Re: FAA To Finally Ditch Floppy Disks & Win-95

On 6/15/25 10:04 AM, Rich wrote:

c186282 <c186282@nnada.net> wrote:

On 6/14/25 7:38 PM, The Natural Philosopher wrote:

On 14/06/2025 17:34, Rich wrote:

Also because (esp.  for EPROMS) the rope memory was much more
radiation hardened than EPROMS of the day were.  Also, for 'long
duration' projects such as that, often each component is designed
and built, then in the end the various parts are pieced together to
produce the "flying candle" you see on launch.  But the "computer"
might have been designed and built five years before that launch,
and minus five years from today might have meant no PROM or EPROMS
were even available at the time.
>

There is another issue I ran into.  On an undersea cable repeater.
IN 1988 or thereabouts the company was overjoyed to finally be able
to use silicon, because up till the early 80s silicon simply hadn't
been around long enough to be deemed suitable for long term use
(>25yrs).
>
Rope memory was tried and tested, Fusible link PROMS had no track
record and EEPROMS were definitely subject to ageing

>
   I'd argue that fuse PROMS didn't NEED a 'track record'
   because they're SO straight-up.

 While true, that's just not how getting things in space work.  They
have to be exposed to the radiation environment they will see in space,
and their success/failure determined from those tests.  I.e., they have
to have a "track record".

   I understand, in theory, but by the mid 60s it was
   possible to emulate cosmic rays and they could have
   bench tested them, artificial aging. NASA especially
   DID have the cash to throw at it. Anyway, ceramic
   base, actual hard elements that either connected or
   were physically zapped. Hard for it to go wrong.

   UV-EPROMS are also pretty stable (and NOT nearly so sensitive as
   some here suggest).

 Stable here on the surface, where we are shielded from almost all of
the space radiation.  But UV-EPROMS work on the same principle as flash
(a charged, insulated, floating gate stores the bits).  UV erases them
because the UV radiation is energetic enouogh to cause the charge to
drain from the floating gates.  Well, guess what, there's lots of
radiation in space that is far more energetic than UV radiation, so
your UV EEPROM will likely lose its stored program (either in short
order, or well before the service lifetime of the satelite it is
installed into).  Both of which make it unsuitable.

   Ummmm ... I'd worry about UV-EPROMS more around Jupiter
   than anywhere near earth - at least within a reasonable
   timeframe. A literal sheet of aluminum foil glued on
   would block some of the deadly particles. Not ALL alas,
   hence the service-lifetime issue.
   EEPROMS are suspect BECAUSE they erase electrically, and
   cosmic rays are almost all charged particles and/or
   produce more charged particles when they hit something.
   In any case, they DON'T use mag core memory anymore on
   space probes and such. Re-writable versions of the
   above-mentioned techs are now 'hard' enough. There is
   also ferroelectric - faster with longer service life -
   which I *think* are even more radiation resistant
   (albeit low-density compared to flash). I used FRAM
   for a few products - config and state info - and they
   worked very well.

   EEPROM was more complicated.  Most Flash is even more suspect.

 The only real difference between the UV model and the EE/flash models is
the EE/flash models don't need UV light to initiate the discharge of
the floating gates.  But the mechanism of storage is the same in all of
them.

   Well, 'similar' ...
   Do look into FRAM however, it's a somewhat different tech.
   If I was hired to do a board for a Mars rover now (I won't)
   I'd seriously want to store the boot-up stuff and state
   info in FRAM. Alas I think the largest ones I've ever
   seen for sale are 512kb