> On Apr 13, 2022, at 9:45 PM, Fred Cisin via cctech <cctech@classiccmp.org> 
> wrote:
> On Wed, 13 Apr 2022, Paul Koning wrote:
>> Indeed.  Though even that is hard for the more exotic formats, if original 
>> controllers are unavailable.  How would you read, for example, an IBM 1620 
>> or CDC 6600 disk pack, given that the machine is hard to find and those that 
>> exists may not have the right controllers?  But both are certainly doable 
>> with a "generic" track extractor engine.  Turning track waveforms into data 
>> then becomes a matter of reverse engineering the encoding and constructing 
>> the software to do that.  This may be hard, or not so hard.  For example, if 
>> you wanted to do this for a CDC 844 disk pack (RP04/06 lookalike but with 
>> 322 12-bit word sectors) you'd get a lot of help from the fact that the 
>> source code of the disk controller firmware, and the manual describing it, 
>> have been preserved.
>> Then as you said the real goal is to recover files, which means also having 
>> to reverse engineer the file system.  That too may be documented adequately 
>> (it is in the 6600 case, for example) or not so much (does such 
>> documentation, or the OS source code, exists for the 1620 disk operating 
>> system?).
> Some projects are well beyond the reach of even the most insane of us.
> I don't think that any of us here today have the ability to build a 
> replacement drive from scratch.  Even with full access to the original 
> construction documents.
> Now, if we had NSA level of facilities, . . .

I don't think a 1970s era disk drive replica is quite as hard as you suggest.  
In my comment I wasn't actually thinking of that, but rather of the possibility 
that you might have a drive and packs, but no computer to connect the drive to.

That said, consider what, say, a 1311 looks like.  It's a spindle and a head 
carriage, each with the levels of precision you would find on a good quality 
lathe.  That suggests the guts of a small CNC lathe, or the building blocks of 
such a thing, could be put to work for this.

One data point: I remember when our RC11 spindle bearings failed (in college, 
1974).  DEC FS was called in, the tech decided to look for a low cost solution 
since the machine was not under contract.  The normal procedure would have been 
to replace the spindle motor, of course.  Instead, he disassembled the drive, 
took the motor to Appleton Motor Service Co., which pulled off the failed 
bearings and pressed on new ones, reinstalled the spindle motor, presto, good 
as new.  He didn't even have to reformat the drive, all the data on it remained 

So the tolerances on drives of that era are not all that severe, not out of 
reach for ordinary skilled machinists.


Reply via email to