• Model 730/750 PCI/ISA to PCI/MCA conversion

    From Christian Holzapfel@21:1/5 to All on Mon Feb 13 12:07:57 2023
    I once came across a sweet Model 730 (6876-TCH) with the 3-slot PCI/MCA riser card. The planar 40H4759 appears dead, and hangs early while counting memory, no matter what.

    Finally I found a close-to-similar but yet different Model 750 (6885-35H) with a 5-slot PCI/ISA riser. It has a different planar 88G4270.

    Reading the tool
    https://www.ardent-tool.com/PC750/#Wrong_BIOS_Flashed
    gave me hope the PCI/ISA planar could somehow be converted to swallow the MCA riser.

    Created a floppy for the PCI/MCA BIOS FLASH program (N2JT50A), booted it.
    Being asked if I wanted to change the serial no. and model no, I entered data from the 3-slot MCA system.
    Then FLSHUPDT.EXE starts preparing the update process, but halts complaining "The POST/BIOS file, $IMAGEV.USF, could not be found on the diskette."
    Which is right: The MCA floppy contains a lot of files, rather named $IMAGEM.BEF/CEF/CFF/DKF/FRF/GRF/... and also said $IMAGEM.USF.

    This sounds to me like the BIOS files for ISA and MCA risers are named differently.
    Well I *could* try renaming all the M to V files and see if it does... But if things go wrong, this could have been my only chance.

    Does anyone recall how this was done switching risers in said Tool publication? Louis maybe?

    Thanks,
    Christian

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Christian Holzapfel@21:1/5 to All on Mon Feb 13 12:37:15 2023
    The BIOS/FLASH floppy for the PCI/ISA models, n1jt89a.exe, contains all similar files, but named $IMAGEV. There is an additional $IMAGEV.BRF that has no complementary MCA $IMAGEM file.

    The actual BIOS images seem to be the .USF files, as they are 131.072 bytes in both cases MCA and PCI, and all other files are much smaller.

    Could be +1 worth trying renaming... Maybe tomorrow. Couldn't deal that desaster today.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From sandy@21:1/5 to All on Mon Feb 13 19:29:40 2023
    Christian

    As I wrote in the article, I myself have not tested the procedure but I didn't see any re-naming steps in the original article.
    During the process, you changed model nember. I don't know if changing model number is just information only in the BIOS screen or do something in flash process. But those who apply "wrong" flash program may not change the value.

    Anyway you need to configure 6885 propely as an PCI/ISA with a 5-slot PCI/ISA system before you apply Flash program of 6886.
    Even if you can successfully flash it up, changing the riser from 5-slot to 3-slot is another challenge.
    My Japanese article was written in Dec. 1999, so the original comment to the New Group by N.G. member might be posted sometime around 1996-1998. Don't now if there are such old log files out there.

    BTW, In US region, it looks like 6885 to 6886 conversion kit was available back then.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From sandy@21:1/5 to All on Mon Feb 13 23:50:06 2023
    Not a log file, but same kind of article dated 1999/12/28 is still available through the search box above.

    Title: Help PC750 flash burnt out
    Martin Adams asked and Dennis Smith replied. https://groups.google.com/g/comp.sys.ibm.ps2.hardware/c/Pmjak-89s5s/m/mBVga5-ymGwJ

    Sandy

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From sandy@21:1/5 to All on Tue Feb 14 01:01:19 2023
    Technically speaking, I think you need a 6886 riser ( 5-slot PCI/MCA) to complete the planar as a 6886. This should work according to Dennis's comment.
    Then the next move is the challenging riser change from 5-slot PCI/MCA to 3-slot PCI/MCA to make it as 6876. You may be the first one who use the 6876 riser right after flashing BIOS.

    Sandy

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Christian Holzapfel@21:1/5 to All on Tue Feb 14 09:06:32 2023
    Thank you Sandy for looking that up!

