Can't see the images. Are you sure your mobo has pins 1 and 2 connected together? If so, then the pinout I described would actually have pins 2-3 reversed.
Pins 2-3 should almost never be closed for ext. battery if there is no separate jumper to disable recharging. The "charging problem" can be easily worked around by adding a single diode to the ext. battery pack so if nothing else you can just connect it in place of the original one. The diode would add a 0.3-0.6V drop depending on type but since a triple AAA pack is nominally 4.5V and the NiCd the mobo had was 3.6 you are sill pretty well matched. Most mobos would work with OK with battery volatages in range of 3V to 6V.

Deunan wrote:

Can't see the images. Are you sure your mobo has pins 1 and 2 connected together? If so, then the pinout I described would actually have pins 2-3 reversed.
Pins 2-3 should almost never be closed for ext. battery if there is no separate jumper to disable recharging. The "charging problem" can be easily worked around by adding a single diode to the ext. battery pack so if nothing else you can just connect it in place of the original one. The diode would add a 0.3-0.6V drop depending on type but since a triple AAA pack is nominally 4.5V and the NiCd the mobo had was 3.6 you are sill pretty well matched. Most mobos would work with OK with battery volatages in range of 3V to 6V.

Well, look here:

The jumper is ON when the battery is connected this is the way it is in EVERY photo of this board on the internet.

The questions is do I jumper 2-3 when I connect the external battery and which pins do I connect it to?

Deunan wrote:

The questions is do I jumper 2-3 when I connect the external battery and which pins do I connect it to?

All I can tell from that photo is that on-board battery + goes to pin 2. Or at least I'm pretty sure of it. There's still the question of what pin is GND, 1 or 4. Can you make sure? Basically do a continuity test between the edge pins and the middle of the PSU connector. Or any other suitable GND spot.

As I've mentioned earlier, all mobos I've seen would have the external battery connected to 1 and 4. And with the on-board battery removed the jumper doesn't do anything so it can stay. But you need to be sure about that pinout so test and find out which pin is ground. The opposite one would then be the + terminal.

Deunan wrote:

Deunan wrote:

The questions is do I jumper 2-3 when I connect the external battery and which pins do I connect it to?

All I can tell from that photo is that on-board battery + goes to pin 2. Or at least I'm pretty sure of it. There's still the question of what pin is GND, 1 or 4. Can you make sure? Basically do a continuity test between the edge pins and the middle of the PSU connector. Or any other suitable GND spot.

As I've mentioned earlier, all mobos I've seen would have the external battery connected to 1 and 4. And with the on-board battery removed the jumper doesn't do anything so it can stay. But you need to be sure about that pinout so test and find out which pin is ground. The opposite one would then be the + terminal.

Yeah I can do this tonight 😀

I have done a fair bit of poking around on the board. Every single trace I've tested works, I have not found a single broken trace, even the ones with minor surface corrosion (still) work fine.

JP6 is definitely the BAT EXT header.

Pin 1 is connected to Pin 3 through a series of two diodes and one resistor
Pin 2 is connected to BATT +
Pin 3 is connected to Pin 1 through a series of two diodes and one resistor
Pin 4 is connected to ground

Pins 2-3 closed, battery is online. External battery connected to Pins 1-4 with Pins 2-3 open, you get ext battery with no recharge.

JP1/2 are definitely connected to the same leg on the oscillator IC and shorting them activates two resistors, most definitely FSB selectors.

JP3/4 are directly connected to pins 64 and 66 on the CPU, which are A13 and A15 addresses. Apparently these have to do with memory addressing? Possibly telling the CPU where and how much RAM or Cache there is?

JP5 is still a mystery, likely Mono/VGA.

appiah4 wrote:

JP6 is definitely the BAT EXT header.

If there is a diode between pin 1 and 3 then jumper on 2-3 should not cause the ext battery pack to be recharged but I suppose removing the jumper would be safer. But, on the other hand, with no jumper it's possible the RTC will drain the battery even when the mobo is powered (and also the mobo might not boot at all without the ext battery in such case).
Conversely with JP6 2-3 jumpered the mobo should boot even without any battery present and RTC will be powered by 5V now except obviously the CMOS settings will be lost.

appiah4 wrote:

JP3/4 are directly connected to pins 64 and 66 on the CPU, which are A13 and A15 addresses. Apparently these have to do with memory addressing? Possibly telling the CPU where and how much RAM or Cache there is?

