A couple of things I'd try at that stage:

- Disassemble the and clean connectors thorougly. Also remove and reseat any socketed chips.

- Verify PSU voltages under load, i e during boot attempts.

Try with another PSU and, long shot, check-replace the 14.318MHz crystal.

Do you have a different system you can try the hard drive in?
(If it needs it's controller - and that is a card, perhaps you could move the card too?)

Might be worth knowing if the hard-drive is actually corrupted -- if you can put the
HD in another system can you boot it? - does it get the same errors?

If the HD is actually trashed, perhaps you could reformat it in another system and put
a simple DOS install in it - if you could get enough to actually boot the system you are
trying to fix, being able to run some diagnostic code on it would really help figure it out!

Or - maybe you could put a different controller and drive in the bad system ... just to be
able to run some diag code.

Thanks for the input

kaputnik wrote on 2025-03-26, 09:27:

A couple of things I'd try at that stage:

- Disassemble the and clean connectors thorougly. Also remove and reseat any socketed chips.

- Verify PSU voltages under load, i e during boot attempts.

I've already tried plugging/unplugging both floppy controller cards in all 16 bit isa slot on the motherboard, no dice.
As for the power supply, I don't have another AT power supply to swap with but I did check voltage and ripple under load with scope, 12V and 5V are on point and ripple is under 50mV so I think power supply isn't causing the issue.

konc wrote on 2025-03-26, 10:08:

Try with another PSU and, long shot, check-replace the 14.318MHz crystal.

I already checked the power supply and it's within specs, as for the crystal I did see someone else mention it in another post, but the motherboard in that post is a 486 with integrated floppy controller iirc. On my motherboard there's nothing integrated. Nontheless I tested the oscillator and it's outputing a clean 14.3mhz signal so I guess it's working? Also tested all 3 other crystals on the floppy controller with the scope, all outputting the rated frequency.

DaveDDS wrote on 2025-03-26, 13:54:

Do you have a different system you can try the hard drive in? (If it needs it's controller - and that is a card, perhaps you cou […]
Show full quote

Do you have a different system you can try the hard drive in?
(If it needs it's controller - and that is a card, perhaps you could move the card too?)

Might be worth knowing if the hard-drive is actually corrupted -- if you can put the
HD in another system can you boot it? - does it get the same errors?

If the HD is actually trashed, perhaps you could reformat it in another system and put
a simple DOS install in it - if you could get enough to actually boot the system you are
trying to fix, being able to run some diagnostic code on it would really help figure it out!

Or - maybe you could put a different controller and drive in the bad system ... just to be
able to run some diag code.

Thanks for the suggestion, that's actually what I tried last night, I moved the hard drive and the mfm controller to my pentium 3 system, disabled the onboard ide controller and the mfm disk actually showed up in DOS. Low level formatted it with speedstor and installed dos 6.22 on it, boots the pentium 3 system just fine. Moved it to the 386 motherboard and now it refuses to boot and throws "missing operating system" error right after the post screen. Also tried the known good ide hard drive with windows 98 from my pentium 3 build, connected it to the multi io board's ide port, set CHS in bios and same "missing operating system" error. Now that's the part further confuses me, because the mfm hard drive used to work on the 386 motherboard (although the dos installation is corrupted), and it boots the pentium 3 system just fine after low level format, but moving back to the 386 motherboard it suddenly stopped working?

Also did more testing with FDD booting, 90% of time it will simply freeze at blinking cursor right after post screen. Interesting part is, 10% of the time it will jump to random screens within BIOS setup. For example, after the post screen it shows a summary of system info and a blinking cursor, floppy then spins up and i can hear head clicking once or twice. Instead of saying "Starting MS DOS", the screen goes black and advanced BIOS setup screen shows up. However the BIOS setup isn't functional and selecting anything will freeze the computer. So I think what's happening is the code read from floppy isn't being executed, and CPU jumps to some random locations in RAM, which explains the freezing and BIOS screen showing up. Could it be bad RAM? RAM test in post screen passes every time. Already tried booting with only 1 bank populated and mixing different sticks from 2 banks, same result. This motherboard uses SIPP ram sticks and they are hard to find..

It almost sounds like the BIOS is corrupted ... self tests should detect this, but you never know...

Can you get a binary image of this actual BIOS somewhere? It might be worth comparing the BIOS
installed on the board with the actual "correct content" and see if it all matches?

I assume you've changed memory, gone down to minimum etc. (you don't need much
to boot bare DOS)

There's not much that can kill such a system in such a way that it seems to power-on and POST
correctly but can't boot known good drives.media

BIOS rom

RAM

BUS

Is this through a video card (or can it be) - video "looking right" would be a good indication that
the BUS is transferring data correctly.

