Hi! They used to be quite common back in the day.
I think circa 1996 to 2001 or so.

The idea was to allow reading your digital camera's pictures on a PC.
Back then, USB didn't really exist and people still used SCSI and parallel ports to interface with media.
It was that time frame when Compact Flash readers had a parallel cable, still.
And when PCMCIA/CF adapters were still in use.

In that time frame, these floppy simulators made sense.
They accepted Sony Memory Stick and Smart Media Card, the latter was like an oversized MMC/SD card.
With a fatal flaw, I vaguely remember: It could die from formatting errors.

Anyway, that floppy works similar to those cassette adapters for the car stereo.
You know, those old style models with the audio cable coming out on one side:

Inside, there is a magnetic write head that works similar to a loudspeaker.
It just doesn't make sound, but magnetizes the surface of a magnetic material.

In case of the cassette, it writes directly to the read head in the car stereo.
Those MP3 cassettes work pretty much the same way.

The diskette is a bit more sophisticated, of course.

It has both a write/read head inside and uses a metallic disk as a medium.

The special software for PC (Win95 up, afaik) allows using the diskette adapter to the fullest (several Megabytes).
It will stream the data in a non-standard way that the special software then re-assembles on the fly.

It's perhaps possible that it can look like a 1,44 MB floppy if the special software isn't used, not sure. 🤷‍♂️

afaik it cant detect head position so no way of working without hacks

I'm now having thoughts how such a thing can be made workable, though it will need a power supply and decent amount of electronics. There are ways to detect head being over a track area, inductance and capacitance are both usable for that though I am unsure how big the timing margins are and what is the extent of the signal possible to extract that way. 80x sense elements, one for each track, to measure capacitance change to locate the head itself and one wire that goes over all of the sensors that carries the data input/output so floppy head is able to read/write regardless of its position. This will result in least amount of electronics in the "floppy" but still a nontrivial amount. In theory the sense elements could possibly be used for R/W too but then more electornics is needed and space is quite scarce. It will be an expensive thing regardless, goteks make more sense in the end 🤣.

80 tiny densely packed sense elements sounds scary until you realize you can route that on a pcb/flex, but how would you sense floppy head exactly? Its a piece of inert teflon with tiny wire element.

There's still a ferrite core with a minuscule gap and some amount of wire in the head (at least on all floppy heads I have seen so far), it will produce some level of disturbance to the nearby electric fields, even if through that coil alone. I'm sure that gap will do some magic both ways 🤣.
With high enough freq (so that even minuscule capacitances start to matter) for the the sense elements and a sensitive enough amplifier it should be possible to measure the proximity of the head and track its movement, assuming whatever comes out the head (or out the simulator) doesn't disturb the process too much. It shouldn't be impossible but I don't really know the practical limits of it all and I don't think I am gonna try to find out either, I have other things in the todo list that have to come first 🤣

EDIT: One doesn't even really need 80 tracks, even far fewer numbers are doable, since one already knows step size and can extrapolate wanted position from just general idea where the head is. This will simplify the hardware a great deal.

rasz_pl wrote on 2022-12-15, 17:33:

80 tiny densely packed sense elements sounds scary until you realize you can route that on a pcb/flex, but how would you sense floppy head exactly? Its a piece of inert teflon with tiny wire element.

Certainly one of my dream retro projects. And the variability in floppy drives makes a seemingly impossible problem pretty much insurmountable. The only solution might be to emit signals for all 80 tracks continuously.