archiving data, was RE: Media Longitevity/Care

Tony Duell ard at
Sun Mar 20 12:32:35 CST 2005

> On Sat, 19 Mar 2005 16:28:45 -0500, Scott Stevens <chenmel at> wrote:
> > On Fri, 18 Mar 05 07:39:43 GMT
> > msokolov at ivan.Harhan.ORG (Michael Sokolov) wrote:
> > > As I have stated many times before, I believe the solution is for us
> > > to open our own factories to make them.
> > 
> > I suspect, rather than 'factories', that it would be similar to being a
> > 'gunsmith.'  There are plenty of artisan machinists who make all sorts
> > of complex machinery.
> Hmm... Drivesmith... one problem, of course, is that the tolerances
> for hard drives are somewhat tighter than for strictly mechanical
> devices such as steam engines and guns.  I know people who make both
> (steam and guns), but I would seriously hesitate to make any sort of
> drive bits newer than, say, an RK05.

Agereed. I think that _most_ mechanical parts for classic computers could 
be made ina good home workshop, with the exception of disk/tape heads, 
and probably bits for inside HDAs. That still leaves a lot that could be 

I certainly think that any part for, say, an ASR33 could be made at home. 
It may not least as long as the original, but do you really need to run 
the machine 24/7? :-)

Which, alas, leaves storage devices as the main problem. Apart from 
heads, I think parts for floppy drives could be made at home. Hard drives 
are a lot worse. Even things like the spindle for an RK05 would be a 
serious project (does that one have a ferrofluid seal in it? Some drives 
certainly do). 

> Having said that, someday I know I'll have to face at least one round
> of this... I have several DF32 drives that I expect have seriously
> worn plating (I've only ever tested the electronics, but I _have_ spun
> the platters up and down, 20 years ago, long before I knew it could
> cause a problem).  Getting a replacement platter fashioned and turned
> to suitable smoothness is completely within the realm of accessible
> tools for steam engines, etc.  It's really no worse than turning a
> brake rotor.  My only real concern is _plating_ the platter.  Setting

Sure. I suspect you could skim the surface of the existing platter rather 
than start from new (I can't believe the thickness is very critical). Of 
course you would have to replate it.

> up an electroplating rig is easy enough (I copper plated a number of
> nickels and quarters as a lad ;-) presuming one can a) come up with an
> acceptable formula for the coating, and b) gain access to the various

I thought it was mostly nickel, but I am not sure.

> soluable compounds to mix up a batch of plating solution.  Even when

Indeed. I know the gold plating is not something to try at home... (the 
only reasonable method uses gold cyanide..). If you could find out what 
metal(s) the plating consisted of, you could probably find details of the 
chemicals to use in some of the old books on workshop methods or 
electrical engineering that Lindsay have reprinted. I am pretty sure I've 
got this info somewhere.

> that's done, one still has to construct the timing track generator,
> but at least that's well documented, so one could build the real deal
> from old DEC bits, or one could simulate a formatter with modern
> circuitry... the timing pulses are not that bad.
> Vince and I have also kicked around what it would take to build a
> modern solid-state replacement for just the rotating media (there's
> already a prototype 74LS replacement for the entire 4-drive
> assembly)... it's honestly easier to replicate the entire drive
> assembly (since that can be done with 100% digital circuits), but the
> whole point would be to have a modern replacement _just_ for the
> platter and possibly the heads.

And it would probably be even eaiser to get the whole machine, storage 
and all, into a couple of FPGAs, or as a software emulator... 

I guess I am odd, but I feel that the peripherals are as much a part of 
classic computing as the CPU. Sure it can be fun to design new 
peripherals for the classics, to link CF cards to a PDP8 or whatever. 
It's also good to keep the origianl peripherals running. 


More information about the cctech mailing list