Source: Dr.
Henning Leidecker/NASA Goddard Space Flight Center
Date: May 2002
A recent failure analysis of
two GPS receivers revealed the presence of tin whiskers on the metal
shield enclosing the receiver electronics. The assemblies were to be used
in high altitude applications and were expected to demonstrate successful
use of commercial off the shelf (COTS) technologies. During failure
analysis of the parts, the analyst noted an accumulation of metal debris
(including solder melt balls) on the bottom of the GPS enclosure. Further
inspection revealed a bunch of ‘twinkles.’
The unit consisted of four
satellite receiver modules mounted to a PC board, two on one side and two
on the other. Each was (roughly) 8 cm by 4 cm by 1cm, and was mounted
within a metal box, presumably for RF shielding reasons. The RF shields
were implemented by a press-on lid. The metal boxes were tin-plated, and
each lid was covered with tin whiskers! It is possible that the side walls
also contained whiskers, but the inspection focused only on the lids,
which were easier to inspect. One lid was apparently more densely covered
than the others.
It is easy to miss these whiskers, which show up only as a sort of
‘twinkle’ in the right kind of lighting. Another analyst was called
over to review the assembly and was told that there were tin-whiskers on
the tops of these tin-plated lids --- and he could not find them at first
(no discredit to him in any way!). Only when he reoriented the unit, to
get just the right lighting, could he see the ‘twinkles.’ Even then,
one could not "see" these as whiskers, until one put one's eye
to within several inches, and got the lighting right, and knew what one
was looking at.
It would have been easy to
brush the whiskers away without ever seeing them, as one was working on
the units. Even removing the lid could result in the loss of many, or all,
of the whiskers, depending on how the lid was handled as it was
removed.
Inspection of one lid using a
binocular microscope indicated 5 whiskers (lengths ranging between 1 and 3
mm) in a 6 mm by 4 mm area. This gives a density of (5 whiskers)/(0.24
cm**2) = 20 whiskers per square cm, and allows an estimate of the total
number of such whiskers per lid: (8 cm)*(4 cm)*(20 whiskers/cm**2) = 640
whiskers per shield top.
There were many places within
the receiver where connectors (apparently consisting of uncovered metal)
were spaced by 0.5 mm and less. The whiskers could have easily bridged
across those connectors.
During the inspection, the
"press-on" tops of the RF shields were not removed, so it was
not possible to confirm whether there were whiskers on the inside surfaces
of the lids next to the receiver electronics. And furthermore without
removal of the lids it was not possible to determine if the electronics
inside may have been protected from whisker induced shorting by (for
example) a conformal coat.
The investigators were
reasonably certain that the whiskers they observed were not producing the
original high-altitude failures that initiated the failure analysis.
However, the whiskers certainly constituted a threat to un-interrupted,
glitch-free operation of those units!