It reading the original HD enough to look for COMMAND.COM also suggests that it's able to
"talk to things".

Have you tried manually moving the floppy stepper to 1/2 way out (with power off) and
then power on/POST/boot... does it step back to track 0 - this would indicate that it's correctly reading the
FDC status and sending good commands to it.

The FDC and HD getting selected when accessed (to boot) suggests that data lines are pretty good.

IRQs

Failure to receive and IRQ when expected, or getting one when not expected can cause things like
this... If you have a scope, I'd look at the IRQ output from the FDC and input to the CPU ... they
should match (CPU may get other ints from things like real-time-clock), so I'd trigger in the FDC.

Also do these tests with everything "out" that can be .. no HD, simple video etc.

Sorry... just re-read your original and see that you've already verified BIOS content.

You mentioned recently "BIOS setup not working" - was that only when the boot attempt randomly
jumped into it - does BIOS setup work as expected when you manually/intentionally enter it?
If so, BIOS would be calling internal subroutines meaning stack (RAM) is working...

If it's actually reading the data from drives, sector checksums should be good or it wouldn't
try to execute...

It could be that a prolonged burst of memory access (like writing a block or executing) is enough
to cause a failure while single/small accesses work.

You've tried all BIOS settings that might change memory access speed?
Any change the memory might be the wrong type/speed?

I suppose it could also be DMA failing.

DaveDDS wrote on 2025-03-26, 15:53:

Is this through a video card (or can it be) - video "looking right" would be a good indication that the BUS is transferring data […]
Show full quote

Is this through a video card (or can it be) - video "looking right" would be a good indication that
the BUS is transferring data correctly.

It reading the original HD enough to look for COMMAND.COM also suggests that it's able to
"talk to things".

Have you tried manually moving the floppy stepper to 1/2 way out (with power off) and
then power on/POST/boot... does it step back to track 0 - this would indicate that it's correctly reading the
FDC status and sending good commands to it.

Yes, video is normal without any visual glitch, in post screen and bios. It's a ATI VGA WONDER 16 VGA card. Just for the sake of it I also tried booting the motherboard without VGA card, same issue. After post ok beep, floppy spins and head clicks once then stopped, so it's not booting even without VGA card.

I didn't try to move the head manually, but during every post the floppy drive makes some noise (head stepping back and forth), if I unplug power cable for floppy drive it will cause fdd controller error, so I guess both fdd controllers are working.

BIOS does work normally when I press del during post, but when booting from floppy and it jumps to bios, the menu will not work. Yes I've dumped and verified both bios chips against the one uploaded to theretroweb, same checksum so i guess bios can be ruled out.

I did play around with memory/bus speed settings in bios, tried to increase/decrease those values but made no difference, still won't boot

I think I'll trace and check IRQ signals on ISA slots tonight, without schematics this gonna be a fun one..

What exactly it does will depend on BIOS.
Just checked with a couple DOS systems I have readily available, both with Phoenix/AWARD BIOS.

With no floppy in the drive, you can hear a 1.44mb drive move slightly .. perhaps 1 step, then notices
the drive is at track 0 - and moves on the the hard drive.

With a 360k drive, it doesn't make any noise/move when the drive is already at track
zero.

One system is a "breadboard" system (built on a piece of wood) on my workbench which I can easily
attach./remove drives. I don't have an open 1.44, but it I move a 360k out to near middle, it
obviously steps enough times to get it to track 0.

The other systems doesn't have easily accessible drives, however if I use ImageDisk Align/Test tool
I can position the head to center/far-end then reset (CTRL-ALT-DEL) the system. during POST will
step the drive enough times to get to track 0.

These tests should be done with no floppy inserted, and the hard drive installed
and set to try and boot - as soon as BIOS determines there's no floppy it should move
on the the hard disk.

You mention "stepping back and forth" - with no diskette inserted, it should go to track 0
and "give up"fairly quickly. You might hear one or two small attempts to step, but nothing else.
If you are getting more than this, perhaps track-0 sensor problems? (but you've tried different drives
and controllers) - (I know this is a "stupid" long-shot - but did you change the cable?

The idea in moving the head well away from Track-0 is to see if the BIOS is correctly talking to
the FDC - basically to test stepping and track-0 detection.

I don't know how much the BIOS reads from it's own screen ... perhaps we have something
loading the bus or otherwise interfering with data transfer during reads - but data gets through on write
(otherwise video on bus should be crap).

