Repair an old 286 Biostar motherboard

[joelgraff]

I have a 286-12 motherboard with a VLSI chipset that I'm trying to troubleshoot.  I actually have two of these - they're identical and both behave similarly.  That is, they don't boot or PPOST.  A Bios POST test card shows no POST codes.  The power rails appear to be functioning (all rail LEDs on the card light up) and of the RESET, FRAME, IRDY, and CLK, LEDs, FRAME and CLK are always lit.  RESET, generally is not (though I've seen it light up occasionally), and IRDY lights up intermittently, but usually goes dark.

The motherboards are the BioStar MB-1212V model, though I think the manufacturer of the computers was DTK - not certain.  The motherboard chipset consists of five VLSI chips as described here:
http://www.dosdays.co.uk/topics/Manufacturers/vlsi.php#VL82C100

The CPU gets uncomfortably hot to the touch within 30 seconds of booting.  I see power on all the memory chips and have swapped them out in a minimum configuration in an attempt to identify bad chips.  I've aslso toned out the traces where the battery was (it has been removed).  There's some solder mask damage from the battery acid, but the traces are intact.  I've also reseated every socketed IC (except the CPU). I also tested the power rails at the power connector and found no shorts.

I've no idea how to go about troubleshooting these - is there a diagnostic rubric or some knowledge wiki pertaining to 80286 mobo's?  I have a scope and can pull spec sheets on the chips and poke around, but at this point, it's just poking around.

My only guess at this point is the BIOS chips are bad.  The other motherboard behaves similarly.  Seems unlikely to me. 

[PKTKS]

The last photo suggest the CMOS battery leaked...

Caused corrosion which may be preventing even the clock to function..

It needs a very very careful cleaning and inspection..

Also the voltages on the BIOS chipsets... usually they indicate
if the the POST will ever succeed..

You have a dead mobo so far.

Paul

[joelgraff]

The last photo suggest the CMOS battery leaked...

Caused corrosion which may be preventing even the clock to function..

It needs a very very careful cleaning and inspection..

I noted that in my original post - there's minor damage to the solder mask but the traces toned out.  None of the nearby components show any evidence of corrosion and the minor mask damage is limited strictly to the area immediately under the battery.  I've seen/repaired much worse damage from battery leakage, tbh.  If this is the cause, it's not obvious to me what is damaged at this point.

Also the voltages on the BIOS chipsets... usually they indicate
if the the POST will ever succeed..

Can you elaborate - point me to a reference?

[fzabkar]

This is the databook for the chipset:

https://mirrorservice.org/sites/www.bitsavers.org/components/vti/databook/1990_VTI_Computer_Products.pdf

I remember hacking BIOS chips of that era. In my case I always encountered a simple 0x00 checksum. You will probably find that the two EPROMs are interleaved, ie one handles the even addresses, the other the odd addresses.

[joelgraff]

This is the databook for the chipset:

https://mirrorservice.org/sites/www.bitsavers.org/components/vti/databook/1990_VTI_Computer_Products.pdf

I remember hacking BIOS chips of that era. In my case I always encountered a simple 0x00 checksum. You will probably find that the two EPROMs are interleaved, ie one handles the even addresses, the other the odd addresses.

That's quite helpful.  Thanks!.  And yes, three AMI BIOS chips - odd, even and keyboard (2x28 pin, 1x40 pin).  Pretty typical.

[CJay]

This is the databook for the chipset:

https://mirrorservice.org/sites/www.bitsavers.org/components/vti/databook/1990_VTI_Computer_Products.pdf

I remember hacking BIOS chips of that era. In my case I always encountered a simple 0x00 checksum. You will probably find that the two EPROMs are interleaved, ie one handles the even addresses, the other the odd addresses.

3 AMI chips, the 40 pin is the keyboard controller.

Standard stuff, you need to verify the power supply rails, make sure power good is asserted, then check the usual, clock, reset lines.

That's quite helpful.  Thanks!.  And yes, three AMI BIOS chips - odd, even and keyboard (2x28 pin, 1x40 pin).  Pretty typical.

2 BIOS chips, one keyboard controller.

Standard stuff to check, make sure the power good line is asserted, the board will stay in reset if not.

Check clock and reset on the ISA bus, a decent POST card will show you these lines, if you can find something like a RACER card then it will come with a set of diagnostic BIOS ROMs which make life a lot nicer.

