Re: I X-Rayed my Apple IIe

["Michael J. Mahon" <mjmahon@aol.com>]

Grant Stockly wrote:
> > After 100 stages of multiplication, getting 6-10mA would require some
> > pretty good sized capacitors at 60Hz, so I'm guessing that this is a
> > high-frequency switching supply.
>
>
> Some new ones are.  Old units are just huge and heavy.  Lifting eyes on
> the top.  :)
>
> But If you double your double and so on you'll only need 10 stages.
> Right?

Typical voltage multipliers don't work that way--they have a stage
for each multiple of the input voltage.  The diagrams you sent me
are just such multipliers of the Cockroft-Walton type.

It has ten stages in five modules, for a multiplier of 10x (minus
losses).

The transformer that drives it steps up the input 25kHz squarewave
from, say, 80 volts to 8000 volts, then the multiplier takes it
to 80kV.

> > I understand--physics was my subject even then.  (I have also always
> > been a lead collector--I'm guessing I have a couple hundred pounds,
> > most of it from thrown wheel weights I found while riding my
> > bicycle.  ;-)
>
>
> Our last cabinet for 450kv 10ma weighed in at around 8,000lbs.  : )

If you want to shield a large volume from high energy photons, it takes
a lot of mass.

Radar has just advised me that if I take a sharp turn right *here* I
can graze the edge of the group's topic.  ;-)

In addition to soft errors ("single event upsets") caused by alpha
particles (of past DRAM fame), cosmic ray showers also produce enough
ionizing radiation to cause DRAMs or even dense SRAMs and registers
to experience bit-flips.

The atmosphere is our security blanket in most cases on the planet, and
error-correcting codes are the backup plan.  The total sensitive volume
of an Apple II, combined with its rather huge transistors, make it quite
safe most of the time without any correction--or even detection!  But a
large server with a thousand times as much "active volume" and much
smaller (more modern) transistors is a lot more susceptible, and relies
more on error-detecting-and-correcting codes to keep its DRAM stable.

A few years ago, a server in the Denver area began to experience error
rates just above the threshold of what the error correction could handle
without special effort--the rate of uncorrected errors was many times
as high as the same system nearer sea level, because it was above about
17% of the earth's atmosphere.

The shielding mass equivalent to the lowest mile of the earth's
atmosphere would be 6 inches of lead, or a little less than 6 feet
of water!

Needless to say, the software was modified to check each memory
word more frequently to correct any errors before they became
multi-bit errors, since we figured we couldn't convince our Denver
customers to install swimming pools above their computer rooms.  ;-)

-michael

NadaNet networking for Apple II computers!
Home page:  http://members.aol.com/MJMahon/

"The wastebasket is our most important design
tool--and it's seriously underused."