Path: utzoo!attcan!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!elroy.jpl.nasa.gov!usc!jarthur!ucivax!orion.oac.uci.edu!ucsd!ucbvax!pasteur!cory.Berkeley.EDU!charless
From: charless@cory.Berkeley.EDU (Charles R. Sullivan)
Newsgroups: sci.electronics
Subject: Re: incandescent light bulb life extender
Summary: failure at turn-on due to hot spot overshoot.
Message-ID: <9859@pasteur.Berkeley.EDU>
Date: 18 Dec 90 01:56:32 GMT
References: <6213@videovax.tv.tek.com> <4078@osc.COM> <1990Dec13.172348.20146@amd.com> <1990Dec13.204415.22627@athena.cs.uga.edu> <schumach.661141590@convex.convex.com>
Sender: news@pasteur.Berkeley.EDU
Reply-To: charless@cory.Berkeley.EDU
Lines: 57

In article <schumach.661141590@convex.convex.com> schumach@convex.com (Richard A. Schumacher) writes:
>Okay, several people have stated that in-rush (high
>initial current) reduces an incandescent's life. How? 
>What's the mechanism?

Before I describe failure on turn-on, let me describe 'normal' end-of-life.
It's got an interesting wrinkle:

A perfect filament would gradually evaporate, equally from all points on
its surface.  Thus it would get thinner, and the resistance would go up.
When it got to be a higher resistance it would run cooler, and evaporate
more slowly.  Because the evaporation is exponential with temperature,
it would never evaporate completely, but just gradually get dimmer.    

In real life, the filament has some defects--spots that are a bit thinner
than the rest of the filament.  These have the same current as the rest of
the filament, but higher resistance (per length).  So they run hotter.
This makes the tungsten evaporate faster, which makes the spot get even
narrower, which makes it run hotter (it is still only a small fraction of
the total resistance, so the current is constant.), and so on....until
that spot breaks.  ( a good reason not to buy low-quality bulbs)

Now consider start-up.  The cold resistance of the tungsten is ~10x
less than hot. (mesure a light bulb with an ohmmeter and calculate 120^2/R. 
It will be ~10x the rated watts).  So at turn-on you get a big current
until the filament has heated up (actually only ~10 cycles of 60 Hz).
The defect spots will not only be heated up more by having a high resistance,
but they will also get heated up faster, because they have less thermal mass
(actually thermal mass to surface area ratio is what matters).  This fast
heating continues not until the hot spots reach their operating temperature, 
but until the rest of the filament reaches its operating temperature.  
(Remember, the rest of it has most of the resistance, and so controls the
current.)  Thus the hots spots will overshoot their steady-state operating
temperature, which is already hotter than the proper operating temperature.

This overshoot is only for ~1/6 second, so the extra evaporation during the
overshoot is not significant for accelerating the aging of the bulb.  But
if the bulb is very close to failure, and the hot spots are really bad,
they can overshoot to the point where the tungsten softens or melts, and the
filament can break.  Hence my assertion that the only time a soft start would
help was when the bulb was close to failure anyway.  

The above qualitative argument doesn't give a number for what 'close to
failure' is.  If it were, say 300 Hrs., that would be significant, but
my understanding is that it is something less than about 8 hours, so
it doesn't make much difference in bulb life.  Thus the only difference
the soft start feature of 'life extenders' makes is that now
your reading light will quit after you are settled into your armchair with
a book, instead of when you first turn on the light.
(The voltage drop across the extender during normal operation
does extend the lamp life significantly, but with an even more significant waste
of energy.)

If you get a compact fluorescent, it won't fail when you turn it on *or* after
you sit down! :-) (until much later, >~10x life of an incandescant).

Charlie Sullivan                            charless@cory.Berkeley.EDU