rebuild mbr? that's what i did to fix the ssd not booting on 486 mainboard.

Good idea! I also think it might be something related to track 0 or the MBR,

If nothin' helps, a secure erase could also help (careful with that!).

Oh and out of curiosity - does MS-DOS 7.x work fine there ?

Thanks for the tips so far!

Today I had some time to test again. I booted the system with boot-floppy, erased the SSD with fdisk, created a single largest supported size partition, formatted it, transferred the system (i.e. executed sys c:) and rebooted without boot-floppy. Result was the well known non-system disk error message. Then I continued with boot-floppy and executed fdisk /mbr (which should recreate the MBR). After rebooting without boot-floppy I again get the non-system disk error message. I reinstalled MS-DOS 6.22, and this too has no effect.

Another odd thing: fdisk reports the disk size as 8025MB although this BIOS is supposed to support disks larger than 8GB (setting: auto detect). If I then manually configure it via cylinders/heads/sectors to be exactly 8025MB then fdisk reports 504MB (although it "knows" that there is a 2047MB partition on that 504MB drive).

To me all of this seems to be somehow VS440FX-BIOS related as I can take this SSD as-is (i.e. without changing the MBR, the partitions or the installed MS-DOS), connect it to the Advanced/ML or P2B-DS and it will boot just fine... Maybe someone with a VS440FX can check how his or her board's BIOS is acting under comparable circumstances?

Just try some other HDD? I have VS440FX and it has 20GB IDE HDD, works just fine without any issues.

Did you set the partition flags "active" and "boot" ?
Is dos the first primary partition?

Yes, DOS is the first and only partition, and per fdisk it is active and bootable.

I already tried everything with another HDD (WD 640GB, briefly mentioned in the initial post). That didn't change anything. And unfortunately I don't have any working IDE HDD available.

What kind of irritates me is the fact that the disk works just perfectly fine with other systems. I can just switch cables and boot DOS with e.g. my Advanced/ML. So the disk, partitions, DOS installation etc. is perfectly fine.

I mean normal HDD for that era, like 10-20 gb size, not 640GB 🤣 😁

What is your bios ? An award bios ?
I know that award bioses had issues with HDDs bigger than 32GB at some point

It has IBM BIOS (most probably Phoenix), I haven't looked at mine with that in mind.

But throwing modern tech onto board of that age and not having a period correct (=compatible) part to test does not help much. With some smaller capacity IDE HDD you could at least eliminate board itself as a culprit and then continue experimenting with parts.

Did you successfully clear track 0 yet ? I'm not sure whether secure erase does this, too.
Here's a tiny tool which can do it, I think. Besides, FDISK /MBR is a bit different in 6.2x and 7.1.
One version only re-writes the MBR if it is detected as beeing broken, if I recall correctly.

Thanks for linking that tool. I transferred it to a boot-floppy and executed it. According to fdisk the disk was now completely empty. Then I created a single FAT16 partition, formatted it with format /s and tried to boot it. Again, the non-system disk error.

Paadam wrote:

But throwing modern tech onto board of that age and not having a period correct (=compatible) part to test does not help much. With some smaller capacity IDE HDD you could at least eliminate board itself as a culprit and then continue experimenting with parts.

Well, the SSD has been and is working on the Advanced/ML board which is basically the same age as the VS440FX. Hence, I do know that the VS440FX (or its BIOS) is the culprit. I wanted to investigate whether I need to configure something special within the BIOS, made some other mistake or the board is defective/has issues in some way. Large drives definitely seem to work... Re: Windows 98SE and Large Hard Drives

do you have any other boot options?

try "other", "removable" or "usb hdd"

you can also create a PLOP-Boot-Floppy and try to boot the HDD from Plop-Boot-Manager.

voodoo5_6k wrote:

Thanks for linking that tool. I transferred it to a boot-floppy and executed it. According to fdisk the disk was now completely empty.

No problem, it also helps to get rid of linux loaders and other mbr viruses.