It's most likely to select correct size of the cache tag chip (8k or 32k). This is also tied to the size of the cache itself. By the looks of it only one jumper is supposed to be closed at a time.
What chips do you have there, exactly? I think the most sane (considering it's an SX board) option is -64 for tag and 2x -256 for data, which would make it 64kB cache. If so, I belive the jumper to A13 should be closed.

Again, a test card would help you more at this point.

Ok, so I got and tried a test card. Here's a video of the test card, where you can also hear the beep in question:

https://youtu.be/ZHDXyhDt2QY

So it stops at 13, which according to several AMI BIOS documents is "Initializing Chipset Registers". What would make it halt here, and can it be fixed?

This link here tells something else:

http://www.bioscentral.com/postcodes/amibios.htm

I see in BIOS prior 1990 (is it though?) that 13 is "System-interrupt vectors initialized" and 14 is "8042 keyboard controller checks OK". That would suggest a broken keyboard controller.

root42 wrote:

This link here tells something else:

http://www.bioscentral.com/postcodes/amibios.htm

I see in BIOS prior 1990 (is it though?) that 13 is "System-interrupt vectors initialized" and 14 is "8042 keyboard controller checks OK". That would suggest a broken keyboard controller.

Interesting, the POST card never makes it to 14 though?

I guess one of the pins on the keyboard controller that should connect to the chipset has a broken trace maybe?

The chipset datasheet is here:
82C291: http://bitsavers.trailing-edge.com/components … eets/82C291.pdf
82C206: http://bitsavers.trailing-edge.com/components … eets/82C206.pdf

Although I can't find an MR. BIOS specific datasheet here's the intel C42 datasheet: http://pdf.datasheetcatalog.com/datasheet/Intel/mXrquws.pdf

I'm currently looking at the pinouts and it's kind of over my head.. So far I figured aside fropm Pin 36 on the 82C291 which does KBDCS# it is largely not related to the keyboard controller. That is mostly what the 82C206 is for, so I guess I ought to look for broken traces between that and the battery area?

I guess there are some traces that look somewhat corroded there, C7-R44 being a top suspect. I will also check beneath the board to see which pins are connected to the KBC and test them for connectivity as well.

Any better ideas at this point?

I would think the problem is in all of the vias around the OPTi chip. The pins one the chip are corroded and the vias look like they're barely conducting.

Tiido wrote:

I would think the problem is in all of the vias around the OPTi chip. The pins one the chip are corroded and the vias look like they're barely conducting.

How should I go about fixing it then? I could just spray some contact cleaner and give them all a good brushing..

Should I try to reflow the legs of the OPTi chip? I've seen videos where this was done by adding a lot of flux to the legs and then just adding some molten solder to them, it seems to just work magically somehow? What else can be done about the via holes?

In case of via repairs, you put little wires though the vias and solder them to the traces on both sides. If anything is connected to middle layer there will be complications.

For corrosion I remove the chip entirely and then use cleaning subststances and re-tin the pins with lot of flux and heat. There's often stuff under the chip that needs to be cleaned off also.

But first you should try to measure the connectivity of various vias with a multimeter, chances are they look worse than they really are.

appiah4 wrote:

root42 wrote:

This link here tells something else:

http://www.bioscentral.com/postcodes/amibios.htm

I see in BIOS prior 1990 (is it though?) that 13 is "System-interrupt vectors initialized" and 14 is "8042 keyboard controller checks OK". That would suggest a broken keyboard controller.

Interesting, the POST card never makes it to 14 though?

Maybe I am dense, but as I understand the last successful test was 13, so it should be hanging in 14. Or is that wrong? Is 13 running, and 12 was successful?

root42 wrote:

Maybe I am dense, but as I understand the last successful test was 13, so it should be hanging in 14. Or is that wrong? Is 13 running, and 12 was successful?

