A more daunting project than using the Kryoflux to image old PC floppies is trying to image the floppies that I used on the Epson QX-10 I had for a couple of years.
There isn’t a lot out there on the QX-10, actually. There’s a page that has some information about the keyboard and components, an entry at the obsolete computer museum with some interesting information, a page in French about one person’s recent experiences with the machines. You can still download the operations manual from Epson. There are some information pages and some disk images.
There’s even an emulator, which was linked to from the “information pages” above. It’s part of a suite of emulators created by Toyisha Takeda, and you can download the lot and run them under VirtualBox running Windows 2000. It needs the ROMS in the same directory.
I got the emulator to start just fine. The keyboard layout is a bit bizarre, so I can’t just type B: to get to the B drive, because the : key is actually where my + key is. I type B+, it hears B:. But it does work. Though it also seems to constantly play an annoying squeal out of the speaker, so I had to turn the volume down.
The reason I’m interested in this primarily is that I ran, for a while, a small and mostly unused BBS on it. It wasn’t really ever advertised widely, but it was a somewhat extensive program, and ran natively on the QX-10. I’d like to see if I can recover it. In order to do that, I need to image the disks, which the Kryoflux should enable me to do. And if it can manage it, it is going to be a minor miracle, since the media in the disks looks terrible.
I imaged stream files with the Kryoflux, but the issue now is how to turn it into an image file that I can use in any sensible way. According to this page, the disk geometry is 40×16 256-byte sectors. When I ran it through the Kryoflux software, using MFM, 40-track distance, and 256-byte sectors, I just got a lot of sector-count mismatches.
When I ran it through with 512-byte sectors, I got quite a lot of green (successfully-read) blocks, but it reports 10 sectors per track, which isn’t right.
A hex editor on the image confirms that I have read some real data from the disks this way.
But I have no idea what I could use to mount this image, and moreover, I don’t really trust it given the “10-sector” problem. Did it find 16 sectors and mix in 1K of garbage into 2-sector chunks? Did it lose 6 sectors? Was I wrong to expect 16 sectors in the first place? (Or is that actually 0×10, in which case it is correct?) I tried mounting it in FreeDOS and Win2K to see if I could use some CP/M disk readers to deal with them, but that didn’t work very well. The QX-10 emulator can read TeleDisk format, and a couple of versions of TeleDisk are available from the QX-10 disk image page, but neither was willing to recognize the mounted image as anything. Nor was Anadisk (linked to here) or the elusive 22disk (which I found a link to here). I even tried using Virtual Floppy Drive to try to fool those programs into thinking that the image I had was a real drive, but they’re all too “smart” and want something in the real (virtual) disk drive.
What I’d like to do is just get the files off those disks and onto a different CP/M machine. One possibility is to run the software one of the Apple ][ machines I now have with Z80 cards in them, that's my current goal. (And in that form, they would also be usable in Virtual ][, since that emulates a Z80 card as well.) But I think I’m still a ways away from being able to do that.