14
\$\begingroup\$

I am building a device that shortens the lifespan of an old style "edison" bulb with an extra long filament as a kind of arbitrary timer.

I am planning to "stress" it (not immediately explode it or anything) by rectifying the current using an unsmoothed Full Wave rectifier assuming that the filament is designed for 230VAC with 60hz duty cycle therefore will do worse under noisy higher voltage DC.

I can also add in a relay turning on and off at 60hz which is very alarming to hear (cool!), would this cause it to cool and heat more times causing metal fatigue? Is there a way to push more current through a bulb and make it shine brighter?

I have a vague understanding that the current pull is the same, as is the RMS.

To Summarise: What causes more wear on a Filament and how can I increase it :D

\$\endgroup\$
  • \$\begingroup\$ I wouldn't expect the rectifier to have much effect at all, my guess is that something like switching on/off at more like 1Hz would be worse for metal fatigue. \$\endgroup\$ – PeterJ Aug 21 '13 at 7:28
  • \$\begingroup\$ @PeterJ: The time constant of the lamp is several seconds, maybe ten. 1Hz will be too fast \$\endgroup\$ – Federico Russo Aug 21 '13 at 7:38
13
\$\begingroup\$

There are three main causes of failure for filament bulbs:

  1. Evaporation of the filament. The more voltage you put across the bulb, the more current will flow through the filament. The more current that flows through, the more heat and brightness in the bulb, causing the filament to evaporate faster and fail. (see chart)
  2. Heating/cooling cycles will also add mechanical stress on the filament. This can be done by turning the bulb on and off at some duty cycle to allow it to heat and cool.
  3. High current inrush can also make a bulb fail early. Inrush is when the bulb is switched on and the bulb driver produces a very abrupt voltage transition.

enter image description here

The above image (source) was in reference to halogen lamps, but the failure/brightness curves should be the about same shape as for filament bulbs in general.

Burning out the bulb

Overvolting the bulb seems like the most straightforward way to cause controlled yet random failures. By this method you can control the lifespan of the bulb from months to seconds, depending on the voltage you set. The further above the bulb's specified voltage rating, the quicker it will fail. With a variable trasformer, you can easily dial in the voltage and approximate lifespan of your bulbs. (image source)

enter image description here

Heat cycling can't give you this kind of control, and generating precise inrush current is needlessly complex for your application.

One of the things that makes it so hard to predict the lifespan of the filament is that very small imperfections or defects in filament can have a dramatic affect of lifespan:

Small variations in resistivity along the filament cause "hot spots" to form at points of higher resistivity; a variation of diameter of only 1% will cause a 25% reduction in service life. The hot spots evaporate faster than the rest of the filament, increasing resistance at that point—a positive feedback that ends in the familiar tiny gap in an otherwise healthy-looking filament.

Source: Wikipedia

\$\endgroup\$
  • \$\begingroup\$ Thanks! Nice Graph! I did some preliminary tests, and rectification doesn't seem to make it shine noticeably brighter, I guess the current remains the same? Im trying to imagine a way to pull more current through it without adding more large bulbs or motors or burning a hole in the table :S \$\endgroup\$ – Oliver Kellow Aug 21 '13 at 7:40
  • 1
    \$\begingroup\$ @OliverKellow, I added to my answer to present how I would get more current to flow through the bulbs. \$\endgroup\$ – travisbartley Aug 21 '13 at 9:16
  • \$\begingroup\$ Filaments are pretty simple things. Rectification just changes the direction of voltage and current flow. But since the power is a product of the two, and -1*-1 = 1*1, reversing both does not affect power. (LEDs are not simple. In that case, changing the direction of the voltage does far more than just change the direction of current) \$\endgroup\$ – MSalters Aug 21 '13 at 10:15
7
\$\begingroup\$

You're more likely to shorten the relay's lifetime than the lamp's.

Rectifying is useless, as the peak voltage and power dissipated (and thus the heat) will be the same. The relay makes only sense if you allow the lamp enough time to sufficiently cool down. The best thing to kill it is to switch the relay on when the main's sine reaches its maximum on a cold bulb. The peak current then can easily reach ten times the nominal value. If the lamp hasn't cooled down enough the peak may only be twice nominal, and it won't do much harm. I think you'll need to let it cool down for at least half a minute.

Lamps used in darkroom enlargers are often rated at a lower voltage (e.g. 190V instead of 230V) to give a higher color temperature. These also have fairly short lifetimes.

\$\endgroup\$
  • 2
    \$\begingroup\$ To be pedantic: Peak voltage will actually reduce by 2 diode drops if a full wave rectifier is used, which might actually make the filament last longer :-) \$\endgroup\$ – Anindo Ghosh Aug 21 '13 at 11:51
1
\$\begingroup\$

It seems that the hotspots and mechanical stress of being turned on and off does the damage, and my rectifying the voltage does next to nothing because of the RMS. While over on undervoltage sounds cool and a good solutuon it's too large for my situation.

I am adding some CFL starters (wired in parallel in case one breaks) inline to have it switching on and off at in-equal intervals averaging about 1hz.

evil laugh

\$\endgroup\$
  • \$\begingroup\$ Did you perform testing to determine that? \$\endgroup\$ – Kortuk Sep 4 '13 at 15:00
  • \$\begingroup\$ @Kortuk Nope, Not scientifically. I built it and it's in an interactive artwork where the light is on constantly and is being stressed (as outlined above) for about three hours a day, and I am waiting for it to blow- its an unusual 60W "Rustika" filament rated at 1000 hrs, but of course the filaments are all different imperfect at some level so I guess its a bit null.. Thanks for your help everybody! Picture Here: embodiedmeaning.tumblr.com \$\endgroup\$ – Oliver Kellow Sep 7 '13 at 16:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.