🚀 go-pugleaf

In article <1991Feb10.021307.10532@uhura.neoucom.EDU> wtm@uhura.neoucom.EDU (Bill Mayhew) writes:
>An associate I used to work with told me about an incident at a
>trade show where a vendor had a prototype board with untaped windows
>on the firmware EPROMs.  A trade paper photographer snapped a picture,
>the flash of which trashed the contents of the EPROMs, causing
>much embarassment for the vendor who was left with a braindead product.

	With the "proper" electronic flash and the "right" circumstances,
the above scenario is quite possible.

	EPROM's begin to erase with exposure to ultraviolet energy shorter
than 400 nm in wavelength.  In general, as the energy becomes shorter in
wavelength, it's erasing ability increases.  Most EPROM erasers use a
mercury lamp having a quartz envelope, with the predominant wavelength
being being the mercury line at 253 nm.

	A typical EPROM, such as a 27256, is rated to fully erase with an
integrated energy exposure of 15 watt-seconds/cm^2 to the quartz window at
a wavelength of 253 nm.

	A xenon flash lamp, as found in an photographic flash, has enough
ultraviolet energy below say, 350 nm to effectively erase an EPROM - given
sufficient exposure.  However, to be effective for this purpose, the flash
lamp must be made of quartz, and there must be no outer plastic filter.
Almost all non-professional photographic flash units use a flash lamp made
with borosilicate glass (e.g., Pyrex), which significantly absorbs the
ultraviolet energy below 350 nm.  In addition, such flash units invariably
have protective plastic filters, which further absorb any ultraviolet
radiation.  The combination of borosilicate glass and the plastic filter
result in almost no energy capable of erasing an EPROM.

	Many professional photographic flash units use a flash lamp made
of quartz - a necessity for higher watt-second ratings and for short
recycle times where lamp heat does not have time to dissipate.  A further
consequence of higher energy found in professional photographic flash units
is that most have no outer plastic filter.

	In addition, the actual circuit design driving a xenon-filled flash
lamp affects its spectral distribution.  Xenon lamps with little series
inductance in the discharge circuit will exhibit *considerable* energy
emission at the Xe(II) lines of 260, 247 and 220 nm.  Adding enough series
inductance attenuates these emission lines so that the spectral distribution
below 300 nm approaches that of say, a 7500 deg K black body.  In this
latter case, there is still *significant* ultraviolet energy between 300
and 400 nm.

	The point of the above paragraph is that ultraviolet energy emission
of a flash lamp can be considerably influenced by the particular circuit
design.

	The result is that many professional photographic flash units
*do* have sufficient radiated ultraviolet energy to erase an EPROM -
provided that the flash head is held close enough to the EPROM and/or
that enough flashes occur.

	Since professional photographic flash units have energy ratings
between 50 and 250 watt-seconds, it does not require much imagination 
to see how an integrated energy dose in the watt-second/cm^2 range
could occur in a small number of close-up shots.  To cause failure of
say, a microprocessor-based system does not require complete erasure
of an EPROM - in a well-designed system which performs a ROM checksum
test upon powerup, all it takes is the failure of *one* bit to prevent
operation.

	I have had for a number of years a Norman A200B electronic flash
that I have use for serious photography.  This is a portable flash with
energy settings of 50, 100 and 200 watt-seconds, and a fast recycle time
of 3 flashes/second at 50 watt-seconds.  It uses an unshielded quartz flash
lamp in front of a metal reflector.  This flash produces a significant
amount of UV energy.  This flash unit *will* trash bits in an 8755 (remember
those?), as I once learned through accidental discharge of the flash unit
while diddling with a damaged camera connector at a test bench.  The flash
discharged at 200 watt-seconds about one foot away from an 8085-based board
containing the uncovered 8755 EPROM-I/O chip.  The problem was immediately
evident as the microprocessor system crashed, with hitting the reset button
having no effect.  At first I thought it was electrostatic leakage from the
flash, but then the real cause became obvious.  Out of 2K bytes on the 8755,
only a dozen or so bytes had random bits set to 1 - but those were enough.
I was able to duplicate the failure mode using a freshly programmed 8755.
Since more contemporary EPROMS still use FAMOS technology, I have no reason
to doubt that that the "right" photographic flash unit will still erase
bits if held close enough to the EPROM window.

	I do not dismiss the story of EPROM memory loss from a photographic
flash lamp as "electronic urban legend" as stated by another reader.  Such
memory erasure has a sound theoretical basis that is substantiated by
empirical experience.  However, such memory erasure can occur only under a
certain combination of conditions as outlined above.

Larry Lippman @ Recognition Research Corp.  "Have you hugged your cat today?"
VOICE: 716/688-1231       {boulder, rutgers, watmath}!ub!kitty!larry
FAX:   716/741-9635   [note: ub=acsu.buffalo.edu] uunet!/      \aerion!larry