Pwm data storage

A project log for Esot's Trenches

True tales (Trigger Warning!)

esot.eric 10/19/2017 at 07:460 Comments

the last log goes into it a bit... never heard of it before, came up with the idea as a result of a project idea I've had for years.... storing audio on floppy disks.

The original idea was to replace the floppy drive's magnetic circuitry with that from a cassette player... thus recording analog onto the floppy.

That never came around... and since I've been learning a lot about floppy drives and disks as a result of #Floppy "Fun" -- Backing up a unique Kay-Pro disk ... it came to me to record the audio as PWM, making use of the floppy drive's normal circuitry.

Then a realization (in the last log) that using PWM could be a great method for high-resolution and high-density data storage on mediums like magnetic...

Briefly, e.g. a floppy disk records at 500k potential magnetic polarity-transitions per second. As I understand, that limitation is largely due to the fact that any closer and the magnetIsm stored on the media will repel each other.

So that got me thinking... e.g. say you want to store a byte of data... that's 8 bits... on floppies, each data bit is stored as two magnetic transitions... thus data is stored at 250kbps, so 8 data bits requires 8/250,000 seconds... or 31.25kbytes/sec...

But, what if we didn't store data bits, and instead stored pwm... 8bits of pwm data could be stored in two magnetic transitions... those transitions can't ever be any closer than 1/500,000 second (tn)... but... say you allocate the first tn to always being high... the last tn to always being low... now the middle can contain the high to low transition anywhere inbetween (not limited to multiples of tn seconds...).

So, say three tns are used (minimum) for each pwm-value... with the right circuitry we've shoved 8 bits of data in 3/16ths of the space required to store each bit individually.

Further, there's nothing (analyzed yet) that would limit that to 8 bits... why not 16? 32?

It's the position of the transition that determines its value.

... anyways...

There's plenty to be pondered... disks don't spin perfectly smoothly at perfect rpms... (thus usually using two magnetic transitions per bit, one driving a clock)...

The circuitry would have to be highly precise to determine between each of 256 edge-timings in one tn...

Etc. Etc. Though, plausibly acheivable... or modifiable... (4bits rather than 8?)

... anyhow... did some searching tonight and came across numerous articles mentioning that PWM is used to store data on magneto-optical drives... so guess I'm not too crazy in this line of thought. But... I've yet to see anything describing the implementation, only mention of it.