Really weird problem - does the system happen to have an on-board serial port? or can you put one in on a card?
If you really have to, we might be able to make a modified BIOS rom that instead of
trying to boot - launches a very simple monitor/debugger talking to the serial port...
That would at least let you read/write, load in some test code and actually see what the CPU
is seeing - Not easy (I think I only ever had to do this once) - but if you can have the BIOS at the
right place, everything should be set up before your code launches.

I've got MON86 a pretty powerful 8086 monitor with lots of useful tools in about 4k.

I also have HDN86 a "Hardware Debug Monitor" which gives you very basic read/write/display
abilities and requires very minimal hardware to be actually working (It doesn't even need any RAM)

Depending on BIOS, the system might see and launch a ROM on a network card to perform
network boot (I think this was the way I put a monitor into a system once) - much easier to do, I
have some cards where the netcard has a std 28pin EPROM socket, and using a ROM emulator
you can try it till you get it working without having to burn-and-burn ROMS.

My drives are 3.5/1.44mb and 5.25/1.2mb, took off the cover for the 3.5 inch one and I can see both drives move their heads from track 0 to inner most position and back to track 0 during the post, which i think is normal (same thing happens on my pentium build).

If i press keyboard at "insert system disk and press any key" prompt, the head just nudges a little bit then goes back, again i think it's normal. If a bootable floppy is inserted, floppy spins up and head nudges a little bit again (probably trying to locate track 0), then system hangs. Sometimes the floppy drive will keep spinning forever with activity light on, sometimes it stops after a few seconds, seems random.

Now aside from floppy disk drive and controller, I think I found something more interesting. Since floppy drive is having issues, I checked IRQ 6 pin on ISA bus with a oscilloscope, and I can see pulses issued by the controller. Following the trace, all IRQ pins go into the 82C206 chip, which according to datasheet is a DMA, IRQ, timer and RTC controller. IRQ 6 is pin84 on the PLCC package, and sure enough i can't see anything on this pin. Pulled this chip out again and cleaned all pins with contact cleaner and eraser, put the chip back in the socket and verified continuity, everything seems good. However, now the floppy drive won't recognize any disk, whatever I put in the floppy drive it will always say "diskette boot failure, insert system disk and press any key".

Booting from floppy drive seems a dead end, I decided to go down the hdd boot route again. Took out the old mfm drive and plugged it into my pentium build, low level formatted it again and reinstalled ms dos on it, made sure it can boot the pentium motherboard properly. Moved it to the 386 motherboard, changed CHS in bios, and right after post screen I can see drive activity led flashes, seems like it's trying to read track 0. Then it stopped and threw an error on screen that says:"Disk I/O error, replace the disk and press any key".

Now this is interesting, since the mfm drive and controller can boot my pentium motherboard properly, the disk i/o error must be caused by something on the isa data bus. Floppy disk booting problem might also be caused by this isa data bus issue, not floppy controller, floppy drive or ram.

Is there any guide on troubleshooting isa data bus issues like this? This is an early 386dx era board with a 8-chip chipset, and I've read online that these chips are unreliable hence all of them are socketed for easy replacement should they fail. If any of these chip went bad this board is toast, might as well get a later 386sx board for this 386 build. At least the 20mhz 80386 and 80387 processors are good, I think...

Some BIOS do exercise the drive by stepping all the way in/out during POST.
Not a "feature" I like particularly because drives have a Track-0 sensor, but no Track-max sensor...
So if a 40-track drive is misconfigured as 80-track it will bang against the stop
many times at startup.

Does sound like it might be an IRQ-DMA problem... Hopefully there is some board/trace damage
you didn't notice as those system chips can be hard to source/replace (you're lucky they are socketed)

Checking data bus is hard if you can't get code running ... my HDM monitor has nice
fast repeating read and write functions which let me put known bit patterns on the bus
for tracking with scope.

Well it's hard to come by vintage hardware locally from my area, and I really don't want to spend 100 plus shipping to buy one from ebay that are being sold for parts.

Already did some basic troubleshooting with multimeter and scope, can see waveforms on all data pins, no short on them and diode readings are all the same. Don't have a logic analyzer and i guess that's as far as I can go.

Will definitely check traces again, but doubt I'll find anything, this motherboard doesn't have NiCd battery and there's 0 corrosion on it. Maybe if I'm lucky I can find another board with the same chipset and swap out those chips. Again thanks for the help

Symptoms like thic can be caused by a missing "DRQ2" line, if the connection between ISA-bus and bus- / peripheral-controller is interrupted or if the controller is defective. If someone pulled an ISA card while the system was running, shorts between -12v / -5v and DRQ2 can occur and damage the controller´s interface circuit.