    According to that article, it should be as simple as flashing it over. Involved BIOS versions back then are the same: n1jt89a for PCI/ISA and n2jt50a for PCI/MCA.
    Looking at the disks, the include differently named files, and seem to validate for a system of their specific type.
    So I really wonder how a conversion was ever possible to that fellow...

    It does not seem to matter whether the PCI/MCA riser is 3 or 5 slot. I have both here, and they are interchangeable on a configured PCI/MCA system (another 6886). Comparing them visually, it seems they have the same ICs onboard, the same count of PCI
    slots (2 slots + bridge chip as an additional device), and only differ in the number of MCA slots they offer.

    I did an upgrade of the existing PCI/ISA BIOS to the latest version, but migrating from that to MCA is still prohibited. Maybe I need an older version of the flash utility... will try.

    Flash chip on the 6885/6876/6886 boards is U13, which is an ST M28F101-150K, pretty standard 1 MBit flash memory.
    If I dare to rename the flash image and trick FLSHUPDT.EXE, I may recover the chip with my programmer. Maybe should place a socket first...

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Christian Holzapfel@21:1/5 to All on Tue Feb 14 09:55:57 2023
    Just winged it.
    Renamed all $imagem.??? files to $imagev and bootet the modified PCI/MCA update disk.
    It sees the trick:

    "The Build ID in the image file is not valid for this system!
    Press Enter to continue anyway. Press ESC to cancel."

    Time to dare...

    "The POST/BIOS code in this system is a newer version than that on diskette.
    If you continue, the older version will replace the newer version in the system.
    Press Enter to continue anyway. Press ESC to cancel."

    Yes, be assured.

    <updating>

    "POST/BOIS update completed."

    Then turned off the system, swapped the PCI/ISA riser for the 3-slot MCA one, removed the BIOS battery for a minute, turned on...
    Happy initializes the video system, POSTs, counts RAM and asks for a refdisk. Should be straight forward for now.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Christian Holzapfel@21:1/5 to sandy on Tue Feb 14 09:24:09 2023
    sandy schrieb am Dienstag, 14. Februar 2023 um 04:29:41 UTC+1:
    During the process, you changed model nember. I don't know if changing model number is just information only in the BIOS screen or do something in flash process. But those who apply "wrong" flash program may not change the value.

    I used the PCI/ISA flash update disk to modify SN and Model to the 6876 type, but the PCI/MCA update disk doesn't seem to care.
    So unfortunately this is not it...

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From dens71ta@gmail.com@21:1/5 to Christian Holzapfel on Tue Feb 14 09:59:21 2023
    On Tuesday, February 14, 2023 at 11:06:33 AM UTC-6, Christian Holzapfel wrote:
    Thank you Sandy for looking that up!


    As I recall when doing the Select-A-Bus conversion it required flashing the BIOS
    prior to swapping the riser out. Flashing the BIOS and powering up without installing
    the matching riser caused the system to be unusable. I still have my PC Series 750
    6886 but it's been many years since I've done a lot of circuit board swapping on
    those systems.

    It's possible the planar you are using is unsupported or there is a solder/component
    issue on the board after all these years. When I started my PC 750 up back around
    2017 after a very long slumber it was exceptionally glitchy. A bake in the oven to
    reflow the solder seemed to fix it.

    -Dennis

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From sandy@21:1/5 to All on Tue Feb 14 20:19:32 2023
    As I recall when doing the Select-A-Bus conversion it required flashing the BIOS


    HI, Dennis. Good to see your name here again.

    Just for a record.
    I investigted my PC750 section in the Still Crazy articles.
    It was dated as 991209 but the date was took from my original daily log files ( dairy, actually ). When I published my PS/55 web page, I prrobably added wrong BIOS info at the end of the PC750 artcle.
    I started my PS/55 English page in Sep. 2002. So, most likely "the report in the N.G. posted 2 or 3 year ago" should be your advice to Martin. I should have to include the source precisely including the url.

    Sandy

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)