VOGONS


First post, by quicknick

User metadata
Rank Oldbie
Rank
Oldbie
SXboard.jpg
Filename
SXboard.jpg
File size
430.27 KiB
Views
1257 views
File license
Fair use/fair dealing exception

Ok guys, so i had this board arrive to me a few days ago. It was seriously corroded - thick and dark green stuff covered many of the traces and vias around the battery on both sides of the board. I decided to try and repair it, since it's my only 386SX board and because... why not?
I tested it as it was - of course it was completely dead. No beep, no codes on the POST card, nothing.

SXthumbnail.jpg
Filename
SXthumbnail.jpg
File size
53.82 KiB
Views
1257 views
File license
Fair use/fair dealing exception

(Photo from the seller. So small you couldn't see the extent of the damage. Or the bent pin on the 387SX 😠)

It took me a few hours to do the following:
-remove all components from affected area
-clean the board, (water/vinegar/isopropyl), use fine sandpaper to remove all corrosion and solder mask, use needle to clear the vias from gunk
-reinforce a grand total of 43 vias using legs cut from scrap ICs, bent and soldered on both sides
-reinforce all affected traces using solder, and thin copper wire where needed
-replace the components, using new diodes, resistors and capacitors. I had to reuse the transistors and inductors because i had no replacement for them.

SXfront.jpg
Filename
SXfront.jpg
File size
434.66 KiB
Views
1257 views
File license
Fair use/fair dealing exception
SXback.jpg
Filename
SXback.jpg
File size
394.51 KiB
Views
1257 views
File license
Fair use/fair dealing exception

After the attempted repair, and using the original (Award) bios, the board seems as dead as before. On very rare occasions the POST card displayed some codes (EC, C4, FA, FF) - but i got no beeps and no video initializations, and the FF code seems a bit nonsensical as it was displayed only once, right away at power on.

Searching the net i found another bios (AMI) for the 386SX / OPTi 82C291 / 82C206 combo. Burned this one on a fresh EPROM, and using this i get the following behaviour at each power on: POST card cycles through a few codes (04, 05 ... all the way to 13, and then 35 where it gets stuck). The thing is that it seems rather slow at this, it takes quite some time to get to code 35 (around 20 seconds, or 40 if i short the Turbo pins), so i left it for over an hour to see if it moves past 35, but it doesn't. According to the book, code 35 stands for Display mode set complete. BIOS ROM data area about to be checked.
Memory below 1MB calculated.
.

So, at this point i'm open to any new idea, as i'm out them...

Reply 2 of 10, by keropi

User metadata
Rank l33t++
Rank
l33t++

I have done the exact same thing on a 386DX board that I have and mine is also dead after the repair.
It's a shame, your's is a 386SX board with option for cache... not that common!

🎵 🎧 PCMIDI MPU , OrpheusII , Action Rewind , Megacard and 🎶GoldLib soundcard website

Reply 3 of 10, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

That looks like it's going to be a lot of work to fix. Too bad, as this 386SX board had cache.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 4 of 10, by CkRtech

User metadata
Rank Oldbie
Rank
Oldbie
quicknick wrote:
It took me a few hours to do the following: -remove all components from affected area -clean the board, (water/vinegar/isopropyl […]
Show full quote

It took me a few hours to do the following:
-remove all components from affected area
-clean the board, (water/vinegar/isopropyl), use fine sandpaper to remove all corrosion and solder mask, use needle to clear the vias from gunk
-reinforce a grand total of 43 vias using legs cut from scrap ICs, bent and soldered on both sides
-reinforce all affected traces using solder, and thin copper wire where needed
-replace the components, using new diodes, resistors and capacitors. I had to reuse the transistors and inductors because i had no replacement for them.

Did you walk the repaired areas with your multimeter? (As in - put on some soothing music, poor yourself a drink, get a notepad, create a plan, and then walk it)

What voltage are you getting at the Vcc pin for the BIOS? Did you have leakage go under the CPU socket? Some of that trace repair looks like it goes right up to the edge.

