SpeedStor - Example of MediaAnalysis overwriting 'dynamic' information


Below, by 'geometry', I mean the count of cylinders, count of heads, count of sectors per track.  For example, the Seagate ST-225 drive has a geometry of 615 cylinders, 4 heads, and 17 sectors per track.

I have the following:

•  IBM 5160 (IBM XT)
•  Western Digital WD1002A-WX1 controller, containing F300 Super BIOS   (Some WD1002A-WX1 do not have this BIOS)
•  An MFM drive of 612 cylinders and 4 heads


The F300 Super BIOS on the WD1002A-WX1 gives me the ability to 'dynamically' low-level format the drive, and that is what I am about to do.

If you do not know what I mean by 'dynamically', then read the information at here.

I boot to a DOS boot diskette.

I execute the low-level formatting code in the WD1002A-WX1, choosing to dynamically format the drive.  The code asks me for, in amongst other information, for the number of cylinders and heads that the drive has.  I inform the code that there are 612 cylinders and 4 heads, and the code starts low-level formatting the drive (a process that takes a few minutes).  After the last track has been low-level formatted, the code writes/stores the drive's geometry information to a track on the drive, a track that the drive reserves for that geometry information (the track referred to in the following as 'track-used-to-store-geometry').  Then, the code reboots the computer.

When the computer reboots, the initialisation code in the BIOS ROM of the WD1002A-WX1 reads the geometry information stored on the aforementioned track.

The boot process continues and I boot to a DOS boot diskette.  I then run SpeedStor 6.03

SpeedStor 6.03 shows me that the drive has 612 cylinders and 4 heads, information that SpeedStor indirectly gets from the WD1002A-WX1.

So, all okay so far.

My MFM drive is old and I know that, over many years, it has developed a few bad areas of the platters.  I decide that before I go through the partitioning and high-level (DOS) format processes, that I will get SpeedStor 6.03 to perform a 'media analysis' of the drive.  That will do a good job detecting defective tracks (track is a combination of cylinder and head).  I will then get SpeedStor 6.03 to mark those defective tracks as 'bad'.

NOTE:  'Media analysis' destroys high-level data (partitions, DOS structures, files, etc.).  Because I have nothing on the drive yet, I am not concerned by that.

In SpeedStor 6.03, I navigate to the 'MediaAnalysis' functionality. I choose the default options, which is ALL cylinders and ALL heads.  'MediaAnalysis' then does its work, a process that takes quite a while (over an hour for my drive).

'MediaAnalysis' finishes, and I see that on my drive, it has detected 35 tracks that it considers as defective.  Detection only; they are not yet maked as 'bad' on the drive.

I use 'LockDefects' to mark those defective tracks as 'bad' on the drive.   [ManualSetup] [Initialize] [LockDefects]

I restart the computer, booting to a DOS boot diskette.

I run SpeedStor 6.03

PROBLEM:  SpeedStor 6.03 now indicates that the drive has 4096 cylinders and 64 heads !!!!

What happened?  Well, the 'MediaAnalysis' functionality of Speedstor 6.03 overwrote the track-used-to-store-geometry.  That was unexpected.


Experiment

I experiment.

I note that the 'MediaAnalysis' functionality can be performed on only part of the drive.  I perform multiple runs of {low-level formatting of entire drive, followed by media analysis of part of the drive}.

I discover that the track-used-to-store-geometry is not overwritten when I exclude {cylinder=0, track=0} from the media analysis.


Way forward for my WD1002A-WX1

First, I low level-format the drive.

Then, when I run the the 'MediaAnalysis' functionality of SpeedStor 6.03, I exclude {cylinder=0, track=0}.
Then, I use the 'LockDefects' functionality to mark the defective tracks found as 'bad' on the drive.
Then I partition then high-level (DOS) format the drive.