The Great Capacity Swindle

[platon42]

bloodline wrote:

zipper wrote:

I remember formatting the amiga disks in a special matter it would give capacity of 1MB...
I don't remember what i used to format the disks in that matter though....

diskspare.device, upto 984 kB, is one contender.

If you rewrote the trackdisk.device to use 8B/10B encoding instead of MFM... then you could proably incraese a standard floppy capacity to 1.4Megs...

I once (in 1998) wrote a "tbc.device" which used "three bit encoding" for two data bits instead of MFM (two encoded bits for one data bits) and theoretically allowed 16 blocks per track. Using 82 cylinders, you could reach 1312 KB per DD disk. Unfortunately, I only managed to get the encoding stable for the first 40 cylinders -- probably due to precompensation after this mark, the reading of the data was not stable. I gave up the project as floppies were becoming more and more obsolete at that time anyway.

[platon42]

bloodline wrote:

 Unfortunately, I only managed to get the encoding stable for the first 40 cylinders -- probably due to precompensation after this mark, the reading of the data was not stable. I gave up the project as floppies were becoming more and more obsolete at that time anyway.

Would precompensation really affect this? I would suggest that perhaps it was more likely that you might have made a calculation error that only showed up when the numbers got big enough. Not wishing to cast aspersions upon your code or abilities :-)

Well, the errors occurred exactly on the edge to cylinder 40 and *all* tracks after that were bad. AFAIK the floppy controller uses two or four different "zones" with different precompensation, switching at exactly half through the disk. I'm pretty sure it was no calculation error (and 40/80 is not even a power of 2, nor a big number).

[platon42]

Zac67 wrote:
I understand you used no write precompensation at all? Then that's probably the cause. Local velocity of the medium increases constantly inward - on track 40 you have a ~35% higher speed, rising to nearly double speed on track 80. I'd guess even three different precomp zones would be necessary. Nice project though. ;-)

I used the trackdisk.device RAW (MFM) writing routines to avoid having to hack paula registers directly, so they should cater for precomp automatically... but who knows why it failed. It was a nice experience getting that low level