Displaced Gamers (YouTube) - DOS Gaming Aspect Ratio - 320x200 || The History of 240p || Dithering on the Sega Genesis with Composite Video

Reply 6 of 10, by quicknick

User metadata
Rank Oldbie
Rank
Oldbie

@Deksor - although not many, there are some vias that do not cross fom one side to another, but to one of the inside layers. Five of them, under the BIOS socket, i've been able to trace and they have continuity. Of course it is still possible that another via, that doesn't show visible corrosion, might be broken.

@keropi, Anonymous Coward, noshutdown - i didnt know that cache is not so common on 386SX boards. Now that's an extra reason for me to keep trying 😀 . There are many similarities between my board and the ZIDA one below (mine just doesn't have the local bus connector and it's a tiny bit smaller, but the cache sockets and jumper settings are identical), and it seems the second tag ram socket is only needed when you have the full 128KB cache. My board came with three cache chips which i pulled out to have one less variable when troubleshooting (two W2464AK-25 and one MCM6264CP15 for tag ram, making for 16KB of cache if i'm not mistaken).
http://stason.org/TULARC/pc/motherboards/Z/ZI … OPTI-386SX.html

@CkRtech - checked all repaired tracks before powering the board, found and corrected a short between two adjacent ones. If by getting a notepad you are suggesting i need to go beyond the repaired areas (e.g. from origin to the end of each track) - you are right, but it might prove impossible. The board is packed with components and there are just too many vias under some of the chips and connectors and it's hard to follow each track.
The BIOS chip is getting the normal 5 volts, all of it's address lines have continuity to the address pins in the ISA bus and all data lines and OE pin have continuity to the corresponding pins of the two OPTi chips.
There was some small corrosion up to the edge of the NPU socket, and there might be some underneath it. But for now i don't have any tool to desolder it (but i guess the job can be made easier by cutting it in small pieces before 😁 )

Reply 7 of 10, by quicknick

User metadata
Rank Oldbie
Rank
Oldbie

Solved today. My troubleshooting attempts were hampered by two aspects:

- the original KBC is faulty
- the board doesn't post without all 4 SIMM slots populated, which I think it's kind of strange for a 386SX board.

Today I tried it for the first time with 4 modules, and it came alive... A series of beeps in the speaker, which I attributed at first to the lack of video card. But after installing a VGA, the same beeps and "A20 Gate Error" displayed in big fonts on the monitor. Exchanged the KBC and everything works. Both AMI and AWARD BIOSes report 16KB of cache, but CACHECHK says there isn't any. Speedsys doesn't test for cache/RAM speed because it needs 8MB, and I only have 1MB sticks, so I'll have to source 4MB ones.

Reply 8 of 10, by mcobit

User metadata
Rank Member
Rank
Member

That got me once before, too.
Had an 386 board with an OPTI chipset and it won't post until all 4 SIMM slots were populated with the same size of ram stick. That is not that uncommon with SIMM modules though. They are grouped in banks and a bank always has to be fully populated, so if all 4 slots belong to the same bank they all need to be populated.

Reply 9 of 10, by CkRtech

User metadata
Rank Oldbie
Rank
Oldbie
quicknick wrote:

the board doesn't post without all 4 SIMM slots populated, which I think it's kind of strange for a 386SX board.

Extremely common and "normal" for a 386 board. I am sorry that one of us didn't ask after seeing that first photo. I must have just passed over it.

Displaced Gamers (YouTube) - DOS Gaming Aspect Ratio - 320x200 || The History of 240p || Dithering on the Sega Genesis with Composite Video

Reply 10 of 10, by gbeirn

User metadata
Rank Member
Rank
Member

Not a SX board. That only has a 16bit address bus meaning it should only need 2 Simms at a time. Anyway, good job OP. I will admit I have given up on less damaged boards. If you haven’t, invest in a usb microscope, I use that and 24” monitor to blow up areas on working on in great detail. I also use it to zoom in and look under sockets and chips to see if I need to desolder them.