Check the ISA bus doesn't have any shorts to gnd or 5V.

Place the board down on a flat surface and press down on the CPU, those chips used to 'walk' out of the sockets.

[x86guru]

Check to see if only one row of DRAMs is a supported configuration. Also, are the memory chips located in the lower left SRAM/cache or are those DRAM? I cant see them well enough in the pictures to tell. If they are SRAM/cache, then maybe there isn't enough DRAM populated. If they are DRAM, then that's probably the bare minimum needed so try removing the single row of DRAM chips above and see if it will start post without them.

[joelgraff]

Check to see if only one row of DRAMs is a supported configuration. Also, are the memory chips located in the lower left SRAM/cache or are those DRAM? I cant see them well enough in the pictures to tell. If they are SRAM/cache, then maybe there isn't enough DRAM populated. If they are DRAM, then that's probably the bare minimum needed so try removing the single row of DRAM chips above and see if it will start post without them.

The RAM is 1 MB of 44256 chips.  I reconfigured it for 512K and set the jumpers appropriately to no effect.  I'm guessing the 41256's below are parity chips?  Not sure - I've never worked with a memory configuration quite like this. I tried a minimum memory config on the thought that maybe I had a bad chip.  No luck.

[joelgraff]

2 BIOS chips, one keyboard controller.

Standard stuff to check, make sure the power good line is asserted, the board will stay in reset if not.

Check clock and reset on the ISA bus, a decent POST card will show you these lines, if you can find something like a RACER card then it will come with a set of diagnostic BIOS ROMs which make life a lot nicer.

Check the ISA bus doesn't have any shorts to gnd or 5V.

Place the board down on a flat surface and press down on the CPU, those chips used to 'walk' out of the sockets.

The CPU gets pretty warm to the touch.  I couldn't pull it to check but it's well seated.  I did test resistance on the power connector between the power rails and ground - no shorts.

BIOS POST card typically shows RESET off, and CLK is always on.  Are those pins that should be checked at the CPU itself, though?

As for power good, I don't know where to check that - again, CPU?  I did swap a jumper to determine whether power good came from the motherboard or the PSU - no change.

I'm still thinking the AMI BIOS chips need to be reflashed...

[x86guru]

Can you probe out the address lines on the BIOS chips? IIRC, the reset vector is 0xFFFF0.  If there is garbage in the BIOS, then it is probably going to double/tripple fault the 286 causing it to fetch from the reset vector again so if you see the reset vector frequently on the address lines, then the CPU is probably resetting itself over and over.

[fzabkar]

The RAM is 1 MB of 44256 chips.  I reconfigured it for 512K and set the jumpers appropriately to no effect.  I'm guessing the 41256's below are parity chips?  Not sure - I've never worked with a memory configuration quite like this. I tried a minimum memory config on the thought that maybe I had a bad chip.  No luck.

They're 256K x 4 DRAMs:

http://bg-electronics.de/datenblaetter/Schaltkreise/KM44C256A.pdf

I don't think SRAM cache was available in 286 platforms. The original IBM PC-AT had a memory array consisting of 4 rows and 9 columns of 128K x 1 DRAMs. Each byte had a parity bit.

[joelgraff]

Can you probe out the address lines on the BIOS chips? IIRC, the reset vector is 0xFFFF0.  If there is garbage in the BIOS, then it is probably going to double/tripple fault the 286 causing it to fetch from the reset vector again so if you see the reset vector frequently on the address lines, then the CPU is probably resetting itself over and over.

I'll give that a try when I get a shot.

[CJay]

The CPU gets pretty warm to the touch.  I couldn't pull it to check but it's well seated.  I did test resistance on the power connector between the power rails and ground - no shorts.

BIOS POST card typically shows RESET off, and CLK is always on.  Are those pins that should be checked at the CPU itself, though?

As for power good, I don't know where to check that - again, CPU?  I did swap a jumper to determine whether power good came from the motherboard or the PSU - no change.

I'm still thinking the AMI BIOS chips need to be reflashed...

Power good is on the power connector.

Reset is good.
CLK is possibly good, difficult to know as you can't see an LED flicker at 12MHz, have you a 'scope?

Is there any activity on address/data lines?

