On 01/26/2014 02:31 PM, Fred Cisin wrote:
On Sun, 26 Jan 2014, Chuck Guzis wrote:
BUT, it would be relatively straight-forward to create
a disk where entire
TRACKS are "micro-stepped" out of position - instead of "are you reading
more of the even-numbered V odd-numbered sectors?", it would be "are you
reading more of the even-numbered V odd-numbered TRACKS".
For data recovery, I had a couple of drives that I could move out of
alignment, although admittedly, I had no calibration, nor software control
of the movement, both of which would be essential for this task.
In the bad old days, we made our own 100 tpi alignment diskettes. We
took a Micropolis drive, removed the stator from the stepper motor, so
that it was really nothing more than an empty shell with bearings and a
shaft. We then coupled a precision 100:1 reduction drive to the shaft
and another stepper and mounted the whole shebang onto a 3/4" slab of
aluminum. It took forever to step across the disk, but the result was
good enough for consistent calibration. A laser-interferometer setup
for reading position exactly would probably have been overkill.
Note that Micropolis drives were known for their leadscrew positioner
accuracy, thus the drive choice. They were slow--15 msec. track-to-track
probably because they used 4 steps per cylinder.
Microsoft continued along that line well into the 80s. The last design
I own is an 1115-series drive where the spindle motor, heads, and
positioner are all part of a solid assembly that pivots when the door is
opened. The stationary part is essentially just the hub "cone". To
compete with the taut-band school, it has a MOS Technology MCU to
provide buffered seek.
Note that leadscrew is also the positioner type used for almost all 3.5"
drives. I suspect the accuracy needs of 135 tpi floppies is a bit more
than can be reliably provided by a taut-band-and-capstan setup.
--Chuck