[NTLK] Storage card specifications

Robert Stewart robertdylanstewart at gmail.com
Fri Apr 8 14:58:24 EDT 2011


Magneto-optical media has an incredible shelf-life due to the way it works. Essentially, the drive uses a laser to heat tracks on the disk to their Curie point, thereby wiping any magnetic alignment. While the material is cooling, it then uses an electromagnet to hold the cooling bits in the desired state. When it cools, the disk is effectively composed of permanent magnets. That's why the disks don't have a problem around magnets that would eat floppies. They're read using the Kerr effect with the same laser running at a lower power.

One neat trick is *how* data is recorded. The electromagnet flips to the 0 polarity and the drive uses the laser to set the entire track to 0. It then flips the electromagnet to 1 polarity and goes over the track with the laser, but varying the laser's power so it only heats the bits that should be set to 1. It's a really clever trick.

I would not be surprised to hear about magneto-optical media lasting over a century.

--

Robert D. Stewart - AC5ZH
Sent from my phone

On Apr 8, 2011, at 13:24, Dan <dan at dbdigitalweb.com> wrote:

> Interesting, sounds like Flash memory's 'dirty little secret'.  Recently
> on this list there was a discussion how a Newton "never forgets".  I
> have a newt with data over 10 years old and everything is still there
> and the data has not been touched for that long (or approximately).  Of
> course everything is backed up as well, but still very interesting.  I
> will definitely research this further.
> 
> We all know about magnetic based drives, the data can eventually fade.
> Flash can as well.  Sounds like the longer term solution is CD/DVD,
> however over the years there has been discussion on how long they truly
> will last.  And the earlier CD's most likely won't due to the early
> techniques.  Most of the better quality CD/DVD's now say 100 years life
> (at least).  But then again how to prove that and a company just could
> be making those promises thinking they won't be around in 100 years.
> 
> Another consideration, the data storage formats change over the years.
> Records, tape, CDs, Flash etc and we all know that record players are
> not as easy to come by these days.  It is likely that the data will be
> moved over to a new medium long before the medium itself fails.
> Especially since it becomes easier to move great quantities of data
> quickly as technologies improve.
> 
> -Dan
> 
> On 4/8/2011 3:16 AM, Bradley Loeding wrote:
>> Unfortunately, Flash will eventually lose its data just sitting in a drawer.
>> Sad but true. The information is stored, not physically, but as trapped
>> electrons whose presence or lack thereof on each cell's floating gate
>> determines the logical state of the bit.
>> 
>> Just as you can tunnel electrons onto the gate quickly at elevated
>> potentials, so too do they slowly leak away. This is a quantum effect and is
>> temperature dependent, among other things. You'll get a wide opinion as to
>> how long it will take before enough of the buggers shuffle off and drop each
>> cell below sense threshold.
>> 
>> www.cl.cam.ac.uk/~sps32/*Data*Rem_CHES2005.pdf
>> (2nd page)
>> 
>> One major flash manufacture specs its passive data retention at only a
>> decade. Some chip fabs claim 50-100 years, which is a bit hard to believe.
>> Truth is, no one really knows, but I'd personally get a bit nervous after
>> 15-20 years.
>> 
>> Ironically, we can store the Library of Congress many times over on a single
>> drive, yet the next generation won't be able to read it without active
>> maintenance of the data... If anyone wants true long-term storage, break out
>> the pen and ink.
>> 
>> -Bradley
>> 
> 
> ==================================================================== 
> The NewtonTalk Mailing List - http://newtontalk.net/
> The Official Newton FAQ     - http://splorp.com/newton/faq/
> The Newton Glossary         - http://splorp.com/newton/glossary/
> WikiWikiNewt                - http://tools.unna.org/wikiwikinewt/
> ====================================================================



More information about the NewtonTalk mailing list