First post, by TheDudeLasse
Hey folks, I’m fighting a strange issue on an IBM Aptiva 2144 and hoping someone here has run into something similar.
SPECS:
Motherboard: Universal Scientific Ind. (USI) Pro-57A (OPTi 82C895 chipset): https://theretroweb.com/motherboards/s/univer … -scient-pro-57a
VGA: Oboard Trident TGUI9680-1
CPU: Pentium 200 MMX (running @166Mhz)
RAM: 64 MB EDO RAM
This system used to run fine with a Onboard Trident VGA + Voodoo2 + Vibra16 setup. After swapping the Vibra16 for an AWE64, the machine suddenly stopped booting with any PCI bus-mastering card installed. Worked a couple of times, that failing to initialize PCI cards.
Here’s the behavior:
RAM check + IDE detection completes (no image, i can hear it)
After 10–15 seconds I get a long continuous beep
No video from any PCI GPU or the onboard VGA
POST never completes
What works / doesn’t work:
Onboard Trident VGA works fine without the Matrox Millennium or Hercules Stingray installed.
Voodoo2 will NOT initialize properly either. It get's past the post screen and is detected in Windows when onboard VGA is used, But fails to initialize / no 3D acceleration.
ANY PCI bus-mastering card → Matrox Millennium, Hercules Stingray 3D, PCI NICs, USB cards, etc. → all cause the long-beep crash during POST (with no image on display)
All PCI slots behave the same
What I’ve tried:
Clearing CMOS (battery + jumper)
Leaving the board unpowered for long periods
Removing all RAM/CPU/cards and re-seating
Trying different RAM (EDO/FPM)
Reinstalling the original Vibra16
Testing the board outside the case
Trying every PCI slot
No change — the moment the BIOS reaches PCI enumeration, it hangs and eventually gives the long continuous beep.
So right now it looks like either:
The PCI bus-master arbitration logic has died,
OR
The Aptiva’s ESCD/NVRAM got corrupted in a way that CMOS clearing doesn’t fix.
Has anyone seen this kind of failure on IBM ptivas?
Is there a hidden jumper or trick to fully clearing NVRAM/ESCD?
Or is this likely a dying motherboard?
Any advice would be greatly appreciated — this one’s been a beast to troubleshoot.
