Very interesting link, thank you. Those manual scans are very good much better then the ones I found on google.

So it seems this board has a deep memory of a failed battery state going by that russian post, and like I suspect something to do with bios/odin chip preventing it posting in normal circumstances.

I got rid of the keyboard error lock by booting up with no keyboard then plugging it in, entered bios went past post did a few hard and soft reboot and keyboard working ok BUT this is all with the cmos clear jumper on, still getting blocked booting with no battery or external coin in a holder which I measured 3.2V.

going to have a break maybye for a day now.

so next on agenda is
1. check if keyboard error comes back after board fully drains after a day
2. check regulator output inside cpu socket
3. dump the bios image to pc
4? I got a barrel battery that has charge I might solder it on original pins if get bored

I don't have a post card unfortunately but I have been meaning to get one for a while they are pretty cheap

Typically the 4-pin battery connector is wired like this:

1 - external battery (+)
2 - internal battery (+)
3 - to RTC/CMOS chip
4 - ground

The 1-4 pin can sometimes be in reversed order, always check. This pinout is very common on 386+ mobos, earlier ones might be different, but usually will accept external battery on 2 outside pins. For normal operation you need a jumper on 2-3 because without it the RTC/CMOS will not get power. This jumper is not needed if external battery is connected to pins 1 and 4 though, in fact shouldn't be there if the original battery has not been removed in that case. Somethimes there might be another jumper nearby to select if the internal battery is single-use or rechargeable. There's 3-4 diodes around to make sure this works and the battery is getting charge but not powering the rest of the PCB.

Powering the system up with jumper set to CMOS clear (3-4) is very much not recommended!

So crossed a few things off my list
1. tried a working barrel battery - no difference
2. read the bios in external reader looks ok, got a crap uv light so can't erase in order to upgrade to 33V
3. Tried measuring voltage on cpu pin 6, shows 5V for every voltage config, reading voltages direct off regulator give low values like .8V or 1.5V.. as was mentioned previous its not a simple regulator so it may not show true voltage.
Tried 3V pin setup on the 5V cpu and it wouldn't post (with clear cmos pin) it might be working but can't confirm with multimeter - going leave this for now

Thanx for the external battery pin explanation, it got me thinking. Yes definitely shouldn't boot with the pin on, whatever damage is done has been done. I booted with the cmos clear pin on and as soon as it posts take pin off.

System is stable, this makes me think the cmos clear pin does something to push a successful post, once it posts it works. However a reboot won't post after pin is removed.

So trying to think logically this pin will most likely bypass some bios check... most likely the rtc. This board is funny a barrel battery and separate rtc unlike all those dallas modules 2 in 1. So right now I am starting to believe the odin chip next to the bios is halting post.

I checked the voltages on the pins and most are live guessing others are ground. A few 5v pins and few small voltage but without a pinout this tells me nothing.

So yeah thread has totally gone 180 from a voltage regulator to odin chip. This is what ill be focusing on, I feel the solution is very close

Oh yeah forgot to point out that on boards that use the dallas rtc odin those won't boot if the chip is removed

But did you try jumper on 2-3? Does it boot then?

Clear CMOS jumper only shorts RTC/CMOS chips VCC line to GND, nothing else. Is why it's dangerous to power the board with it in place, the chip will be shorting all it's bus signals to GND since every modern CMOS I/O pin has 2 protection diodes built-in. In other words it's not realy this chip but the mobo chipset and/or CPU that can get damaged by this. There isn't any voltage-sensing (for battery health status) done here AFAIK.

With so much damage from the battery spill chances are the ODIN chip might have a pin or two not making contact with the PCB. It will confuse the hell out of the BIOS/mobo so it might not post. Resolder all the pins and don't skimp on flux (clean it afterwards though). Again, a code from a test card would tell you much more.

Also, what is exactly "CPU pin 6", where you measured the voltage? I find it odd that there would be any voltage there present with the regulator transistor taken out. Maybe it's a I/O pin?

Thanx for the info, great explanation on the logic behind it. I won't be booting with the clear cmos chip no more!

Yes pin 2-3 is what I tried many times too (use internal battery) tried it with and without the working barrel installed no difference.

You are correct I need to focus on the odin chip, I did reflow those pins but pretty quickly, I need to go back to them again. There was corrosion hiding behind the chip legs too which I cleaned as best as I could. I may end up removing the whole odin chip to give everything a super good clean and new solder on the pads.

