Rotating memory data recovery

[Tom Jennings]

On Mon, 2004-09-13 at 08:44, Paul Koning wrote:

>  John> What can the actual drum clock be...  I dunno...  another
>  John> reason to slow the drum; see below further...
> 
> No, don't do that.  The heads might not fly if you do that, and you'd
> end up with an instant head crash on every head.

As I wrote earlier, they are contact heads, but you are correct, running
them at reduced speed would prevent them from lifting ever, rather than
just the speed up/down time duration.

(The heads are pressed against the platter with a small spring, with a
few grams of force, to counter-act the lift. There's an adjustment
procedure in the hardw. man. for setting it.)


> Digitizing at 8 bits, more if you can get it easily, and minimally 8x
> the data clock rate sounds like the way to go.  That's clearly more
> than the original used.  The argument for doing so is that it will
> give you a way to recover data that would not be recoverable with the
> original electronics.

Agreed. This would be 3-channels of 8-bit at 640Ksamples/sec to get the
timing tracks, or 35-channels for timing+data.

The thing is, after RTFM and this discussion I'm leaning towards
disabling writes, machine bring-up, then record timing tracks as digital
data:

  a) I have to spin the platters under the heads (note1)
  b) I can reasonably assure writing will not ruin the data.
  c) There's a good chance the platter is OK; if not this path
     should cause no more harm than any other approach.

(note1: There is a way to recover the data without spinning it, but it's
a little out-there: remove the platter, clean, apply that magnetic fluid
(sic) used to make bits visible, and *optically* make a copy of the
disk. It's not *that* crazy, but it's extremely unlikely I'll attempt
this. The disk is only about 80 bits/inch, it's not too crazy.)

If the timing tracks turn out to be bad, or read amps are off etc, I can
get more exhaustive.