PM, but I'll post here as well: a likely cause is that the original chip was one of the Dallas modules that didn't have the "clear CMOS" pin. Some motherboards will tie this position to ground or use it to route another signal if they used a full 24-pin DIP footprint. The solution is to cut off the pins that are not required for your situation: easiest thing to do is match the missing pins on the original module. Just snip them off the GW-12887-1 or bend them up if you don't want to cut.

    Quote Originally Posted by T-R-A View Post
    Ruud makes a good point about the likelihood of a bad IC. While glitchworks is known for making some good stuff, it's possible something slipped by them. We've had serious problems with "fake" RTC's (not the 12887-which we don't use) at work from various vendors not working in some of our repaired equipment. Only truly good source we found was Digi-Key (which was significantly more expensive, but worth it)...
    The DS12885s for the GW-12887-1s come from Mouser, so it's unlikely there's a bad chip.

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
  •