Eh no if anybody here is dense that's probably me 😀 But here is the manual for the card: http://www.elstonsystems.com/dl/manual_postcard_32.pdf

It says:

"User Guide
1. Insert the post code card in PCI or ISA slot. Power on the machine. The post code will
show on display. The left Post code display monitor the real time and right post code is for
previous one. "

Which would suggest left two digits are the stage the PC is currently in, and the right two digits are the stage that it last completed.

This quote seems to confirm it:

"POST Codes
When the machine is turned on, the hexadecimal display should show the various POST codes as
the system executes (unless it has a rare BIOS that does not display POST codes).
If the machine does not boot, system POST has detected a fatal fault and stopped. The number
showing in the hexadecimal display on the Diagnostic Card is the number of the test in which
POST failed. Refer to Appendix A for a listing of POST codes. "

When I look at Code 13 for AMI it says this:

"AMI (13)DMA controller#1,#2,interrupt controller#1,#2disa- bled. About to disable
Video display and initialize port-B. Chipset initialize/auto memory detection about
to begin. Replace first memory SIMM.(13)Chipset initialize/auto memory detection
about to begin. Check first SIMM.(13) Interrupt vectors initialized."

DMA controller, Interrupt #1 #2 disabled are shared with step 12 so that means it's certainly not the problem here. So it's about chipset initialize/auto memory detection. I'm guessing it's complaining about the memory in 1st SIMM?

Just an observation. In the pictures you posted, I noticed that you only populated the two simm banks that are the furthest away from the edge of the board.

Is bank 0, bank 1, etc. clearly labelled on your board? Isn't bank 0 perhaps located on the outer edge of the board, since some motherboards require you to populate bank 0 first and then populate bank 1 and so on.

Since this is a 386SX, it should only require two 30-pin simms. However, try populating all the simm slots with the same type of memory and see what happens.

Lastly, if you have access to an EEPROM burner and you have a monochrome display card & monitor, you could try the Supersoft/Landmark diagnostic BIOS ROM (the AT version), which you can download from here: http://www.minuszerodegrees.net.

jesolo wrote:

Just an observation. In the pictures you posted, I noticed that you only populated the two simm banks that are the furthest away from the edge of the board.

Is bank 0, bank 1, etc. clearly labelled on your board? Isn't bank 0 perhaps located on the outer edge of the board, since some motherboards require you to populate bank 0 first and then populate bank 1 and so on.

Since this is a 386SX, it should only require two 30-pin simms. However, try populating all the simm slots with the same type of memory and see what happens.

*Faceplam*

For whatever reason I ASSUMED that the lower two would be Bank 0. That's always the case with Socket 3/5/7 AT boards I work with, I never thought this would be the case.. But checking Stason.org for similar Opti 386SX boards, it seems the upper two are actually Bank 0! I will definitely try this tonight. If I get a clean boot I will be so happy 😁

jesolo wrote:

Lastly, if you have access to an EEPROM burner and you have a monochrome display card & monitor, you could try the Supersoft/Landmark diagnostic BIOS ROM (the AT version), which you can download from here: http://www.minuszerodegrees.net.

I have no monochrome/hercules cards and no eeprom burner 😢

I have a board with the same chipset (different layout though), it doesn't POST without all four SIMM slots populated.

quicknick wrote:

I have a board with the same chipset (different layout though), it doesn't POST without all four SIMM slots populated.

I will populate all four tonight. There are boards with 8 SIMMs and for those Bank 0 and 1 are 4 SIMMs each so trying that also makes sense..

Ok, it's ALIVE! Not completely, but it's alive nonetheless.

Video of the POST process: https://youtu.be/cebsE0l83M4

Video of the Debug Card: https://youtu.be/-uiw0GQ-JUo

So basically it boots and POSTs painfully slowly for whatever reason and ends up with a keyboard error and no lights on the keyboard..

The questions in my mind are:

1. Why is POST so slow?
2. I guess this means there is a broken trace between the keyboard and the controller and I missed it?

On my board the keyboard controller was defective (among other problems). On yours, I see what looks like corrosion on two vias near the KBC (above R43 / D2). Check to see if you have continuity between KBC pins 1 and 39 and the data/clock pins on the keyboard connector.