[Coco] Re: Disk controller summary?

[Mark Marlette]

As I recall, the high density clock for 3.5" disks is 500KHZ.

Go to www.atmel.com and look at the AVRs. Powerful little units. I did
at 4MHZ model here at work to read magnetic stripes for door control.
AVR was $2.38. Most instructions execute in one clock cycle, 32
registers, ram, flash. Dev kit from Digikey is $80 and comes with a few
AVRs. Hard to beat.....




Quoting KnudsenMJ at aol.com:


>

> In a message dated 3/9/05 6:47:39 PM Eastern Standard Time,

> jdaggett at gate.net writes:

>

> >Only  if you add an external PLL to demodulate the FM and MFM

> >signal off the  disk. I am not sure a digital PLL will work  or one

> in

> >software.  Though I maybe wrong on this point.

>

>

>

> I was once part of a reverse-engineering project where we had a

> competitor's

> 8" floppy disk, with no idea of its format or underlying OS.   We

> built an

> interface to just sample the flux transitions under the read head,

> which came

> thru as a string of 1's and 0's.  A friend built the hardware  that

> sampled

> these data and saved them as a file.

>

> I wrote a simple software state machine to track the flux transitions

> and

> decode into actual 0s and 1s in terms of the MFM encoding.  Didn't

> take  long to

> combine these into bytes, and figure out it was CP/M.

>

> The moral is: you don't need a PLL, just a system that can run 2 or 3

> times

> faster than the disk's clock rate, so you can oversample the flux

> changes.

> Your software must be smart enough to realize that some pairs of 1s

> or 0s mean

> that your ahead of the game, so you throw out the second 0 or 1 and

> re-synchronize your state machine.  That's the phase-locked part of

> it.

>

> Do any of these PICs, AVRs, or HC11s clock fast enough to sample the

> flux

> transitions?  ISTR they were in the 100s of KHz range, so probably

> can.

>

> BTW, never heard of the AVR before, but it sounds like a cool family

> of

> devices, maybe good for building MIDI routers?  --Mike K.

>

>

> --

> Coco mailing list

> Coco at maltedmedia.com

> http://five.pairlist.net/mailman/listinfo/coco

>