MFM vs RLL or ST-225 vs ST-238R (was Re: 3.5" floppy drive in IBM XT?)

--- Joe <rigdonj(a)cfl.rr.com> wrote: Does that mean that you confirmed that your controller is an RLL controller? There's no substantial difference between an ST-238R and an ST-225 except a) the label, and b) some manufacturing confidence that a tighter encoding scheme will not drop bits. The data cable holds analog signals. This is significantly different than an IDE drive (whose cable is all digital). The density of storage is entirely dependent on the nature of the controller. The quality of retention is dependent on the nature of the drive. People did attempt to stick an ST-225 on an RLL controller. Results were mixed. It was a "cheap" way to increase storage 50%, but if your drive had some weak spots, you would develop lots of data errors. Others chose to put ST-238Rs on MFM controllers. Results there were much better. You only got 21.4MB of storage (marketing MB, AFAIK), but you weren't taxing the surface as much as Seagate said you could. So, yes, your sources are correct - the ST-238R *does* do 32MB, but only when attached to an RLL controller (which you might very well have). -ethan P.S. - there were other arrangements than 614 x 4 x 17 when ST-506/ST-412 drives were used in non-PeeCee applications. One that immediately comes to mind is the Tandon drive (TM602S or TM603S) used in the Commodore D9060 and D9090 IEEE-488 hard drives. Due to the nature of Commodore "DOS" and its floppy origins, they chose not to change the block size of 256 bytes per block. Its internal OMTI controller did tracks of 32 sectors of 256 bytes for a total of 8192 bytes per track, not 8704 (17 x 512). Yes, to keep it on topic, the ST-225 will work in a D9060, but will only use the first 154 tracks). (see http://www.softwolves.pp.se/misc/arkiv/cbm-hackers/2/2264.html , among other things you can find googling for D9090)