The plot thickens. I've hooked a Gotek floppy emulator, with a bootable DOS 6.22 image. It boots correctly from it. After booting, I can access drive C: and run programs from it without a problem. I've also ran a scandisk on C: and no errors found. But I still can't boot from C:. 🙁

FWIW, C: is a CF card with a 2GB DOS partition on it. It works fine in all machines I have tried it on, except for this one.

Tried the secondary IDE channel, the BIOS autodetect routine finds the CF correctly, but then can't boot from it, and after booting from floppy it can't find a HDD at all. It's like it weren't connected.

Well, last test for today. I ditched the CF card, and took a 4GB Seagate microdrive. Put that in a CF/IDE adapter and connected it to the primary IDE. Booted from floppy, partitioned the microdrive with a 2GB DOS partition, formatter it and transferred DOS to it. Set the BIOS to boot from C: again, and it still doesn't. This time instead of the "j" it shows me a string "Error!" and hangs.
I guess something must be wrong with the IDE ports on this motherboard.
So for now I can only boot from floppy, although after boot I can use drive C: normally for some reason.

If anyone has any other ideas, I'm open to them.

I'm a bit confused whether a voltage regulation is on board - or not.
Please try a 5 Volt 486 CPU first.

Fair enough. I put in a i486SX2-50, reconfigured the jumpers and powered it on. I got the exact same behavior. Does not boot from IDE. Boots from floppy only.

And the same happens with a DX4ODPR100 overdrive, which is also a 5V part (has its own on-board voltage regulator) https://www.cpu-world.com/CPUs/80486u/Intel-DX4ODPR100.html

Switch out the fake memory sticks. This one has non-standard chips on it made me feel this is fishy.

Use FPM 4MB or 8MB non-parity memory sticks please.

Cheers,

I don't understand what you mean by "fake" memory. These have been tested and work fine in many other motherboards.

But here, just to eliminate any questions of "fake" memory, I have replaced the memory modules with this. I hope you approve of it. 4M of FPM memory.
Same error.

stamasd wrote on 2022-01-21, 22:23:

Well, last test for today. I ditched the CF card, and took a 4GB Seagate microdrive. Put that in a CF/IDE adapter and connected […]
Show full quote

Well, last test for today. I ditched the CF card, and took a 4GB Seagate microdrive. Put that in a CF/IDE adapter and connected it to the primary IDE. Booted from floppy, partitioned the microdrive with a 2GB DOS partition, formatter it and transferred DOS to it. Set the BIOS to boot from C: again, and it still doesn't. This time instead of the "j" it shows me a string "Error!" and hangs.
I guess something must be wrong with the IDE ports on this motherboard.
So for now I can only boot from floppy, although after boot I can use drive C: normally for some reason.

If anyone has any other ideas, I'm open to them.

Maybe there's a problem in the boot record. Try to boot from floppy and then fix using command

1FDISK /MBR

P.S. Also you can find something useful in topic:
IDE to Compact Flash as MS-DOS boot drive.

I have tried a newly partitioned and formatted drive, with a newly written MBR and fresh install of DOS 6.22, and it still doesn't boot. And the CF card which doesn't boot on this motherboard, boots perfectly fine on other systems.

(to clarify: I booted from floppy and used from floppy: fdisk, format and sys to make the C: drive bootable. Even with this newly made C: drive it doesn't boot)

Well. Have you tried to copy something from a HDD to a floppy on the mainboard? If you manage to copy then IDE controller is OK.
IMO, issue could be in mechanics of LBA/CHS or a wrong boot record.

Yes, copying from CF to floppy and from floppy to CF as well as running programs from CF works perfectly fine after boot. It's only booting from CF that doesn't work. The same applies to mechanical drives (Seagate microdrives are spinning HDDs in a CF form factor).

The above applies only to the primary IDE channel. A HD or CF on the secondary IDE channel is detected by the BIOS autodetect routine, but cannot be booted from, and is not accessible after boot at all; as if it weren't connected. I can't explain how the BIOS sees it before boot but not after.

That better on memory you are using instead, oh well on issue not resolved.

What I do is:

fdisk; delete all the partitions, create primary parition and answer yes for boot. Then format it via format c: /s/u. Reboot to make sure works, Install DOS via 3 disks or copy *.* C: while still in A: boot disk after format. On same computer you are setting up the storage on.

Cannot do this on another machine as this will render this not working.

Cheers,

pentiumspeed wrote on 2022-01-22, 19:37:

That better on memory you are using instead, oh well on issue not resolved. […]
Show full quote

That better on memory you are using instead, oh well on issue not resolved.

What I do is:

fdisk; delete all the partitions, create primary parition and answer yes for boot. Then format it via format c: /s/u. Reboot to make sure works, Install DOS via 3 disks or copy *.* C: while still in A: boot disk after format. On same computer you are setting up the storage on.

Cannot do this on another machine as this will render this not working.

Cheers,

That is exactly what I did. On the same machine. It does not boot from IDE.

