60ns 4MB sticks, nice. It is possible one of those chips is faulty, that would explain why it counts different values and perhaps why it was sold as 4MB set - maybe it's how it was detected in other mobos. However I did have some of these sticks with cracked solder joints between chip pins and the PCB, that's easy to repair with soldering iron and some flux. So inspect and make sure it's all soldered properly, and also check if there's any deep scratches that can break connection, etc. There are memory testing programs for DOS, like Gold Memory, try that as well.

As for the floppy drive, I had some weird issues with those when the CF card I tried to use in place of HDD was somehow not liked much by the BIOS. I know, weird, one would expect the error to be HDD related then, but no, the BIOS reported FDC failure and indeed it wasn't working properly (but in truth neither was that CF card in the end). Solved by using different brand card (both were 512MB). If your BIOS also has 504MB limit try a smaller HDD or different card to see if that fixes the issue.

Progress! The IDE controller is working now and floppy / HDD are working both. I managed to boot a dos boot floppy, and it could read from the hdd. However, the board is not stable. It boots quite reliable, hanging only sometimes at "13" or displays a "keyboard error", but most times it works. But it won't work for longer than a minute or so, and eventually it hangs. Everything stops, nothing shows any reaction. The cursor keeps blinking (but afaik that's done completely on the vga, so a halted cpu wouldn't affect that, right?) but I can't even toggle caps lock and the floppy stays in its last position and repeats that. I tried the other ram kit, no change. Sometimes it doesn't even reach the bootloader and dies right before showing the soft errors (which is every time the bad cache).

What could be the problem here? 🙁

I bought a CF card as replacement for the HDD, do I have to think of anything? What do I set in the BIOS? (and I bought an CF to 2.5 inch IDE so I have to wait until monday, when the correct one arrives -.-)

If two sets of memory are not stable (and assuming you tried a conservative settings in BIOS, like 1WS for R/W) then I would focus more on cache. These do die too sometimes, you noticed they can run pretty warm (again, I'm assuming you've tried slower settings in BIOS for cache as well).

I deactivated Cache in BIOS, as it won't post with the modules installed. I tried swapping them, it's always the upper module that's getting hot. So it seems to be the board or the socket and not the module.
After I found the option to deactivate cache in the bios, it seemed to be more stable than before, now booting always into the boot disk, but eventually still dies, at the latest when I tried to start the norton commander on the hdd. While it's responsive, the speaker outputs a whirring sound, which stops when the system halts.

Another thing I don't like: if I bent the board at the slightest to its behind, it wont boot. (--)
I got mire stability, when I layed the upper part on a small thing, so it bends the other way.

Edit: the board will hang immediately if I lay a finger on one of the cache sockets causing the board to bend slightly.

Then you have a cracked solder joint, broken trace or a via. These are pretty difficult to find and do not assume it's a problem located in the place you apply pressure to - reason being this is not where the PCB flexes actually. Get a good magnifying glass (or an inspection microsope if you have access to one) and go over all the pins of those two big mobo chips. Then the bottom, as I've said before, look for any cracked solder and also deep scratch marks.

As for that one chip being hot - it's either something shorting out or it's not driven properly by the chipset (again that could be due to missing/bad connection).

Will do! I assume, this also causes the instability of the board?

I did a first look onto the connections and did not find any obvious broken thing. But I do not have any experience with this, so didn't know where to look.

My CF card arrived (2GB AGFAPHOTO 120x) and I cannot get it to work.
Adapter is a DELOCK CF to IDE adapter.
If I set it to master, the board won't boot, displaying either garbage on the screen, or dying at "40".
If I set it so slave, it will boot, but I dont know the correct settings (cylinders, heads and sectors) so cannot access it.

When I access the CF with a card reader on my PC, everything works.
I copied a working DOS image to the CF with dd and made the disk bootable with "CardTricks".

I tried the adapter in another PC (Asus P5Q, Intel Core2Quad, *maybe* 2009?)
It's "Marvell"(? cant remember xD) IDE controller finds the card as PATA CF-Card 2GB, the jumper setting seems to be irrelevant.
It only said at first, that the disk was invalid.
When I booted from the DOS floppy, it could access the disk, but would only show wrong data. It did read something from the CF, but that seemed like random bytes from it.
"dir" showed different information every time, and eventually only "File not found", the disk label was every time random ascii.

After many tries with "sys c:" and "fdisk /mbr" I managed to access the card correctly! Every information was there, and after I booted from the floppy and executed autoexec.bat, it was just like the system on the card. But when I tried to boot directly from the card (which worked!), it would not find anything on the card, dos booted, said file not found for autoexec, and I get a basic shell with no functionality than dir and ver, dir showing some of the files and folders correctly, and the second "dir" ends with "file not found".
scandisk with surface scan started from the floppy doesn't find anything.

