12
\$\begingroup\$

I have been hearing a common saying that if one keeps switching the lights on and off you will probably damage the light bulb itself, since every time you close the switch there would be a sudden rush of current through the circuit. Given that we are talking about modern light bulbs you would find in a normal household environment (incandescent /fluorescent/LED), will repeatedly switching it on and off cause long term damage to the light bulb?

I personally do not think it will because of the fact that the initial rush of current will not even have enough energy to cause any noticeable effect. That's what I believe, but I'm not sure if that's true or not. Aren't those lights in decorations and signs also flashing all the time? I don't see them wear out faster.

\$\endgroup\$
  • \$\begingroup\$ I vaguely remember reading many years ago about a factory whose managers had determined that it was cheaper to leave the fluorescent lights on 24 hours a day than to turn them off during the day. The extra energy costs were more than offset by the lower lamp replacement rate. \$\endgroup\$ – Pete Becker Apr 3 '15 at 14:12
12
\$\begingroup\$

It depends on the type of lightbulb!

Halogen, incandescent, florescent, and vapor lights all use tungsten filaments that heat up and emit electrons via thermionic emission. In that sense, they are similar. However, the method to "turn on" the lights varies.

Incandescent bulbs are simply turned on once and left on. The inrush current is on the order of 12 to 15 times the peak current if not limited by the methods described in the application note.

Florescent bulbs operate by a "starter" and "ballast" design. The filaments heat up more gradually since the starter (D in the diagram below) has to switch multiple times in order to kick-start the electrons flowing through the tube, not just one time like the incandescent light.

enter image description here

Basically, the starter (a bi-metallic switch) heats up and opens periodically, causing the magnetic field generated by the ballast (G) to collapse and release an inductive kick into the tube. If the kick isn't strong enough, there won't be enough electrons to sustain the circuit through the tube and the light will flicker. The light will only sustain when the magnetic field is strong when it collapses. For an animation of this, check out "How a Florescent Light Works".

Anyway, the idea is that the tungsten element undergoes thermal shock every time the light is turned on. I conjecture that the thermal shock is less for a florescent than for an incandescent, since the florescent lights are not immediately heated up to full throttle because the starter has to try multiple times to start the light (usually over a period of several seconds). Either way, turning on the light every time does damage the filament and will result in long term damage.

The LED however, is the only type of light emitting device out of the list that doesn't use a tungsten element. It uses a PN junction instead. This means that the LEDs require much less voltage and current, meaning low power consumption compared to the lights with filaments. As such, LEDs won't be damaged at all by switching, since there is no filament to damage and the power going through the bulb is lower. In fact many applications switch them at high speeds using PWM which they handle with no problem.

Also, check out MinutePhysics' great video on modern lights for a short explanation of how these lights work!

\$\endgroup\$
  • \$\begingroup\$ Thanks for the detailed answer! So does that mean turning it on and off actually does more damage than keeping it on for the same amount of time? Since most light bulbs are powered by AC current anyway, won't that also contribute to the overall damage? \$\endgroup\$ – Derek 朕會功夫 Apr 3 '15 at 4:36
  • \$\begingroup\$ I am not sure but I think the inrush current damages the filament more than if the light was already heated up and left on for the same amount of time. Over time the tungsten oxidates due to the extreme heat inside the bulb and gets thinner, but it is the thermal shock that does the real damage. I guess think of it as a rubber band. You use it to hold something together, and it can stay there for a long time happily. But every time you stretch it, it gets damaged by the tensile force. Eventually you go to stretch it one last time and it snaps. \$\endgroup\$ – FullmetalEngineer Apr 3 '15 at 5:04
  • 1
    \$\begingroup\$ The AC current doesn't really contribute any "damage". The electrons don't damage the filament when they switch direction. It's the thermal shock and the heat that damages it. \$\endgroup\$ – FullmetalEngineer Apr 3 '15 at 5:05
  • \$\begingroup\$ It's unexpected for me that they didn't make some sort of special circuit to resist this inrush. Won't simply putting an inductor somewhere in the circuit can already stop any sudden change of current? \$\endgroup\$ – Derek 朕會功夫 Apr 3 '15 at 7:32
5
\$\begingroup\$

According to the U.S. Dept. of Energy:

  • It is best to turn off incandescent and halogen bulbs whenever they are not needed, due to their high consumption of electricity.
  • For a compact fluorescent bulb, a rule of thumb is to leave it on if you leave a room for 15 minutes or less (depending on several factors).
  • For LED lighting the operating life is unaffected by turning it on and off.

https://energy.gov/energysaver/when-turn-your-lights

\$\endgroup\$
2
\$\begingroup\$

General rule of thumb is every time you turn a light on and off it will shorten its life span, but this also applies to leaving lights on 24/7.

Inrush Current: An example of inrush current is an LED downlight fitting with 9w (0.0375A at 240v) will have any average inrush current of 7A for 300ms (not enough time to trip a circuit breaker breaking contact at 400ms).

Thermal Expansion: The more faulting factor is the temperature stress (thermal expansion) on the driver and control gears (Ballasts, LED control gears, transformers etc.). Every time something heats up (which is anything electrical due to resistance), it needs to cool back down. This causes expansion and contraction on cable joints, soldered or terminated, causing faults and will ultimately result in PCB's (printed circuit boards) to burn out resistors, dislodge cabling from contacts and arc contacts/cabling. This is why you see control gears fail consistently in LED fittings that have 50,000hour lamp lifespans.

This is a common occurrence in circuit breakers and fuses. As time goes on, screw fastened terminations will begin to expand, pushing termination screws to undo, but when it contracts, there is a arc gap between terminals. This causes a hot joint.

Sorry for the long winded reply but I have been asked this question in the past.

\$\endgroup\$
  • \$\begingroup\$ Where did you get your 7 A for 300 ms numbers from? The LEDs themselves don't have any cold inrush so the only thing to account for that would be internal capacitance after the rectifier. 7 A for 300 ms = 2.1 C (coulombs) and then \$ C = \frac {Q}{V} = \frac {2.1}{240} = 8,750\ \text {µF} \$ with a 400 V rating and that would be much larger than the lamp. \$\endgroup\$ – Transistor Sep 22 '18 at 15:52
  • \$\begingroup\$ lighting.philips.com/main/prof/indoor-luminaires/downlights/… This is direct from a datasheet from Phillips \$\endgroup\$ – Bradicul Sep 22 '18 at 16:31
  • \$\begingroup\$ Wow! That's mental. I might ask a question on that myself. I can't imagine what can be drawing 1.75 kVA on a 9 W lamp. Thanks for the link. \$\endgroup\$ – Transistor Sep 22 '18 at 16:45
  • \$\begingroup\$ Yeah I know, don't worry I checked myself first time I heard it too. Unfortunately most commercial buildings projects managers (major works) are using these specs as a guide for green saving and daylight harvesting but neglect to mention the finer details. New construction works in Sydney Australia are currently using these in individual tenancy fit-outs as a cost saving alternative. Cant wait until phase balancing for the neutral takes it toll on morning startups and afternoon sunrise/sunset inputs from the KNX/Cbus/Dynalyte systems. \$\endgroup\$ – Bradicul Sep 22 '18 at 16:55

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.