The Digital Archivist Trainees had the opportunity to attend the “Copy that Floppy” workshop organised by the Cambridge Future Nostalgia team on October 9, which provided an introduction to floppy disk imaging for digital archivists and digital preservation practitioners. This blog post outlines some of the key takeaways from our experience, and a full guide to floppy disk imaging produced by Future Nostalgia can be found here.
A floppy disk is a type of media which stores data on a magnetic-coated soft plastic disk in a hard plastic case. Popular in the 1970s–1990s, floppy disks come in several sizes: 8-inch, 5.25-inch, 3.5-inch, and sometimes 3-inch. While the number of 8-inch and 5.25-inch floppy disks sold in this period remained relatively stable, the number of 3.5-inch floppy disks sold rose dramatically in the 1990s. The Future Nostalgia team predicts that there will be a significant rise in the number of 3.5-inch disks in future accessions, and therefore creating the capacity to image 3.5-inch disks in particular before this influx should be a priority.
Early floppy disks came in single-sided and double-sided formats, meaning that data could be reliably written on only one or both sides of the disks. It is also important to try to identify the “density”, or the way the disk was encoded and magnetised, as this affects how the disk can be read. 3.5-inch double density disks have a hole only in one corner, whereas 3.5-inch high density disks often have two. 5.25-inch disks are more difficult to identify as double or high density, and 8-inch disks are also sometimes single density. The disk manufacturer and type of computer used to write data can also affect the way the disk can be read (e.g., Mac data can be difficult to read on a non-Mac system and vice versa). Common disk manufacturers included Apple, Amstrad, and IBM.
Floppy disk drives that are compatible with the various sizes of floppy disks can be used with a “controller” to read disks on a modern computer. A controller is a piece of hardware that manages the connection between the disk drive and the modern machine, and crucially, it can read “flux-level data” from the disk. (Some 3.5-inch disks can also be read with a USB floppy drive, but these drives cannot read flux-level data, which can help recover some information when a disk is damaged or degraded.) In the workshop, we used a “Greaseweazle”, which is the most commonly used floppy disk controller, that runs with a Python package of the same name.
In teams, we each assembled a workstation to read various sizes of floppy disks. The Future Nostalgia team provided drives, controllers, and cables, as well as some test disks and workshop participants also brought in their own disks that they had been hoping to read. Excitingly, one member of my team brought in a stack of 3-inch Amstrad floppy disks which tend to be rarer than their 3.5-inch counterparts. We used a 26- to 34-pin ribbon cable to connect the 3-inch drive to our controller and a USB-C cable to connect the controller to a PC. The Amstrad drive also required us to use a flipped power cable compatible with an Amstrad drive to connect to an external 12V power source. Luckily, the expert at our table warned us this was necessary―a regular power cable or a power connection directly to the 5V-compatible Greaseweazle would’ve fried the drive or the board!
Despite everything being connected in a way that should have worked, the Greaseweazle software returned unexpected errors when trying to read the disk. Floppy disk drives and cables are fickle and will sometimes work or not work in the same set-up―it’s worth taking things apart, putting them back together, and trying again. Eventually, we discovered that the controller was unhappy with its connection to the ribbon cable and we had to instead connect it to a different port on the same cable. When that was done, the Greaseweazle was satisfied and we were able to image some Amstrad floppy disks! The first step was to take a flux image of the disk and view it using an emulator. From this flux image we were able to tell whether the disk was damaged (fortunately it was in good shape!) and how many tracks were stored on it. We then were able to convert the raw flux image data into a disk image, and extract some of the text files saved on the disk. It turned out that the stack of 3-inch disks contained research notes and bibliographies compiled by an historian of Anglo-Saxon history from whose archive they came.
My colleague Evie’s team ran into one of the most interesting cases of the day, which amassed a small crowd of practitioners looking over her shoulder while she was imaging a disk. Curiously, the flux image kept returning data for only one side of the double-sided disk. The suspicion that we left with was that the user had first written the disk using both sides of a double-sided drive, but had later overwritten data on only one side by using a single-sided drive. Unfortunately, that meant that the oldest data was lost―but it generated a lot of speculation as to how to go about recovering as much as possible. Floppy disks are complicated, and they and the machines needed to read and write them were expensive. Users found creative ways to reuse and reformat disks, which means that sometimes manufacturers’ labels are misleading when imaging disks today. The Future Nostalgia team estimated that they have success imaging disks about 50% of the time due to degradation or damage, so it was an authentic experience not to get complete data off of all of the disks we saw.
This workshop was a fantastic crash course into floppy disk imaging, and many thanks to the Future Nostalgia team for inviting us along!
