First post, by GPA
I am continuing my dive into 486 era hardware and this forum has already helped me to start and run an old Intel 486 VLB board, but i have recently faced a strange issue with a "new" Intel made 486 PCI board, based on 420EX chipset.
The board seem to run just perfectly fine under the DOS and under the Windows up to a point when PCI video card is initialized. What i mean is, everything in DOS, including games like Doom and Quake, and various benchmarks, runs fine. There is one thing that worries me though. The clock speed is somewhat unstable. I am not sure what causes that, but when any program determines the clock speed of the CPU, it would vary from 100.0 to 100.6 MHz. I have tried 2 different Intel DX4 CPUs that work fine and stable on my other 486 boards, and their clockspeeds are very consistent on other boards. But on this particular board the clockspeed drifts a bit. Not much, all within 0.6 MHz, but if i run the same benchmark multiple times, i never get the same CPU clock reading. When i run memtest, it measures access times to different addresses in the main memory. And usually all the access times are the same, no matter what address is accessed. But not on this board. Every address shows slightly different access times. I believe this happens because of the clock drift. Anyway, it doesn't affect any real programs, all the games etc run perfectly smooth without crashes or other glitches.
Windows installs quickly and smoothly. It runs all the benches, including SuperPI, and many others, perfectly. But as soon as i install a video card driver and reboot, the PC doesn't load Windows any more: it starts, everything goes well up until when the VGA is getting initialized. The PC reboots at this point. It boots well in Safe Mode though. Or if i kill the video card driver and return to Standard PCI VGA, it works as it should.
I have tried 2 different PCI VGAs: Savage4 from Diamond (Stealth SIII 530) and Riva128 from ELSA. They both work with the same memory and the same CPU on other 486 boards. But they both cause this board to reboot when Windows initializes them.
I am getting to the point where i would change the main quartz oscillator, but i am far from being certain that it causes any problems: although these things are fragile, they usually either work perfectly or not work at all. More than that, i have not found a clock generator chip on the board yet (have not been looking for it, should be easy though), but old clock generator chips are usually difficult to source. This is all that comes to my mind anyway, but i do not understand why the board works perfectly except for that VGA initialization, i really cannot understand how it can be connected to a tiny clock drift... Probably the real problem is elsewhere...
Does anyone have any ideas at all please? Maybe someone faced a problem alike in the past?
I would appreciate any help. Thank you.