Inspired by @foone's amazing thread on double sided floppies (), a brief story about Amiga floppy drives
Floppy control on the Amiga involves four separate chips, but the interesting one is Paula. Paula was primarily known as the sound chip, but also handled receiving the raw data stream from the floppy drive.
Unlike PC floppy controllers, Paula didn't have any inherent assumptions about the format of the disk - Amigas can read both MFM and GCR encodings (and anything else that you come up with), while PCs are limited to MFM and have hardcoded assumptions about synchronisation marks
(This is why PCs can't easily read Amiga disks, even if they're using MFM. There's an amazing hack where you use two drives connected to the same controller, put an Amiga floppy in one, a PC floppy in the other and then synchronise off the PC floppy but read off the Amiga one)
Anyway, Paula was capable of handling a bitrate of 500kbit/sec, which was all that was necessary for handling a 300RPM double density disk. But Paula read disks an entire track at a time, and when high density floppies arrived that was a problem - each track was now twice as big.
The net result is that the bitstream from an HD floppy is twice that of a DD floppy. Paula couldn't cope, and redesigning it to fix that would have cost money. So Commodore chose a different solution.
The Chinon FB-357 (and FZ-357) are PC HD floppy drives. The FB-357A is a special Amiga-specific version that, when it detects an HD disk, drops its rotation speed to 150RPM and so maintains the same bitrate as a DD disk. Working HD floppies without needing to redesign Paula.
Unfortunately, it turns out that these drives are no longer available (who could have predicted), so there are hacks to modify the more common FB-357 into FB-357A specification (web.archive.org/web/2007011813…). But it's also tough to find appropriate FB-357s!
There's an alternative approach (aminet.net/package/docs/h…) which involves hacking the timing circuit in arbitrary PC drives to run it at half speed. But then the drive is sad that it's running at half speed.
There's a magnet on the drive flywheel that passes a sensor once per revolution, which is how the drive knows how fast it's spinning. So you stick another magnet 180 degrees around the flywheel from the first, and now each revolution generates two signals and the drive is happy.
Anyway the moral of the story is that designing flexible hardware is great but if you aren't willing or able to fund development to keep up with technological progress you're going to end up sad and your users are going to end up gluing magnets to things the end
