idea for a universal disk interface

[Tom Gardner]

Agree that we are talking about two vastly different projects.

Personally I think the universal disk reader is doable and interesting but
expensive.

 

Start with a clean bench having an air bearing variable speed motor and a
universal mount to which various pack/cartridge adapters can be mounted.

Add a laser controlled horizontal positioner with a Z height adjustable
probe head station  (manual Z height at first but maybe automated if volume
dictates)

Add some sort of head load/unload

Pretty straight forward stuff

 

A modern TMR should easily read the much wider track of old iron oxide but
the issue is flying high and head crashes.  So there needs to be some
research here.  Perhaps a surface (or track) mechanical buffering process
would be sufficient or a ridiculously wide by current standards TMR head or
such a head with an adjustable flying height read element or some
combination should be workable.

 

Track following may or may not be a requirement.  If the read element is
small enough maybe just positioning it in the center is enough.  Or maybe
just once around synthetic run out compensation would work given the very
wide track/ very narrow reader.  If you have to go full track following
that's not much of a problem with the few embedded servo old iron but
implies a second head  reading the servo disk coupled to the reader head -
doable but one more head to crash and a whole research project on the
various track following systems and associated hardware.   Probably doable
with DSPs since the transition rate on the servo info is pretty low

 

Finally you get to interpreting the raw data; this might be relatively
straight forward since the data rate of old iron is fairly low so the analog
signal could be highly oversampled, 10x maybe, which makes the decoding
pretty straight forward if you know the format and recording code.

 

I once built a disk pack servo writer, class 100 clean room, slit the
concrete foundation, excavated and filled with sand, put an air bearing dc
motor with a 3330 spindle mount in a granite slab and used a laser
positioning micro stepping actuator to write the servo surface of 3336 class
disk packs.  Feels like a similar project

 

Tom 

 

 

 

 

-----Original Message-----
From: Paul Koning [mailto:paulkoning at comcast.net] 
Sent: Tuesday, April 19, 2022 5:30 AM
To: Mike Katz; cctalk at classiccmp.org
Subject: Re: idea for a universal disk interface

 

 

 

> On Apr 18, 2022, at 6:44 PM, Mike Katz via cctalk <
<mailto:cctalk at classiccmp.org> cctalk at classiccmp.org> wrote:

> 

> Which is more generic.

> 

> ESDI, SMD or SCSI.

> 

> In my opinion, SCSI is as close are you are going to get to a universal
interface.

> 

> As for reading raw data from a drive.  The newer the drive, the higher the
bit density and lower the strength of the magnetic fields and hence the
lower the flying height.  You have to deal with linear (or horizontal)
recording, perpendicular recording, Heat assisted magnetic recording,
microwave assisted magnetic recording.  The latest technologies are
approaching 1TB (yes that's TB) per square inch.

> 

> If you spin the platters too slowly you will not be able to distinguish
individual magnetic fluctuations from noise.  What do you propose as your
maximum data density (in BPI) and what is the minimum speed you will need to
accurately decode it.

 

I know about some of the modern drive magic, but I wasn't talking about
those at all.  My comments about recovering raw signals from disk surfaces
is for much earlier disks, especially removable packs.  In more recent disks
you always have the drive if you have the disk since they are the same
thing.  (I'm ignoring "data recovery" services here that deal with
mechanically failed drives; that's a specialized business and as you said
it's increasingly difficult with modern drives.)  

 

If you consider 1960s through 1980s you're likely to run into disk packs for
which the drives may be hard to find.  The mechanical tolerances of those
devices require care but are not crazy difficult, as my RC11/RS64 example
last week was meant to illustrate.  The bit densities are not all that high,
nor the track densities.  Consider for example that the track positions on a
1311 drive are set by mechanical detents, and are low enough that no servo
system is used at all.

 

My mechanical skills and tools are perhaps a bit better than some of the
readers here, undoubtedly worse than others.  I could sketch a "spin table"
that could handle, say, an RK05 pack.  Without a milling machine I can't
build it, but that could be fixed, and I could refine my skills to make it
work.  Do I plan to?  No, but if an interesting enough pack showed up I
could imagine doing it.  The RA60 pack I have would be a bit more of a
stretch -- more platters and higher densities, not to mention lack of
documentation -- but it's still on the edge of possibilities.

 

So I think there are two different possible projects here.  One involves a
generic controller/emulator for common disk to controller interfaces, like
ESDI or SMD.  (Or SCSI but those are off the shelf items, right?)  I'd
imagine there would be plenty of takers for that, just as there are for the
MFM and floppy emulators.  The other, more difficult and less needed, is the
pack reader that I was discussing.  There is a little overlap between the
two but not much.

 

                paul