My plan is to de-solder the 75LS254 , evaluate the chip, repair the traces, then reinsert a known good chip.

I've read a comment about this board that suggests something curious. Sounds like there are versions that require the parity chips and versions that don't require the parity chips.

Any idea if the difference is motherboard layout, different chipset revision, or is it just a BIOS thing?

seeing the state of the pins on the LS245 I would not recommend bothering with it at all - chances are corrosion is creeping inside it
just snip it off the board (thus making desoldering easier and less stress on the already stressed pcb area) , install a socket and a new chip and see what happens

keropi wrote on 2025-06-26, 19:31:

seeing the state of the pins on the LS245 I would not recommend bothering with it at all - chances are corrosion is creeping inside it
just snip it off the board (thus making desoldering easier and less stress on the already stressed pcb area) , install a socket and a new chip and see what happens

That's good advice. But I just got a Hakko FR-301 for my birthday with a 0.6 mm head so I had to try. Miracle device. Cleared all 20 holes in less than a minute. Just sick. But the corroded pins were not coming out so I had to cut them and give them a twist to get them out. So you were right in the end. The chip wasn't going to make it

5 Trace Breaks---

typical acid damage...
use a fiberglass pen to clean the affected area - it will reveal all copper breaks easily - the soldermask on the area needs to go so you get access to the copper to perform the repairs

Took me a while to find that fiberglass pencil. Turned out I put it with my soldering stuff. Maybe I should have looked there first! I cleaned things up best I could

Only took a minute to find my donor wire from the sacrificed floppy drive

I considered making a video of the soldering, but it would have been more slapstick humor that instructive. There are now another pair of 3mm length tiny copper wires somewhere in my carpet. Still I was pretty pleased with how it turned out. I know it works best if you do your runs "via to via", but it was awkward to do that here so I just did short runs and counted on the socket to keep things in place.

Next came the socket. I accidentally inserted a 16 pin socket at first. I had soldered two pins before I realized it. How did I miss that? Fortunately it came out easily. Once I got the correct socket, I had a hard time getting the corroded sockets to take the solder. Probably didn't clean it out well enough. Singed the board a bit, but it seemed to take well enough to risk applying some power. I checked for shorts first. Easy to remember to do that ever since that time the tantalums blew up in my face.

Next thing you know, Huzzah!

Next up-- New battery and MR BIOS!

There were too many repaired traces by the battery ground so I took it in a different direction. I can never remember the correct way to do the diode. Let me know if I messed it up.

That repair was too flimsy. It was going to break off some day and an unfortunate time.

I tried installing a "not-a-varta" replacement, but the battery via's didn't have enough metal left to take solder.

Instead I was able to use a near by via and solder the battery minus to the ground plane. Still hind of stupid, but I don't feel like it's going to fall off every time I handle the board.

I put some of my wife's nail polish on the exposed metal bits since I don't have any solder mask.

CR2032 + Diode will fall below min voltage very quickly on that kind of board. Most 286 and 386 board need 3 to 3.3V at the min. Would be better to use that with 3AAA or 3AA in a remote pack velocro'd inside the case somewhere. If you use something like Energizer Lithium AAAs or AAs they will last forever. Remember AT boards have no standby power, unlike ATX. So the battery is running the entire time it's turned off, even if plugged in.

At least in the variant of this board that I have (P.C. MATE 986-12), the battery is used only for powering the RTC. The CMOS settings are actually written to a separate 93C46 EEPROM, not to the RTC RAM.

So, if you don't mind the time/date being reset every time you could even run it without a battery.

EDIT: the RTC, at least on my board, is an HM6818A (clone of the Motorola MC146818A) and has an operating voltage of 3 to 6 volts according to the datasheet. So what kingcake says is correct, as soon as the CR2032 voltage drops a bit the RTC may stop ticking when the machine is off.

Thanks for the info. I tried to do it with a "Not-A-Varta" using a SOT-23-5 hot exchange voltage controller instead of a diode which would give me a little more voltage headroom , but couldn't get it to work because the metal was gone from the battery vias and the solder had nowhere to stick. What about stacking a pair of CR2012 cells?

I tried modifying the board to run at 20Mhz by putting in a 20 mhz chip and copying the jumpers and oscillators from this board, no luck, no post codes, no nothing.
https://theretroweb.com/motherboard/image/986 … 31607910363.jpg

I decided I'm lucky that the board is stable at 16Mhz with 2MB of chips and stopped messing with it.

Here's the build:

• The board is at home in this slim line case that seemed to be from the right time period.
• The case had a Multi-IO IDE controller w/ dark grey cables so I kept that.
• Found a home for a stray Mach8 combo ISA card that works great with windows 3.1 in standard mode.
• I printed out a face plate for the empty bay and put in a 1GB Transend CF that is correctly identified by the MR BIOS.

What should I do with the original ROM? Tape it inside the case?

Should I "retrobright" the yellowed plastic front?

douglar wrote on 2025-08-20, 01:07:

I tried modifying the board to run at 20Mhz by putting in a 20 mhz chip and copying the jumpers and oscillators from this board, no luck, no post codes, no nothing.
https://theretroweb.com/motherboard/image/986 … 31607910363.jpg

I decided I'm lucky that the board is stable at 16Mhz with 2MB of chips and stopped messing with it.

What speed are your memory chips? 60/70ns like those shown in the picture should work well for a 286-16 with 0ws or 286-20 with 1ws. I don't think you can set the wait states on these boards. So the one from that photo is probably running with 1ws, mine came with 80ns memory but it's a 286-12 so it's kind of consistent with these being fixed to 1ws.

douglar wrote on 2025-08-20, 01:07:

The board is at home in this slim line case that seemed to be from the right time period.

I have a very similar case but with a slightly different front which includes a MHz display and it says '386' under the display so I had to build a 386 in it 😆, but yeah, it's from 1991-92 so it matches well with your board. I assume you have the proper riser? It's a nice case, a bit cramped but very compact.

douglar wrote on 2025-08-20, 01:07:

Should I "retrobright" the yellowed plastic front?

If you feel it's worth the effort then it would definitely improve the looks, mine is not yellowed at all but on the other hand the metal cover has a few scratches, I'm debating whether repainting it or not, it's just so hard to find a matching paint color!! But personally my first upgrade would be to add a 5.25'' floppy drive, I feel they go hand in hand with a 286.

TheMobRules wrote on 2025-08-20, 05:48:

What speed are your memory chips? 60/70ns like those shown in the picture should work well for a 286-16 with 0ws or 286-20 with 1ws. I don't think you can set the wait states on these boards. So the one from that photo is probably running with 1ws, mine came with 80ns memory but it's a 286-12 so it's kind of consistent with these being fixed to 1ws.

The chips are labeled 60ns but only 25% have packaging that looks like Siemens packaging from that year. I bought sizeable lot of chips at a low price and then found ones that work thruogh trial and error. When there is a bad memory chip, I get to MR BIOS post code 02 or 03. When I tried the 40mhz crystal, there were no post codes. might be a bad crystal or maybe the CPU with 20 mhz markings is damaged or relabeled as well. Anyway, at this point I’ll take a stable 16mhz over YAHWC ( “yet another half working computer”)