Uncreative Labs

Joined: 20 Nov 2004 Posts: 589 Location: Shandong, China

My 386 system is based on a 386/486 motherboard with vlb slots, and I was really hoping to switch to a more authentic 386 board. I have two other boards based on opti 495slc and ALI M1429. The 495slc board has a good layout and supports 256kb cache, but for whatever reason refuses to work with my upgrade chips and only likes the straight 386s.

The ALI M1429 board was produced very late, and has an AMI bios date of 4/4/93. This board even supports writeback cache! However, there are problems with it. It had an am386-40 chip soldered in place, and even though it has an additional 386 socket, plugging chips in there don't work. I ended up slicing off the surface mount chip and discovered some corrosion underneath. Now the board is acting flaky with the PGA 386 installed (usually random resets or a20 handler errors). I think that the corrosion is probably the cause of it. I'm not sure if I'll be able to repair the damage...which is too bad, because it's a damn fast board (supports and caches up to 128mb of RAM too).

I ended up just sticking with my weird hybrid board but I did make some good progress. I was able to get my blue lightning module working. Although I have the software that matches it (some weird japanese pc98 crap) it refuses to detect the chip. This is probably because I don't have a real PC 98 machine. I managed to get the kingston lightning software going with it though. I had tried this software in the past without success, but this time it somehow worked!?! Though I techically have a 3x25 (75MHz) chip, I am unable to use the 3X multiplier with the kingston software. I currently run in 2x33 (66MHz) mode which is pretty bloody fast. According to all of my benchmarks, integer is even faster than a real intel dx/2-66. I tried overclocking the system to 2x40 (80MHz) but I have had no success at all. The system won't boot at that speed which is really a bummer. 3x33 (100MHz) was my goal, but if it doesn 't do 80MHz I think it's not going to happen.

Another great thing is that I got the VLB working properly while using the blue lightning chip. I am currently using the VLB version of the Mach64 which is now faster than the ISA version (before it was the other way around). I am not sure why this is, but it's certainly nice!

Joined: 01 Oct 2006 Posts: 36 Location: USA

Quote:

It had an am386-40 chip soldered in place, and even though it has an additional 386 socket, plugging chips in there don't work.

I've never seen a 386 motherboard with both a soldered chip and an empty socket. Could be for upgrading in which case I would think it would disable the soldered in CPU. Tearing it out may have caused a short somewhere though, might check it over carefully but my guess would be it's toast.

Joined: 20 Nov 2004 Posts: 589 Location: Shandong, China

I didn't exactly "tear it out", I carefully cut it off making sure I didn't damage the board traces. I went over my work with a magnifying glass. Mistakes are repairable. The problem as I said is that there was corrosion under the PQFP 386 chip. My guess would be that the company who made the board (generic taiwanese) used a crappy machine and trapped some air underneath. I suspect scraping away the corrosion caused a short or an open circuit. I haven't been able to find it yet. I'll have to try some strong hydrogen peroxide to remove the remaining corrosion to see what happens.

As far as I can tell this board had no provisions for actually using the extra PGA socket when the PQFP 386 is installed. I tried inserting DLC chips into the socket but they were not detected by the board (reports 80386 and does not detect presense of cache). This board has very few jumpers. I do not have the original manual, but I tried them all in every combination possible and could not disable to the onboard chip. Since this is an extra board I thought I'd practice my surface mount skills. I think I did a pretty good job though. If I wanted to I could resolder another 386 chip on there, but I don't think it would fix the corrosion problem.

Joined: 01 Oct 2006 Posts: 36 Location: USA

Yeah, bad choice of words. I know you took it out carefully but won't that still leave a lot of open circuits with no chip there (not counting any damage from the corrosion)? I'm guessing it wasn't designed to be taken out so if that is an upgrade socket it may not work without the original chip there. That's just a wild guess though, it may just be a dead motherboard.

Joined: 20 Nov 2004 Posts: 589 Location: Shandong, China

As far as I can tell the two chips are connected in parallel (tested with a meter), so removing one chip would not cause any open circuits. However, the corrosion may be causing short-circuits, which would definetly cause a problem. I don't personally believe that the PGA socket is intended to be an upgrade socket, otherwise the manufacturer would have provided jumpers to disable the onboard CPU (all other boards similar to mine in the TH99 database have this).

Joined: 01 Oct 2006 Posts: 36 Location: USA

Well as you can tell I'm no expert on this type of board but I hope you get it working.

Good luck Smile