Re: Did Commodore cheat with the quad density floppies?

On Freitag, 4. Januar 2019 19:56:43 CET smf wrote:
> On 04/01/2019 05:14, William Levak wrote:
> > It's not more efficient. It's more accurate. In MFM, if you have a
> > long string of zeros, reading accuracy is limited by the accuracy of
> > your clock. GCR eliminates this problem.
> 
> Neither MFM or GCR can have "long strings of zeros", they were both
> invented to eliminate that particular problem.
> 
> MFM is less efficient because it requires 2 transitions per bit, while
> GCR needs less (depending on the particular encoding type, IIRC
> commodore used 8 to 10).

I  think you're mixing up FM and MFM. FM needs one or two transitions per bit 
(one for 0-bits, two for 1-bits).

MFM needs at most one transition per bit (only for 0-bit that follows a 1-bit, 
no transition is needed)

> However MFM is more accurate because it guarantees there will only be 1
> zero transitions, GCR can have more.

Again, FM needs has a maximum one 0-bit. 

MFM actually varies it between 1, 2 and 3 0-bits in a row. But as it writes 
with double the frequency, it results in 4us, 6us and 8us between transitions, 
which are equivalent to 0, 0.5, and 1 0-bits in a row at the write speed of 
GCR.

See here for a comparison: https://extrapages.de/archives/20190102-Floppy-notes.html

> 
> This allows MFM to be clocked faster and still be reliable.