[fzabkar]

According to this web site ...

https://stason.org/TULARC/pc/motherboards/B/BIOSTAR-MICROTECH-INTERNATIONAL-CORPORATION-286-MB-3.html

... you have populated Bank0 and Bank 1, plus their parity chips (P-0 and P-1).

[joelgraff]

According to this web site ...

https://stason.org/TULARC/pc/motherboards/B/BIOSTAR-MICROTECH-INTERNATIONAL-CORPORATION-286-MB-3.html

... you have populated Bank0 and Bank 1, plus their parity chips (P-0 and P-1).

Right - currently set to 1 MB.  I pulled BAnk 1 and populated Bank 0 for a total of 512K and swapped chips around hoping maybe I could eliminate a bad RAM chip and be on my merry way, but no luck.  It's not a RAM issue, I'm reasonably certain.

[joelgraff]

The CPU gets pretty warm to the touch.  I couldn't pull it to check but it's well seated.  I did test resistance on the power connector between the power rails and ground - no shorts.

BIOS POST card typically shows RESET off, and CLK is always on.  Are those pins that should be checked at the CPU itself, though?

As for power good, I don't know where to check that - again, CPU?  I did swap a jumper to determine whether power good came from the motherboard or the PSU - no change.

I'm still thinking the AMI BIOS chips need to be reflashed...

Power good is on the power connector.

Reset is good.
CLK is possibly good, difficult to know as you can't see an LED flicker at 12MHz, have you a 'scope?

Is there any activity on address/data lines?

On the CPU or the BIOS?  I don't know what kind of chips the BIOS chips are, so I haven't looked up the spec sheets yet.

Anyway, I have a scope, but I haven't tried looking at anything yet.

[fzabkar]

The BIOS ICs are either 27256 (32Kx8) or 27128 (16Kx8), as selected by JP11.

[james_s]

It's been a long time since I've messed with a 286 but I don't remember the CPU in those getting particularly hot. It wasn't until the faster 386 chips that I remember them getting too hot to hold my finger on, and the higher end 486 was the first I ever saw that needed a heatsink. Does the CPU get its 5V source directly from the power supply? I don't remember when I started seeing local voltage regulators on the motherboard, toward the end of the 486 era I think.

[CJay]

On the CPU or the BIOS?  I don't know what kind of chips the BIOS chips are, so I haven't looked up the spec sheets yet.

Anyway, I have a scope, but I haven't tried looking at anything yet.

Either, but if there's activity at the BIOS chips then there has to be activity at the CPU and you've verified the address bus and data lines.

The BIOS chips won't be flash, they'll be good old UV EPROM, probably 27C128 and I wouldn't rule out them having got corrupted over the years.

As you've got two of these boards pulling the BIOS ROMs from both board (label them so you know which came from which) could be a good way to check the contents, it's *highly* unlikely they'll both have gone corrupt in the same way so if they verify against each other then it's pretty likely they're good.

BIOS wasn't updated often on boards of that era because it was a pain to pull the chips, erase them, reprogram them, relabel them even if you could find a later or different version.

The 286 will be fed 5V direct from the PSU, they can get warm in normal use,  more so if there's a problem somewhere so you need to know what those address and data lines look like.

Put your multimeter on continuity and check the ISA bus for shorts to GND or 5V.

You can use the same technique to quickly check for chip shorts on the TTL and chipset, just put one probe on GND and sweep the other down the chip legs, a quick look at the PCB traces and chip data shold tell you if you expect a short or if you've got a fault, again, you've got two boards so you can compare one to the other.

FWIW, I worked repairing PC boards like this, up to Pentium and yes, it's really nice to have elegant, thought out solutions and puzzle over the datasheets but the majority of boards were fixed with simple, unsophisticated techniques like checking for shorts, reseating, resoldering and visual inspection.



 

[Per Hansson]

Both Youtube channels Necroware and CPU Galaxy have some nice videos where they repair old mainboards, it might give some inspiration.
Also this community site has info on many old mainboards like yours: https://www.ultimateretro.net/en/motherboards/1050

[PKTKS]

Also the voltages on the BIOS chipsets... usually they indicate
if the the POST will ever succeed..

Can you elaborate - point me to a reference?

Near the RTC you have a XTAL which should be inspected using  a SCOPE.
It must be functional

All the BIOS chips as mentioned on other posts should have proper working voltages..
If you have a dead chip or dead (corroded) rail it will fail the test..

Paul