8 Apr
2007
8 Apr
'07
11:56 p.m.
On 8 Apr 2007 at 12:21, Dave McGuire wrote:
> > The largest MFM drive ever made was
_probably_ the Maxtor 2190
at
> 190MB unformatted. ISTR that there was a logical
limit to the
> number of cylinders due to the control protocol (or perhaps
> limitations in the PC BIOSes of the time) and physical limits to
> the number of sectors/cylinder and the number of platters (the 2190
> had 15 or 17?)
PC BIOSes generally supported only 1024 cylinders, but some BIOSes
extended the cylinder number by using a couple of high order bits in
the DH (head) register. So you could use your Maxtor XT2190 with
1224 cylinders if you knew how.
16 cylinders was the only hard controller limit; but even the XT2190
used only 15 heads (as did the Atasi/Priam 519). The controller
sector register is 8 bits wide, giving one a possiblity of 256
sectors per track and the cylinder was specified by a 16 bit register
pair, so the limit there would be 65536.
The issue that doomed MFM was the slow data transfer rate (5MHz).
The only reasonable way to put more sectors on a track (other than
using RLL) would have been to slow the spindle speed. MFM/ST412
just got too old.
Several ST412 interface drives had enough slop in the interface to
insert an extra 256 byte sector on cylinder 0 to record the drive
geometry. This was usually called sector 0, but any sector number
could have been used outside of the normal BIOS range.
Cheers,
Chuck