Do you know what hard drive geometry the BIOS shows to DOS?
Do you have other hard drive geometry translation options available?

One possibility is that the BIOS uses 256 heads and DOS does not work with it and later BIOSes used up to 255 heads only to work around this.

But based on the 8025 MB size, it uses 1023 cylinders, 255 heads and 63 sectors. Which is a bit odd, since it should go up to 1024 cylinders, totaling some 8032 MB, but not alarming as many implementations adjust the cylinder count to one less than it should be.

If you enter the parameters manually, it may think you want to use those parameters with the drive directly, but it won't work as ATA protocol supports only up to 16 heads (but 16k cylinders), so it then forgets about the geometry translation and uses 1024 cylinder limit, 16 head limit and 63 sector limit so you only get 504 MB.

That's interesting, I also heard of the 1024 limit in DOS 5/6 (?).
Does this also affect DOS 7.1 ? I mean DOS itself, not FDISK (didn't W98's FDISK also got a patch for +8GB ?)

Btw, there was a similar post about OS/2 and the >64 MB setting. It was about OS/2 3.x beeing using outdated
PC BIOS INTs to retrieve the size of total memory, and so it ultimately got confused and reported 64MB (or less).
http://www.os2museum.com/wp/windows-nt-3-1-an … mory-detection/

Jo22 wrote:

That's interesting, I also heard of the 1024 limit in DOS 5/6 (?).
Does this also affect DOS 7.1 ? I mean DOS itself, not FDISK (didn't W98's FDISK also got a patch for +8GB ?)

No because DOS 7.1 supports FAT32 file system and can access disks using LBA (extended int 13h calls) so it supports disks larger than 8GB.

DOS 6 uses standard int 13h calls so max disk size is 8GB, and because of FAT16 file system each partition may be only up to 2GB. But it requires BIOS to use LBA or geometry translation for disks larger than 504MB to work.

I see. But in theory, depending on cluster size, et cetera, FAT16 can support +2GB partitions, too ?

Anyway, happy Halloween everyone!

Jo22 wrote:

I see. But in theory, depending on cluster size, et cetera, FAT16 can support +2GB partitions, too ?

FAT16 itself yes, but many implementations that access FAT16 filesystems (including DOS 6 and its tools, digital cameras with SD cards, etc) do not support FAT16 partitions over 2GB. They have problems supporting 64kbyte cluster size, because 128 sectors per cluster was considered as signed byte so it's -1. Up to 2GB works because it has 64 sectors per cluster (32kbyte sector size).

Jepael wrote:

Do you know what hard drive geometry the BIOS shows to DOS?
Do you have other hard drive geometry translation options available?

Currently, it is configured to "Auto Detected". Other available options are "Standard CHS" (for drives with fewer than 1024 cylinders), "Extended CHS" (for drives with more than 1024 cylinders), and "Logical Block" (LBA). The BIOS does not show what selection it actually performs when I choose auto detected. Since LBA is supported, I could try forcing the BIOS to use it as this is the latest IDE translation mode it supports.

Jepael wrote:

One possibility is that the BIOS uses 256 heads and DOS does not work with it and later BIOSes used up to 255 heads only to work around this.

But based on the 8025 MB size, it uses 1023 cylinders, 255 heads and 63 sectors. Which is a bit odd, since it should go up to 1024 cylinders, totaling some 8032 MB, but not alarming as many implementations adjust the cylinder count to one less than it should be.

If you enter the parameters manually, it may think you want to use those parameters with the drive directly, but it won't work as ATA protocol supports only up to 16 heads (but 16k cylinders), so it then forgets about the geometry translation and uses 1024 cylinder limit, 16 head limit and 63 sector limit so you only get 504 MB.

Thanks for the explanation, very interesting indeed. When I have some time within the next few days I'll try to manually configure it using the 1023/255/63 setting to see how that gets interpreted. When I first configured it manually, I used the 16 heads and 63 sectors variant.