drives can be shipped ready inked as D1 thus avoiding the assembly technician having to think about
anything but LEGO mechanics. Talk about a sledge bummer to crack a walnut, the Walnut presumably
belong the assumed size of the technician’s brain.
I rang all the people who might know or be willing to find out what I had to change or where
the limitation bodge or option was, i.e. MS-DOS, the AT BIOS, or the floppy controller etc. The
word from the IBM camp (dealers cloners, IBM research staff etc.) was “the AT only supports 2
floppy drives”. Great. Sounds like floppy drive interfacing became a state secret. Watch out ohms
law, if IBM ever discovers you, you could become an unsupported feature.
By this time the lid was well and truly off my machine with bits spread out everywhere and I
was able to observe the behaviour of the drives and sure enough as
Dave said in his article
the heads were double stepping and therefore all their time banging the end stop. In the
ensuring experiments I achieved every permutation one could conjure, except for DDDS 730K. These
included 730K at 9 high density sectors per track (leaving 6 sectors worth unrecorded), 80 track
single-sided 730K, and even double stepping on the 40 track drive. These did not result from
spurious entries in the CONFIG.SYS file. The formate simply did not necessarily obey the manual.
My revised theory became:–
After boot, at system initialisation the disk drives are addressed, and if present, analysed
for number of tracks. This information retained a least initially.
The parameters from CONFIG.SYS are read into the DPBs for user configured drives (D: to L: in
When any disk is logged in (even to be formatted), it is checked for density of recording (if
The drive is also checked by driving line 2 of the floppy bus into the high density state and
if 360 R.P.M. is detected on the SECTOR line the device is deemed HD otherwise it is double
(i.e. normal these days) density.
MS-DOS decides which type of drive it must be and places the appropriate parameters in the DPB.
It follows from 3:, 4: and 5: that with an unrecorded disk in the HD drive it will never see
that drive as a normal density drive.
It follows from 6: that one cannot format a blank disk at normal density on a high density
drive unless one can convince the system that its a normal density drive by some other means.
It follows from 3 that a 730K formatted disk can be written to on an HD drive and that a
previously formatted 730k disk may be reformatted 730K..
In the absence of some user induced disk parameters, a normal density disk will be presumed to
be 40 track.
The conclusion from the foregoing is that to get a 730K disk formatted from scratch one must:–
have the right disk parameters,
not have done any disk work which might overwrite these,
fool the system into thinking the HD drive is normal density.
One way of achieving c) is to intercept line 2 of the floppy bus, and this is effectively what
I’ve done (I actually wired a switch in to one of the link options on the drive)
I now successfully format 730K by rebooting, throwing the switch and then formatting. I can
then put the switch back to normal and continue to operate at 730.
I’m sill working on the IBM/floppy bus problem and I would like to learn how to get deeper
into the operating system to find a more satisfactory solution.