One of the major challenges is the lack of experience with the 8” drives and floppies. Thus, a trial and error was the approach we used. The knowledge needs to be unearthed step-by-step hoping to find the relevant bits and pieces to get the stuff working in the intended way. In parallel we tried to get in contact with an expert of legacy computer systems who runs a computer history museum of his own and actually owns 8” equipment. We hope that with a wider hardware and software selection coming from the legacy systems a more systematic approach will be possible.
To rule out hardware related issues like cabling, the first suboptimal connection, which was pretty long cable with open end which might lead to signal reflections, was shortened and a second edge connector cable was produced. The new one ruled out any misfitting of the connector as it was an original 50pin.
Challenge: The disk geometry and other parameters
For non-familiar users the options for the disk geometry remain a bit opaque (beside the recording mechanism, but that one is presumed to be MFM and cannot be changed). For example, generally the drive supports 77 cylinders (Should equal to tracks – on a single sided disk tracks are cylinders, but in some cases of double sided documentation the number of tracks is double the cylinder count. In these cases, they say tracks per surface.), but the IBM specification (type 2D – printed on the BASF disks – possible configurations taken from Wikipedia) have a special index cylinder and then 74 more cylinders to offer. While the index cylinder configuration remains the same the remainder can differ (significantly). Index cylinders/track 0 in IBM formatting often are of FM, the data tracks in higher density floppies use MFM instead.
More tools to try: FDIO utility
As many of the floppies were Hewlett Packard #92195A which were 8” disks especially for the HP 9895 (A) drives, we thought we should try fdio which is a set of utilities for the HP 9845 (nb it is a different model).
One of the options in this utility is to analyze unknown media. The analyze command used
fdio -analyze -d pc8hd a: Where pc8hd = 8” high density.
Unfortunately this did not work, response was:
FDC analyze: --- testing sector geometry for side 0 ---
FDC analyze: testing modulation & data rate...OK
FDC analyze: testing sector size, start sector and sectors/track...failed
FDC analyze: could not get valid track info, probably media not formatted
Possible settings for the disks are:
Start sector: 0
Sector size: 256
Start sector: ?
Sector size: 128
Finding out about the Nashua Disk
The label on the Nashua test disk reveals very little. There is a part number (FD-1D-WP) which sent us on a web search and luckily we were able to discover that this is a single sided double density floppy disk from some offer. The specs on the original manufacturer’s box listed on the side of the carton are as follows: FD-1D WP-R (the item number), S.S. / D.D. SOFT which translates to single sided (!), double density and soft sectors. This explained the failures of the previous low level formatting verification experiment: A single sided disk can actually be formatted double sided, but reading from the empty side will fail. Changing the parameters to single sided with setfdmprm got us much further into the verification.
This brought us to the raw floppy command fdrawcmd. In our tests we were reproducing the experiments of fdutils homepage and using information on the interpretation of outputs from this side:
fdrawcmd readid 0 rate=1 need_seek track=0
no disk change
The readings look reasonable as the first byte (0: in output) is below the magic 7 (above this value the reading produced an error). In byte 5 we see the sector under the head when command was issued. In byte 6 the sector size encoded (0 equals 128 Byte, 1 256 Byte etc.) The extraction of the second track (data) produced meaningful output too:
fdrawcmd readid 0 rate=1 need_seek track=2
no disk change
Byte 3 is 2 here, which says: Plain disk with singlespaced tracks. Good thing here: We were actually able to read different tracks than just the first one (track 0). All these runs reliably produced proper output if we used the proper configuration for that particular disk. If we get output like:
no disk change
then the command was not able to make anything out of the disk. This happened when we changed the settings for the Nashua disk. Unfortunately, that result was all the same for all possible relevant settings of the disks in question).
The story so far
Up to now the disks we are keen to read did not reveal anything. Suspicion could be that they actually might contain nothing or bit rot ate the recordings. Time to sum up and structure the problem:
- The cabling seems to be OK as we can run some analysis and low level format (to a certain degree) the only disk we have for experimenting. We cannot high level format (mtools) it, though. It is still not entirely clear how the setfdprm and fdformat actually relate to another. We were not able to set arbitrary values and for some settings a SD setting jumped to a QD in result. Additionally, the actual formatted capacity we expected from fdformat did not occur but some lower value.
- We cannot finally say, if the combination of the rather modern floppy controller and the prehistoric drive actually work together. They do it to a certain degree as the low level formatting shows the same track number output both in the tool output and the seven segment digit on the dbit adapter.
- Unfortunately, we do not have a second disk or setup of another machine to prove either the floppy drive or the floppy itself to be reliable. The drive could be e.g. out of “focus” and thus can write/read a fresh disk, but not the other disks formatted in an other drive.
- Track 0 (or other tracks) of the floppy disks we are interested in, are not accessible with the tool set we have on hand or we we unable to find the proper settings or tools. Or the recording used is totally different from what the drive is able to interpret.