As the subject implies, I've got a PCjr with a
floppy drive that is having
trouble reading disks and fails the internal CTRL-ALT-INS diagnostics.
Symptoms include a very loud/bad noise at first seek (rest of seeks sound ok),
A failed track-0 detector can cause a drive to make loud noises at first
seek, and to misalign on subsequent track accesses - if the drive hits the
physical stop before the track-0 sensor trips, then the head will "rattle"
against the stop until the controller gives up trying (usually around 77-80
steps). If the physical setup of the drive is such that the stepper is able
to "make a step" past track-0 in this case, the alignment for subsequent
accesses will be off as well.
Another related question: When I was first getting
started with personal
computers 25 years ago, I seem to recall that track alignment was a common
problem and could be fixed by using a calibration diskette and special software
that you could monitor as you turned the alignment screw. Without one of those
factory calibration diskettes, is it even possible to align/calibrate a floppy
drive for track alignment?
You can't do a proper job of drive alignment without an alignment disk and a
scope.
As with most things, there are fine to not-so-fine lines between "proper" and
"passable". I've built a Align/Test function into ImageDisk which will let
you
do a reasonable job of track-to-track alignment of the drive to a known good
disk (alignment is at best only as good as the disk you are using).
With a known good disk in the drive (I would suggest using a factory original
diskette from a high-profile software vendor, or a disk freshly formatted on
a NEW high-quality drive):
- Run ImageDisk, select the drive etc. and execute the A)lign/Test function.
- Step out the middle of the disk. 'S2' will step to track 20.
- Execute 'A'nalyze. This will identify the format of the track, which will
be displayed at the top of the screen.
- ImageDisk will continue reading the track, and will beep at about a 2hz rate,
with a tone which is 500hz + 100hz for every unique sector which it reads
that matches the cylinder id of the track you have stepped to. It will also
show counts of matching and non-matching cylinder sectors.
- If your disk is far out of alignment, you may need to start at the inside
or outside edge to get matching sectors. In really bad cases you may need
to fiddle with the drive alignment in order to get to this point.
- Once you are reading sectors, listing to the audible tones while slowly
adjusting the position of the head will allow you to determine the limits
of where the head can read the track without taking your eyes off of it.
Set the position to the 1/2 way point between these limits.
- Repeat on inner and outer tracks, working toward the best compromise that
you can achieve.
- To adjust the track-0 switch, use 'Z'ero (recalibrate) and '+' (step
out)
to move the head back and forth between track-0 and track-1 at each track
ImageDisk will beep as described above, and display a running count of the
matching and non-matching sectors read from the tracks. When the sensor is
set correctly, the counts will be correct (all matching, no non-matching)
and the head will perform a single step in each direction, without banging
against the stop.
- If the head "rattles" against the stop on 'Z', move the track-0
sensor
slightly toward the center of the disk.
- If the head does not step out to track-0 on 'Z', then move the track-0
sensor slightly away from the center of the disk.
The adjustments are interrelated, in that the drive has to find track-0
before it can reliably seek to any track - some going back and forth is
to be expected (you get MUCH better at it after having done it a few
times).
Nowhere near as good as using a proper alignment disk, however I have used
variations of this technique a number of times on drives that have worked
reliably for many years afterward. As always, your mileage may vary.
Regards,
Dave
--
dave04a (at) Dave Dunfield
dunfield (dot) Firmware development services & tools:
www.dunfield.com
com Collector of vintage computing equipment:
http://www.parse.com/~ddunfield/museum/index.html