As I intended to find out the configuration I tried many tools which claim to find the settings. (ideinfo.exe, idediag.exe and idthatide.exe) On the P5Q, none of them worked. IDETHATIDE failed with a "Ready error", the disk never got ready, idediag doesn't find anything, ideinfo.exe said, the bios reportet 1 drive, hardware scan didnt find anything. Could be the "too new" PC.
So I booted the floppy on my 386 with the cf adapter connected as slave, bios was set to empty (when I set values like 4095 cylinders, 16 heads and 63 sectors, it dies with HDD controller failure).
One time, the board was ok long enough to execute ideinfo.exe, and the screen instantly turned to garbage (will attach a foto of it, from the vga bios "garbage version"), the only line I could read was "sectors: 63", so that did work to some point! After it finished, I got a garbage version of the command prompt and the board froze 🙁

Is the adapter broken? The reviews on amazon stated, that it would not work on slave setting, scrambling data in this mode, but master would work 😒
https://www.amazon.de/gp/product/B001C852NO/r … 0?ie=UTF8&psc=1
As I wanted to have only one IDE device on the board and it was the only one which was cheap and available, I wanted to try...

You likely won't get that card to work properly on this system. Pre-PCI mobos have BIOSes that can only use CHS mode and that places a limit of about 504MiB - this is because the BIOS call parameters can only accept up to 1024 cylinders, 16 heads and 63 sectors (and each sector is 512B). Newer BIOSes can have certain extensions (so-called LARGE type in CMOS setting) and then LBA came and fixed most of these problems for a long time.

If you don't have auto-detection option in BIOS that offers LARGE then you're ouf ot luck. You need a smaller card, pick a 512MB one - these don't even have the 512MiB and probably will offer just shy of 500, so you won't hit the 504 limit. Now you might be thinking to just enter lower numbers in CMOS and use this card partially - that might not work, some HDD controllers will trip the BIOS anyway. Even if it does, you might not be able to read the contents propery under a current OS, so it's probably a waste of time to even try. But feel free to experiment.

Only once the BIOS is happy with the card can you format it under DOS, do FDISK /MBR and transfer the system files. It won't work properly otherwise.

Ok, I will try a smaller card.
So does this mean, that when I accessed my 2GB HDD it just saw the first 500MB? I could enter the correct values and the bios calculated the correct size!

Unless your BIOS supports the LARGE extensions it is simply not possible for DOS to access anything above 504MiB. But if it accepts such values, and boots, maybe you can try software solutions. There are TSRs that will patch the BIOS code, these need to be loaded as the very first thing during OS startup. Usually it also means limiting the first HDD partition to below the limit to make sure anything you put there will actually be accessible during boot. But all this take some work, skill, and DOS memory - so perhaps a smaller card will just suffice.

I see that the board you are using has seven 16bit ISA slots. Unless you have plans for every single slot, I'd recommend the easiest and most compatible solution to get modern, easily available drives and CF cards working is to find a 3Com 3C509 network card and write the XT-IDE BIOS to an eeprom that you then place in the socket on that network card.

You don't have to worry about settings in your BIOS, or sizes of CF card; you could go and buy (and use) a new 128GB card at that point if you like - and drag and drop content to/from it in any modern PC.

If you're not up to the task of writing an eeprom and configuring a network card, then there are pre-made XT-IDE interface cards available; you don't have to get one with a CF interface if you don't want one.

Sorry to necropost on this, but I think I've identified the chipset that Morse labeled as 91A320/91A321: an OPTi 82C391/82C392 combo. I found this out when I tried an MR BIOS for the OPTi 82C391... plugged it in and it came right up, and runs great!

That dirty TAG socket requires a funky 24 pin 16Kx4 SRAM in order to enable write-back cache mode. A Logic L7C165PC-20 is what I am using on mine. With it, the off-cache access drops from 110uS to 71uS according to Cachechk.

Sorry for this big, big necro, but I want to post an (final) update!
After some time has passed, I got to know some people at the faculty of electrical engineering, who have a small lab at the uni, where they could help me out.
The problem was, that the board was booting and freezing after a minute or so, and that the cache wouldn't work (and generate much heat when put in).
We found some corrosion in the socket of the keyboard controller (the IC itself was fine!), so we desoldered it and replaced it. This indeed fixed the problem of the freezing system! So I was able to install an OS to the CF Card. Which was awfully slow, as there was no cache installed. But working! *-*

The cache was easy - we analyzed the problem with the cache, there was just a bad chip (Chip enable hanging at 2.3V or so, so between high and low for TTL). Replaced that, the other cache chips were working. The speed improvement was more than measurable xD

Fun fact: The soldered 386 was not broken! We removed the socketed one, and it still worked. I wonder, how to switch the board to the socketed one, as there is no obvious jumper...

So now I got a fully working Morse M3 with a 80386! Thank you all for your help and have a nice day! <3