Frank Miles wrote:

Tektronix was, during that time, strongly discouraging all new designs from
using tantalums. IIRC they had been taken to court over a case in which
a 465 'scope (the original, not the plastic follow-ons) had spontaneously
ignited and had resulted in an expensive fire. Forensics revealed that
a tantalum power-supply bypass cap had started the conflagration. The
drive to reduce tantalum usage was driven primarily by this liability issue,
more than component cost. If you wanted to use a tantalum, you had to
justify its usage to the component/design review committees -- which wasn't
difficult if you had good reasons and your design was solid.

-frank
(ex-Tekie)

I was there at the time, too. Tantalums were essentially verboten unless
the source impedance supplying the tantalum cap was at least 3
ohms/volt. That's because it was found that the short circuit failure
mode was aggravated by high inrush current, so the source current had to
be limited. One of the chief reasons we had been using tantalums in the
first place is that they have very good bypass characteristics up to
quite high frequencies -- so a single capacitor could handle a very wide
range. When the source impedance was high, the capacitor didn't need to
be so good in the first place, and of course adding a physical resistor
in series with a supply bypass pretty much defeats the whole purpose.
Consequently, the 3 ohms/volt rule pretty much eliminated tantalums as a
viable choice for most applications. Fortunately, it was at just about
the same time that very big improvements were made in aluminum capacitor
technology. As the aluminums shrunk in size, they became much better at
bypassing higher frequencies. So they took over from tantalums pretty
rapidly. There was a glitch for a while, though -- boards were being
cleaned with Freon at the time, and it was discovered that Freon could
migrate past the seals on some or most aluminum capacitors and corrode
the aluminum, leading to poor reliability. The solution adoped by some
manufacturers was to add a rubber seal at the lead end of the capacitor.
That increased the length of the leads between the outside of the
capacitor and the inner body, increasing the lead inductance and
decreasing the capacitor's high frequency bypass capability. . . but
that's just another example of the day-to-day problems an engineer faces
and has to overcome.

Incidentally, I got a Tek 1502 TDR on eBay not long ago. It had a
shorted tanalum power supply bypass capacitor.

A couple of other anecdotes -- A time base plugin I designed had gotten
through the entire extensive pre-production test phases, accelerated
life tests, etc., and was in pilot production. I walked past the
production line technician's bench every day, and began noticing several
tantalum capacitors of the same type in the replaced-component box. They
had come from a sweep circuit I had essentially copied from an
instrument which had been in production for some time. Puzzled, I
analyzed the circuit carefully, and discovered that at an extreme
setting of one control, the tantalum cap could have a very small reverse
voltage applied. I modified the circuit to eliminate the possibility of
any reverse voltage of any level, and the capacitors quit failing.
Servicing data from the instrument I had copied the circuit from showed
noticeably reduced reliability of the capacitor, also. The lesson
learned is that tantalums won't tolerate _any_ reverse voltage. If they
don't fail immediately, a disproportionate number will fail eventually.

The other anecdote involves a QRP rig. As a crude reverse-voltage
protection, I had reverse-connected a 3-watt diode (actually, a 36 volt
zener I had lots of) across the power supply terminals. My battery
supply normally had an-line fuse which would blow. Just before Field Day
one year, the fuse holder broke and I didn't have a spare in the junk
box. I'd never blown a fuse in 20 years of Field Days, so went without.
The battery was a 12 volt, 5 Ah sealed lead acid unit, capable of a few
hundred amps if shorted. As I'm sure you've guessed, that was to be The
Year of the Reverse Connected Supply. The wires to the battery
immediately melted out of their insulation, burning some holes in the
tent floor. I managed to disconnect the battery without getting burned
and before a real fire started, and checked the damage. The rig's
(recently installed) power switch was fortunately off, so the innards
didn't get any reverse voltage. The diode had gotten so hot that the
plastic case had fractured and probably burned -- it was gone. The
diode's solder joints had melted, and the two separated diode leads were
dangling. But there was still a dead short across the terminals -- a
small 6.8 uF dipped tantalum capacitor was also across the terminals,
and it had become such a good short that it hadn't gotten hot enough to
explode. (The power supply wires were something like #24 or #26, so
they'd limited the current.) My guess is that it went short just as soon
as the diode opened, and made a better quality short than the diode had.
The fuse is now back in place (along with new diode and capacitor), so
of course I haven't reverse connected the supply since.

Roy Lewallen, W7EL