Well I managed to have this board boot from the primary IDE channel. I had to blank a CF card by writing all zeros to it, then partition/format/install DOS, and then it booted. It did not boot if I did this without erasing it completely.

That is the good news. The bad news, the second IDE port is bad. Any drive attached to it is seen by the BIOS autodetection routine, but cannot be booted from. Moreover, after booting from floppy, the drive on the secondary IDE is not accessible. And if I attach drives to both the primary and the secondary IDE channels, it prevents it from booting even from the primary IDE with the message "hard disk failure". If I then remove the drive from the secondary IDE and leave only the one from the primary, it boots normally.

Oh well. I guess one working (and bootable) IDE port is better than none.

(oh and FWIW I reinstalled the "fake" memory sticks which work great on this board and pass all tests)

Next item on the to-do list: install a LDO voltage regulator so I can use 3.3V CPUs. I measured the voltages on the empty regulator pads, they are from left to right: 3.15V - 5V - 5V, which is correct for a LDO regulator of type LT1085 or similar. LT1085 is not easy to find in this day and age, but I ordered a few of a more modern equivalent (LT 1587) which should be here in a few days. It is compatible in pinout and function with the older LT1085.

(if you're curious, calculations for the LDO: Vout=Vref*(1+R1/R2); R1/R2=3.15/(5-3.15)=3.15/1.85=1.7; Vref=1.25V; Vout=1.25*(1+1.7)=1.25*2.7=3.375V, which is about right for a 3.3V CPU. This comes straight out of the LDO datasheet)

stamasd wrote on 2022-01-24, 00:03:

***
That is the good news. The bad news, the second IDE port is bad. Any drive attached to it is seen by the BIOS autodetection routine, but cannot be booted from.
***

Glad to see good news 😀
I don't remember exactly but a second IDE channel would work if IRQ 15 was routed to it. Maybe you can find something in BIOS settings about it.

The BIOS has exactly one setting related to the secondary IDE: enable/disable. It's currently enabled, of course. No other setting related to the secondary IDE.

(and that is more than the BIOS has regarding the primary IDE. There is nothing at all in BIOS about the primary IDE, not even the enable/disable switch)

Crap. I think I won't be able to run 3.3V CPUs on this board. I did some tracings and measurements, and I think that despite there being a pad and holes for a VRM, installing the regulator chip will not work. Let me explain why.

Here are large pictures of the area, both front and back of the board.

You can see the empty spot for the regulator chip, the holes for its terminals, and below it jumper J11 which is supposed to control the CPU voltage, with position 1-2=5V, 2-3=3.3V (if LDO chip is installed).

That is the theory. However, I have measured the voltage delivered to the CPU, with either the jumper in position 1-2, 2-3, and also with no jumper. It is always 5V. Yes, with the jumper completely off the CPU still gets 5V.
I have traced electrically what goes where. Pin 3 of J11 goes to the output of the VRM. Pins 1 and 2 both go directly both to 5V from the power supply, and to the CPU. I can't trace physically the copper traces that do that because I think they're in one of the inner layers of the board, but electrically that's how they are connected.

I think the idea behind J11 was thus: pin 1 connected to 5V; pin 3 connected to the VRM output, i.e.3.3V; and pin2 to the CPU. This way, with a jumper between 1 and 2 the cpu would get 5V, and with the jumper on 2 and 3 it would get 3.3V.

However, somewhere there is a short or bridge between pins 1 and 2. This makes the CPU to permanently receive 5V, and not the VRM output voltage regardless of the position of the jumper.

I have seen other boards where this is done via a solder bridge or permanent jumper, so that in absence of a VRM chip the CPU still receives voltage- and when you install the VRM chip, you have to undo that bridge.

Except that on this board, I cannot see any such bridge. There are no 2 pads connected by a wire, nothing that looks that it could be opened up. Neither on the front nor on the back. See the close-up pictures above and try to find one, I couldn't. I actually examined a much larger portion of the board, both front and back, than is shown in the pictures. Nothing.

If I cannot undo the permanent connection between pins 1 and 2 of J11, there is no way I can use a 3.3V CPU - because in this arrangement the CPU receives its voltage directly from the 5V rail of the power supply regardless of the position of J11.

I think that for some reason, on this board they made the connection between pin 1 and pin 2 via a trace on one of the inner layers of the PCB, which is not accessible.

If anyone has any ideas on what to do, I'm open to suggestions.

Here are links to high resolution pictures:
https://postimg.cc/xkWxQ2Rj
https://postimg.cc/Ln9Mwt5C
(R55 and R57 are the resistors that set the voltage the VRM gives, the rest of the small components around the area are all capacitors)

Strange. Maybe one of the other jumpers currently set on the board is making that connection? You could try measuring resistance between J11 1 and 2 with all jumpers removed. If the bridge is gone, you can then set the jumpers again one by one to see which one bridges those pads.

Also, maybe some circuit is used to switch the voltage based on the VOLDET pin of the CPU (S-4), so you may want to trace that one and see where it goes.

Other than that, I don't know why they would permanently bridge those with an inner trace... sounds really stupid, but hey, stupid things were common on boards from that era!