• Please review our updated Terms and Rules here

486 L2 cache not working

P

Paul-VT

Member
Joined
May 30, 2018
Messages
40
Location
Vermont, USA
I'm working on a QDI V4P895P3/SMTP3 V4.0 486 motherboard. At POST, there is a message that says "CACHE MEMORY BAD, DO NOT ENABLE CACHE!". This message only appears if external cache is enabled in the BIOS. If L2 is disabled, the board boots fine and is stable. If L2 is enabled, the board works but is sometimes unstable but no L2 cache is detected by common cache diagnostic programs. The jumpers and chips shown in the photo may not be correct as I was a'mid-test.

I have tried:
1. Check and recheck all jumpers
2. Check voltages on board and on cache chips
3. Check clock signal and waveform
4. Swapped in known good cache chips
5. Tried different amounts of cache
6. Tried different CPUs with speeds of 66 and 100MHz
7. Tried different RAM.
8. Dumped the BIOS and checked it against one that I found on TheRetroWeb
9. Looked for damaged traces and solder joints on board

Questions:

1. Could this be one of those "fake cache" boards? Seems unlikely as the cache is socketed.
2. Can anyone suggest which ICs or other discrete components may be the culprit?

Thanks
-Paul in Vermontv4p895.jpeg
 
Before his thread gets too hot, I should cut it short. The more I work on the board, the less stable it seems.

The board will now only POST about 1 out of 4 times. When it doesn't POST, my analyzer card's leds are lit properly but no codes appear.

More testing is in order before this thread can move forward.

I was looking on eBay for clear photos of this board for reference and noticed the eye-watering prices. It sure would be nice to fix it.
 
Last edited:
I don't think there is much we can do to help you. You probably have to go thru all the standard trouble-shooting steps.

The only thing I could think of is probably a broken connection on the square OPTi chip. Try pushing it down while powering on the system. If it gets more stable, inspect the pins under a microscope.
 
Are those SIMMs EDO? I can't read the numbers exactly but notice the 1997 date code.
 
Have you tried replacing the jumpers or at least making sure they are making a good connection? I've had issues with many boards this era and those sealed jumpers... I would make sure each jumper isn't loose and is clean.
 
I have tried different simms. Including one that has a sticker indicating FPM ram. Maybe the sticker is wrong? I will check the jumpers for continuity and the RAM chips part number.
 
Unlikely that EDO causes that issue. If a 486 does not support EDO, it would simply not post.
 
Timo W. said:
I don't think there is much we can do to help you. You probably have to go thru all the standard trouble-shooting steps.

The only thing I could think of is probably a broken connection on the square OPTi chip. Try pushing it down while powering on the system. If it gets more stable, inspect the pins under a microscope.
Click to expand...
I think you may be on to something. I pressed down hard on the big OPTi chip while powering on the board. It came to life, POSTed and did not give the cache error. The board reported 256k of cache. As soon as I lifted my thumb off the chip, the system crashed.

Big question now: Should I attempt to reflow the solder for that chip? I'm no expert at soldering delicate SMDs and I don't have a microscope... yet :)
 
Do you have a hot air station and some decent flux? Or do you just have a regular iron, and if so, do you have a flat tip?
 
twolazy said:
Do you have a hot air station and some decent flux? Or do you just have a regular iron, and if so, do you have a flat tip?
Click to expand...
Nope. Just a regular iron with a round tip. After watching some videos on youtube I think I don't have the nerve to do the repair. Maybe sometime in the future if I acquire the necessary equipment. Looks like the legs of the chip and the pads underneath are extremely delicate.
 
Paul-VT said:
I think you may be on to something. I pressed down hard on the big OPTi chip while powering on the board. It came to life, POSTed and did not give the cache error. The board reported 256k of cache. As soon as I lifted my thumb off the chip, the system crashed.

Big question now: Should I attempt to reflow the solder for that chip? I'm no expert at soldering delicate SMDs and I don't have a microscope... yet :)
Click to expand...
If you plan to do more repairs, get a cheap hot-air station. All you need to reflow the chip is flux and hot air.

You can do it with a soldering iron as well, but I would not recommend that. It's much harder to do and you will create bridges a lot, which aren't fun to fix. On the other hand, space is tight around that chip. So even with hot air, you need to be very careful.
 
You must log in or register to reply here.
Back
Top