My first week at MITH has been marked by a firsthand encounter with a tradition of 1980s personal computing and data storage: the flippy disk. Before I get into that though I thought I would briefly introduce myself.
I’m a graduate student at UT Austin’s School of Information. Although I originally set out to study book and manuscript preservation, my interests soon turned to digital media preservation and curation. Specifically, newer media that uses unique digital affordances are of interest: video games, interactive narratives, etc. Not surprisingly legacy media and vintage hardware can play into this quite a bit.
My background is in the humanities, studying literature. But like a lot of folks here, I have at least some technical aptitude when it comes to computing. I fondly remember trying to program an adventure game in QBASIC, using nothing but text prompts, if…else conditions, and GOTO. It was unashamed spaghetti code, but fun.
On to flippy disks. The majority of 5.25″ disks were design to be used on single side only, but users could produce another notch on the opposite edge of the manufacturer’s own. When the user inserted the disk (upside-down) into the drive, the new write-unprotect notch allowed the floppy disk drive to treat the side as writable, thus doubling the user’s storage capacity.
The problem stems from the disk detection used in most floppy disk drives, or at least the ones I have available right now. In addition to looking for the write-unprotect notch, these drives also look for an index hole, a small circle cut near the hub of the disk. Through this hole the drive can sense the disk moving by receiving an index pulse (read by a photo-resistor on the drive). If one inserts the second side of a flippy disk into such a drive, the index is no longer detectable, and drive will not perform read or write operations on the disk.
Pictured above is a hack for emulating the index pulse sans index hole. It’s described in the Catweasel MK4 manual [pdf] (see “reading Flip Disks” section), one of the floppy disk controllers, along with the FC5025, in use here. Many of the technicalities of this hack I’m encountering for the first time. The basic mechanics however are as follows: the wires leading to the photo-resistor are cut and rerouted to a reed switch. The iron rods in the reed switch open and close as a magnet, superglued to the spinning platter (pictured above), nears and recedes it.
The effect is that the drive is still getting a pulse (albeit not an actual index pulse) as it normally would on the regular side of the disk.
Despite all this work by Doug Reside, and on two different drives here as well, I have not been able to read the flip side of the disks provided. As the manual notes, positioning of the reed switch and magnet is a sort of nebulous thing, subject to trial and error. My experience with an ohmmeter and electronics is minimal, so I’m holding off analysis until I can learn more. In addition, some of the settings on the Catweasel are also subject to diagnosis and consideration. On the plus side, the FC5025 reads the regular side of the disks fine.
More generally though, the problem illustrates a stumbling block for digital media preservation. Here the drive is designed to accurately and reliably read and write data to a disk is checking for sector 1 to begin its operations. However, an impartial read of the disk bit-by-bit is enough. Interpretation of the filesystem and files could be left to another program. It would be ideal to read the disk with such fidelity to the fluxes and the timing between fluxes that the resulting data is as close to analog as possible. Basically, reading the information and interpreting the data, for the archivist, should not be dependent on one another, and would not be located in a single device. It seems though that such is often the case though when one encounters something like filesystem mounting troubles, for example. Going in this direction though, I’m not sure how much any device requires some interpretation to read a disk’s information. Perhaps some interpretation is necessary for transfer.
In any case, it’s been an interesting week. Research on 5.25″ floppies really illuminates their physical variety: hard- and soft- sectored, single and double sided, with and without hub rings, one or two index holes, along with numerous capacities. This is aside from varieties in software formatting.
All in all, a challenging start. I’m looking forward to the rest.