ANALOG based encoding using a DSP Floppy controller

Hi Folks I am new to this fourm.

After Seeing this enjoyable video on youtube covering audio recording with a 3.5" floppy drive, I have to ask the following questions. http://www.youtube.com/watch?v=Xpr7B-7BFP4

So if magnetic floppy media can support sound and music, in addition to high speed binary data. Why not use a more modern multi-level encoding and see how much can be crammed onto an old floppy disk?

Having never worked with DSP code before what would it take to build up a 4 levels or more modem and read/write head drivers?

N2TOH said:

So if magnetic floppy media can support sound and music, in addition to high speed binary data. Why not use a more modern multi-level encoding and see how much can be crammed onto an old floppy disk?

Click to expand...

Ever see a Telefunken model 600 recorder? Or a Brush Mail-a-Voice or the Timex magnetic disc recorder? (And you thought the floppy disk was a modern invention!)

The big problem is the limited recording time if you want to use floppy disks.

There's also the fact that most floppy drives (excepting Mitsumi's QuickDisk format) use discrete, concentric tracks, and not a single, record-style spiral. You can see how that effects things when Jeri starts playing with moving the stepper motor back and forth - you don't get forward and backward play, you get normal forward and forward playback stuttering back over successive snippets. So your stepper motor and timing sensor are going to have to be pretty precise if you want truly skip-less playback over a whole side of the disk.

Ah yes the Angular Velocity VS Linear Velocity.
I deem it possible if we still had FDD Drives available to us in large quantities.
Some of the more modern drives are harder to hack with.

The LS-240 had a spiral mode for 3.5" HD disks; that might make a very long running recording time.

I should clarify the application a bit, The intention is to use analog levels to store data on the floppy not audio for the human ear.

The idea is based on the concept that magnetic media can store more then two diferenx fluxes. so using 4 or more levels should not be that tough.

one of the folks in an MITx IRC chatroom suggested 256 levels, so a single byte could be stored in the space of a single bit.

one of the aspects I'm grappeling with is how to implement the hardware.

Well, I'm not terribly sanguine about that. The coating on a floppy disk by design has a pretty "stiff" magnetization curve and the heads are similarly designed--in other words, the thing's designed for saturation-level recording. Consider also, that floppy media is a pretty much hanging-by-its-fingernails level of technology--i.e., there's not a lot of leeway there. It's basically a "trying to do the most with the least" sort of thing. I would think that with more than about 4 levels, the S/N ratio would preclude doing much more.

As far as imiplementation, a good fast DSP might get you somewhere, but be aware that you'll have to completely redesign the read/write amplifiers.

But, hey, have a go!

What about depth modulation? by exploiting the properties of the cross talk that occures when a write head is over driven. I could see and easy way to seperate the tracks from each other by adjusting the data rate, so when the read head is picking up 3 seperate tracks at once it can decern one from the other.

N2TOH said:

What about depth modulation? by exploiting the properties of the cross talk that occures when a write head is over driven. I could see and easy way to seperate the tracks from each other by adjusting the data rate, so when the read head is picking up 3 seperate tracks at once it can decern one from the other.

Click to expand...

Give it a try if you want, but note that most 3.5" floppy coatings are so thin a bright light can be seen very clearly right through the disk. I'm not trying to discourage you--just point out some issues.