Floppy Disk Archiving

Over the last 30 years I’ve built up a large collection of computer software for a range of platforms, including the Commodore 64, Amiga and PC; alongside this was numerous custom disks and Amiga Format coverdisks, one of which had been written to and contained some joystick experimentation code written by 8 year old me in Amos on our Amiga 2000.

Deciding it was time to archive the entire collection, I found theĀ Kryoflux. A small USB device with a floppy connector allowing both a 3.5″ and a 5.25″ drive be attached to it for reading entire disks at their lowest level.


For the disk drives, I tracked down a Panasonic JU-257 and a Teac FD-55GFR, both of which are capable of reading all 84 tracks of a standard floppy disk (some copy protections wrote data to track 80+, which while being readable on most floppy drives, not all could write past track 79).

Floppy drives with the Kryoflux
Panasonic JU-257 and Teac FD-55GFR with the Kryoflux

One of the most appealing features of the Kryoflux, is the ability to archive an entire disk without converting it to a particular disk image format. It will extract every track into a format known as a ‘Kryoflux stream file’, which simply contains the magnetic flux timings read from each track. This is useful because it allows you to extract each disk once, and later determine which file format you should be actually using, without risk damaging the disk by reading them again.

The Kryoflux software allows you to visualise the magnetic timings of each track as its read on a scatter graph, and given your selected output format, will determine if the tracks are valid for that format, empty or contain bad sectors.

Normal Amiga Disks

Here we have “Dune II” disk 1, a normal AmigaDos disk with no copy protection and no bad tracks; as indicated by the green boxes on the track/side view.

Standard AmigaDos disk
Standard AmigaDos formatted disk

If you look at the scatter graph, you’ll see three horizontal bars, these indicate MFM encoding, and looking closely at the bars, you can see vertical gaps which give the track the appearance of having ‘boxes’ across time.

These ‘boxes’ are actually the sectors in the track, and counting them shows we have 11 sectors per track.

Copy Protected Amiga disks

Moving on I attempted to archive Rainbow Islands, and found this disk was significantly different to most disks seen so far, containing only a single valid AmigaDos track (both sides), and a very different pattern on the scatter graph for all other tracks.

Rainbow Islands
Rainbow Islands, a copy protected disk

As it turns out, Rainbow Islands was protected by the infamous Copylock, created by Rob Northern in the late 1980s. Copylock contained a number of different protection schemes designed to make duplication of disks extremely difficult, the first being that the disks are written as one long track, this was possible to do with expensive disk duplication machines but impossible to achieve on a normal machine.

The first track appears as a valid ‘AmigaDos’ block and contains the bootloader which can read and decode the remainder of the disk.

Examining the valid track in the resulting ADF is how we discovered the protection scheme, as it is clearly tagged for us to see.

Inspection of the conversion of Track0
Inspection of track 0

To be able to use this disk, we have to extract it as a ‘CT RAW Image’ (which is supported by WinUAE with the SPS plugin) and after doing so, the game runs just fine.

Rainbow Islands running in WinUAE

Commodore 64

Some of the C64 disks presented a problem, old drives did not rely on the index hole to determine if the disk was spinning, and while all disks contained atleast one hole, the majority did not have one on the other side of the disk jacket.

While the drive can read both sides of the disk without flipping, the top and bottom heads are not alligned to the same position on the disk, as the C64 disk drives are not double sided, these disks must be flipped over for the head to be in the correct position to read the entire side of the disk…. doing so means our floppy drive can no longer find an index hole and believes the disk is not spinning.

At this point in time, I’ve had to settle for just one side of my C64 disks, and ultimately a number of these disks had been modified (World Records in Summer/Winter/World Games caused modifications to many disks), and/or contained bad sectors.

PC Disks

Along with my own known disks, was a box of 5.25″ floppies which had originally been in use at a local factory in the late 1980s and early 1990s.

As a curious individual, I wanted to take a look at their contents, and begin extracting them as MFM sectored disks.

Immediately it was clear something was not right, every second track on side 0 was empty, but every track on side 1 was in use. The resulting IMG file also reflected issues, as most files contained huge empty sections, and one file even contained parts of what should of been a directory listing.

Something was not right with this disk


As it turned out these disks are only 40 track. The default in Kryoflux for MFM is 80 tracks, after creating a custom format using 40 tracks, our disk image was created sucessfully.

Looking Better
MFM 40 track disk correctly interpreted

Curiously, I noticed that the unused uneven numbered tracks on side 1 contained a scatter pattern very similar to the even numbered tracks . After some investigation, it turned out the uneven numbered tracks on side 1 contain the exact same data as the even numbered tracks on side 1, but the uneven numbered tracks on side 0 are completely blank.

I’m still unsure as to why this is, and the only information I have regarding the original hardware is that they where WANG PCs; however there was also 3 disks amongst the collection which have stranger characteristics.

Strange PC Disks

The even numbered tracks on side 1 of these three disks are completely blank, leading to the resulting disk images to contain corrupted files (huge null areas). As my mouse moved across the tracks, I happened to notice that the uneven numbered tracks on side 1 contained what appeared to be a the MFM pattern.

MFM pattern on the uneven numbered sectors
MFM pattern on the uneven numberedĀ tracks of side 1


The first potential fix was fairly obvious, rename the tracks on side 1 by reducing each track number by 1 (deleting track 0 in the process).

As it turned out this approach was succesful, leading to 3 valid disk images, all containing usable files… Perhaps the drive head which wrote these disks was missaligned by one track? but that also doesn’t explain why the other disks have every track on side 1 in use.

Now a valid side 1

Disk Contents

These 3 disks contained files created by Professional Write; an early DOS based word processor, which was able to read the files found on the disk.

Professional Write for DOS, listing the contents of a disk


Most of the contents of the other related disks are Lotus 1-2-3 spreadsheets containing information about rejected parts, nothing exciting, but still interesting from a historical perspective.

Lotus 1-2-3 Pie graph of engine troubles

Disk Quality

These 5.25″ disks had not been touched in 25+ years, and spent the majority of that time in a garage roof which regularily reached temperatures over 40c during summer, yet it was possible to archive every disk without error.

Its seems fair to say that Nashua is a high quality floppy disk.

Nasua 5.25″ Floppy Disk


Unfortunately a number of my 3.5″ floppy game collection was not as fortunate, and numerous disks contained bad sectors; many of these have only been aquired over the last decade, so their history is not known.