Return to the archive index

Re: Implications for wearables

From:
Date: Mon, 13 Sep 1999 22:04:43 -0400

On Mon, 13 Sep 1999, R. Paul McCarty wrote:
> Does this suggest that semiconductors need to be fabricated in such a way
> as to prevent things like EMPs from knocking our your wearable when it
> becomes more then just a gadget you keep with you like a cellular phone?

For the forces involved in EMP, changing the way an IC is fabricated will have
little effect - you're still inducing huge currents in fraction-micron width
wire lines.  You need to protect your silicon at a macroscopic level, namely
line the inside of your cell-phone with a layer of metal (aluminum foil) to
create a faraday cage around the innards.  You also need to keep the antenna
off and the socket similarly EM-sealed.

> What means exist to protect semiconductors from RFs and other forms of
> radiation? I seem to remember semiconductors on satellites being put in
> special cases to protect from outside interference.  Seems like other
> alternatives might be redundant processors which could detect an error
> like in RAID arrays, and redo instructions.  Or how about simply detecting
> the sudden RF signal and putting the CPU in halt?

Even for satellites it's still relatively low-level background radiation being
designed for and protected against, not the spike/surge of an EMP.  The dopants
used differ, and the design methodologies are more rigorous, but the devices are
essentially the same.

Of related interest, some smart-cards make use of an Si shield layer to prevent
EM snooping (I think it's Schlumberger that holds the patents), but again it's
to stop microscopic levels of RF from escaping, not MegaWatts of EMP from
entering.

> I also suspect that the RAM is the culprit here since it uses the smallest
> features sizes and uses charges to store the 1/0 rather then in the cpu
> where you have more steady state currents. Which would suggest your best
> protection may be using CPUs that have higher peak to peak signal voltages
> like 5v instead of 2.2v, and large feature sizes like 0.3 or 0.5. Like a
> 386 or 486.

In a word, no.  CPU voltage is not a good measure of how robust your
memory will be.  Feature size is a factor for DRAM, where the data is stored as
charge in a capacitor array; write voltages in excess of the supply voltages
are generated on-chip, and refresh cycles (a time-dependant factor) replenish
cell levels.  Voltage is a factor for SRAM, where data is stored by active
latches; feature size is less of a factor than material/doping profiles.  And
then there's FeRAM... :-)

Andrew.

-- 

Andrew Plumb
radio:ve3slg
mailto:
http://www.plumb.org/
http://wwp.mirabilis.com/13667980

--
Subcription/unsubscription/info requests: send e-mail with subject of
"subscribe", "unsubscribe", or "info" to 
Wear-Hard Mailing List Archive (searchable): http://wearables.blu.org

+Previous Message in Thread | Next Message in Thread

From Wear-Hard Mailing list Archive (WH)
Maintained by R. Paul McCarty

Archive created with babymail