Using a heat gun would be my last choice, I think it looks doable to carefully bend each leg slightly up off the pad while heating with a iron.

Pin 6 was the 6th pin away from pin1 on the cpu (where the slant on the cpu corner is for inserting into the socket where the slant is also present)

On cpu pinout diagrams I found on the internet it had this pin as vcc among many others. With no cpu in the socket I have no idea how this 5v really works. Do you know which pin on the actually cpu socket corresponds to the voltage? I don't want to risk a 3V cpu to "test"

Use this, and note that pin numbering uses XY system: http://ps-2.kev009.com/eprmhtml/eprmx/h12203.htm

Maybe I got your explanation wrong but 6th pin in either direction (F1 or A6) from the slanted corner is not VCC. The red dots are VCC, and since 486 doesn't have split I/O and core all of them will be what the regulator outputs. The I/O on the other hand will be 5V or thereabout since the original 486 is a 5V chip, so even the later ones need to be 5V tolerant on the I/O pins.

I wouldn't try to lift ODIN just yet, simply add some flux on the pins with a tiny brush and then use soldering iron, touch every one with the tip dipped in fresh solder first. If you get a short then just clean the tip, add more flux, touch again - it'll pull the shorting solder away from the chip to the tip.
If you do lift the chip then make a photo of the PCB before you solder it back in. I do that so that I would know later where the traces go, so I can check them later even if the chip is back on PCB again.

Played a bit with the board and a multimeter, the external battery header seems to be wired exactly the way Deunan described (pin 1 is the one closest to the Odin chip).
Searched through the pics that I took while repairing the board, and there was initially a jumper on pins 2-3. I believe that my first attempts of powering the board after repair were with that jumper in place, but not 100% sure, that's why I will revisit this board when/if I'll have enough free time, to give it the attention it deserves 😀

As I mentioned earlier, connecting the CR2032 holder between [ext.bat. pin 1] and [barrel battery negative] is exactly the same as connecting the 2xAA holder between pins 1 and 4, so I cannot imagine now why it didn't work.

I couldn't find the datasheet for the Odin 12C885, but there is one available for the Dallas 12885 and I think the Odin is a pin-compatible replacement.

In addition to the above link, if you search for 486 datasheets you can find the pinouts both as "top view" and "pin-side view", the latter being useful for when you're taking measurements on the back of the board, with the CPU installed.

As for the VRM, since on your board it is a simple transistor, can't you just remove it and test it with your multimeter?

Also, if you decide to lift the Odin, maybe you could try with a "classic" Dallas 12887, since there are solder pads for it?

this is a good idea, I really like it. But will need to get a socket for the dallas to do it right, since a dallas has a internal battery that would also eliminate the need for a external battery on the 1 - 4 pins?

well with the vrm it needs a input voltage and outputs 3v 3.6 or 4 depending on the pin config so If I desolder it and say run a 5v input voltage I won't be able to see the above outputs since they are controlled with pin configurations. using resistors to alter the output, thats my understanding.

But for now looks like
1. remove odin have a look under
2. order a socket, I got a few dallas rtc modules
3. maybye solder a dallas to.top.of the solder points not all the way through just to test it before get a socket

so just had a little session.
Removed the odi, and was very happy to see a mess and corrosion but it was short lived.

Cleaned the traces they all looked fine under the layer of dirt corrosion but I tinned them anyway.

Cleaned the odin legs pads and everything around. Didn't hold back on the flux and did a few horizontal passes followed by finishes with vertical strokes on the legs.

Unfortunately still no boot, tried jumper 2-3, no jumper and external battery.

From here I have to assume the odin is dead, either shorts from the corrosion under the chip killed it or I did with booting with the cmos clear jumper.

Anyway next session ill try find one dallas chip and solder it on, can't sit around a week waiting for a socket.

quicknick maybye your pin4 on external header simply lost its ground, because like u say the shouldn't be a difference on the ground on external battery or internal

Anyway here is some pics of my adventures

so I already spent too much time on this today but I just had to know, so off comes the odin again and I found a dallas 12887A in another board that I socketed previously.
So in goes the dallas I put it through hole properly and quickly soldered, might as well do it right, right

and....

Thats right got a post with battery jumper 2-3 and feeding power from the dallas, unfortunately thats all I got time for but the odin was the problem.

I probably wouldn't narrow it down or get the momentum to try this, especially soldering the dallas so thank you for everybody contributing here.

Still early to say success because haven't done a full boot, but at least stable post now.

