29 Nov
1999
29 Nov
'99
11:48 a.m.
-----Original Message-----
From: Tony Duell <ard(a)p850ug1.demon.co.uk>
To: Discussion re-collecting of classic computers
<classiccmp(a)u.washington.edu>
Date: Monday, November 29, 1999 9:40 PM
Subject: Re: Needed: 1 IBM 8" alignment disk.
I hear you. This particular drive claimed it did.
> It isn't alignment as I can manually move the
head very slightly back and
> forth which does affect the data coming into the controller assembly but
> does not help it find the actual sector... Therefore, hardware or an
anlog
> adjustment
>
> I would like an alignment disk so I can use "their" tuning procedure to
make
> this drive work quickly. It has a number of
adjustments including :
Clock,
window, gap,
etc...
Most alignment disks do not include such tests (I am not saying that the
IBM one you asked for doesn't). They're just raw drive alignement disks
with catseye patterns for aligning the radial position, an index timing
burst, etc.
If you think there's an electronic problem then you
don't really need an
aligment disk. Start with the spindle speed, which I assume is right
since it almost certainly uses a mains-powered induction motor. Still, it
can't hurt to check that the index pulses occur at the right frequency.
That's all working fine... It's controller has a serious status system in it
where it monitors disk spindle speed, etc...
They'll be a master clock, probably also used for
writing. Check this
with a 'scope or frequency counter. If it's incorrect, find out why.
One master clock.. 1Mhz, good.. A board dedicated to "data timing".
Then take a formatted disk (format one yourself on your
old CP/M machine
if you have to ;-)). Read a sector continously. Display the off-disk data
on one trace of a 'scope and the read clock on another. Does the read
clock seem to be the right frequency? Is it locked wrt the data? If not,
look at that read PLL (or whatever it uses).
Good idea on the CPM unit. I rarely use/see one so I will format one and
start that way.
Index timing is either very important (if, for example,
the controller
expects to see an address mark as the first thing after an index pulse)
or not at all important at this stage.
Not there yet.
At this point you should at least be recovering data
bits from the disk.
So if it still can't find a sector, find out why not. Can it not find an
address mark at all?. Or can it not find the header you're asking for.
This is likely to be a digital problem. As I don't have schematics for
this unit I can't comment any further.
It can't find a sector *most* of the time.. Once in a while I can get it to
read and display a sector correctly... The rest of the time it is in an
infinite loop.
What bugs me is sometimes I can read a sector with a good CRC and the data
is correct... probably *another* bad solder joint... this box has been
filled with them so far.
Thanks for the CP/M idea.
-tony