In the CHS definition BIOSes used at that time cylinders are limited to 1023. So you can not represent 1045 cylinders.
So one could cap that to 1023 / 2 / 40 but then you would loose capacity. So the drivers offer a translated geometry that fits the capacity:

1045 * 2 * 40 = 83600 sectors
980 * 5 *17 = 83300 sectors. This geometry fits within the cylinder representation and you loose only 300 sectors (~150 kB) compared to the actual hardware geometry.

- Kalok has no translation because it is not needed, all CHS values are below the BIOS specification limit.

About the bad sectors:
Of course a HDD can be registered in the BIOS with a lower CHS geometry. So if the Kalok is 676 / 4 /31 you can use any geometry where each value is less or equal the maximum given geometry. Of course one looses some capacity in accessing the drive.
If the actual CHS specification of the HDD can not be selected in the BIOS the user probably has used some other or some tool that patches the correct CHS translation in afterwards. Works if the HDD is a slave or secondary.

There is no quick way to find out:
Try the disk on a more modern PC where you can set 676 / 4 / 31 and just try to read all content, e.g. with xcopy to nul. If there are no errors then this was the geometry used on partition creation.
If that does not work, go through the predefined BIOS profiles in your 286 and try the xcopy to nul until you succeed.

Sometimes it is not as obvious. There are cases where the original system had e.g. a 30 MB HDD which broke at some point. It was replaced with a 120 MB HDD but for compatibility reasons it was set to the same geometry as the 30 MB HDD.

elianda wrote:

About the bad sectors:
Of course a HDD can be registered in the BIOS with a lower CHS geometry. So if the Kalok is 676 / 4 /31 you can use any geometry where each value is less or equal the maximum given geometry. Of course one looses some capacity in accessing the drive.

That's what I tried to do by specifying 615 / 4 / 26 but that didn't work. So I'm guessing it was partioned / formatted with a different CHS.

When you mean trying with "less or equal the maximum given geometry", do you mean "any CHS combo that results in less of equal the total amount of sectors (C*H*S)" , or does it have to be "less or equal amount of cylinders" AND "less or equal amount of heads" AND "less of equal amount of sectors" ?

Because it ended up working with 977 / 5/ 17 but obviously 977 > 615 so I was surprised that this worked.

elianda wrote:

There is no quick way to find out: Try the disk on a more modern PC where you can set 676 / 4 / 31 and just try to read all con […]
Show full quote

There is no quick way to find out:
Try the disk on a more modern PC where you can set 676 / 4 / 31 and just try to read all content, e.g. with xcopy to nul. If there are no errors then this was the geometry used on partition creation.
If that does not work, go through the predefined BIOS profiles in your 286 and try the xcopy to nul until you succeed.

Sometimes it is not as obvious. There are cases where the original system had e.g. a 30 MB HDD which broke at some point. It was replaced with a 120 MB HDD but for compatibility reasons it was set to the same geometry as the 30 MB HDD.

Thx I'll give that a try next time (ended up formatting the drive now and seems to work ... was not really interested in the data here but good to know for a future project.

Let's say you have a 40MB hard drive that is configured in a system using 980 / 5 / 17. Does that mean that every bigger hard drive can be installed in that system with this CHS combo and it will work exactly as the 40MB one ?