So the last item will be finding a definite way to test the regulator, I guess I'll try a few diff 5V cpu first and put it on 3V and if they all don't boot its a good sign

Then the sad time will come to pack it up and look for the next adventure

Nice! Frankly I'd never try replacing that ODIN without a test code telling me the mobo stops at RTC/CMOS test. It's very rare for these chips to actually go bad. But clearly it has.

Anyway, at this point I think the easiest option is just to replace the TIP and be done with it. Testing it in-board might be inconclusive and it's easier to tell if it's shorted or not once it is out. The PCB itself as well as the only other thing that could be wrong is some decoupling capacitor on the regulator output is shorted internally. If that's the case it might be difficult to find which one is it if there's several. But it can't be any of the CPU VCC caps since it works on 5V.

Deunan I went back to the cpu pinout you sent and had a look again slowly and you are correct I was not measuring the vcc pin, so I found the vcc pin this time and indeed correct voltages show up, the regulator is working.

However if I put the ground pin to ground and measure voltages from the vrm they show 5v and 4.5 from 2 outside pins, if I place ground on middle pin and measure the 2 outside pins they show 1.5v and .9v

Doesn't matter I know now to measure direct at the cpu socket that's where it matters

The system looks good and stable now did a long round of benchmarks

So I guess end of thread, hope somebody makes use of the info in the future

p.s final touch will be to socket the dallas battery

treeman wrote:

So I guess end of thread, hope somebody makes use of the info in the future

Glad to see a happy end here 😀
But please, if it's possible, update the tread's title to include the motherboard's model so it will be easy for somebody to find it.

good idea, done

Epilogue:

Re-soldered the external battery header, jumper on pins 2-3,
Soldered the CR2032 holder on the pads of the original barrel battery,
Removed D1 to prevent charging/possible explosion of the new battery.

Everything works flawless now, so I must have made some mistakes when I originally fiddled with this board. Maybe I was tired after scrubbing corrosion and repairing traces, as I tend to do this stuff in one long session, when it hits me 😀

Interesting to see that you can use a dallas chip in there instead of a odin chip + varta.

I think I'm gonna get off the odin chip and put a dallas chip in place because the odin on my board is corroded and the previous owner used his *questionable* skills there once again. As far as I remember, it works, but it looks horrible. In my opinion, that's gonna fail sooner or later.

(by the way, I just received the new VRM for my board so stay tuned on my own thread, and also I think I'm gonna replace all the expansion slots because they all have some corrosion, so I need to check there's nothing wrong underneath ...)

Its all wired up from factory but it was probably more cost effective to see a odin + a varta who knows but the option is there and its a much cleaner one I think if you are going to be desoldering things and putting a external battery.

Anyway I got the socket and installed it, bit of a pain to desolder the dallas after I already put it in but socket is the way to go. Also on this board I would say socket is recommended because things get in the way with a direct dallas on the motherboard and have to bend a few things out the way.

The holes and traces are all there so its a matter of taking off the odin and throwing the socket in the place.

From what I saw on other posts of this board it looks common for the corrosion to attack the legs of the odin.

Anyway some pictures.

quicknick wrote:

Epilogue: […]
Show full quote

Epilogue:

Re-soldered the external battery header, jumper on pins 2-3,
Soldered the CR2032 holder on the pads of the original barrel battery,
Removed D1 to prevent charging/possible explosion of the new battery.

Everything works flawless now, so I must have made some mistakes when I originally fiddled with this board. Maybe I was tired after scrubbing corrosion and repairing traces, as I tend to do this stuff in one long session, when it hits me 😀

Are you sure that it's D1 and not D2?
For me, D2 is connected to the positive pin of battery.

Crikey, what a beauty!

RiP wrote:

Are you sure that it's D1 and not D2?
For me, D2 is connected to the positive pin of battery.

Yes, quite sure that the charging circuit is comprised by D1 and R3.
Cannot check now, but I guess D2 comes between battery (+) and the RTC, so if you will check the anode of D2 will be towards the battery and the cathode towards the RTC (guessing with another resistor in series).
Anyway, it is safe to connect a CR2032 battery only when the voltage is zero across the empty socket, with the board powered on. If you measure something around 5V, either the charging circuit is still in place or there is too much reverse leakage in the other diode, one that comes between the battery and RTC - had that happen on a Soyo board that still tried to charge the battery with 5 µA after removing the charging diode, solved by replacing the (let's call it "D2" for now) germanium diode with a standard 1N4148.