Why are incandescent bulbs blowing/burning out? How to prevent?

I'm using incandescent mini Christmas bulbs in my project. In testing I found the bulbs do not blow until the voltage is around 7 or 8 volts. Lights are connected to my Arduino with a MOSFET. All of the V+ in and GND are bussed together or are in parallel with the power supply.

Every once in a while one or two of the bulbs will blow. Might be workig fine an hour or so and then one blub is dead.

If I replace the blub everything is fine, and another blub blows.

Now when I look closely at the blub I can see a black ring inside the glass of the blub. I'm thinking this is indicating the light blew becase of a voltage spike? (Not sure.)

Am I correct in my assumption the blubs are blowing by a voltage spike?
The arduino is powered separtly/USB to comptuer. The incandescent blubs are the only thing powered by the wall wart.

I'm thinking this is a voltage spike which is just happens to occur when the light is on. What' the best way to project this from occuring?
A zener diode? Something like a 1N5919BG? And do I need a resistor?

Or would a better approach be to use a 9v or 12 v power supply with a 12 to 5 volt regulator.

I am hoping to power the Arduino with the same wall wart that's powering the incandesent blubs.
Wall wart is from a Netgear router. Rated at 5v/2.5a Measured voltage is 5.5.

Biggest question, what was the rated voltage of the lights? Just connecting them to 8V and see they don't blow right away isn't very exact :wink:

And is that wall wart AC or DC? Did you measure it? A low loaded AC wall wart (real transformer) can easily output 10V+.

septillion:
Biggest question, what was the rated voltage of the lights? Just connecting them to 8V and see they don't blow right away isn't very exact :wink:

And is that wall wart AC or DC? Did you measure it? A low loaded AC wall wart (real transformer) can easily output 10V+.

These are the cheap Christmas tree lights which are in parallel/series. I'm in the US so 50 +50. I'm in the US so 120AC/170RMS. I believe the voltage drop/working voltage is 1.5 per light but can't remember at the moment.

I connected the blubs to a variable power supply. At 1.5/2 volts DC about the same brightness as at 120vac. As I increased the voltage to 4 or 5 they appear to be the correct brightness. I suspect these bulbs are made for 220VAC and also sold in 120VAC market.

To my surprise the blubls would consistantly blow at about 8 volts. So I'm thinking driving them at 5VDC should not be a probelm. And for the most they've been working fine.

Wall wart is not a transformer, not heavy enough, there's a chip in there.

I'm wondering if maybe I should add a resistor in series to drop the voltage a bit? Or a zener? Or maybe switch to 12v, and use a bcuk voltage reducer. And that will give me 12 to power the Arduino.

What is "120AC/170RMS"? In the US you have 120VAC RMS. And there are 50 series sets of 2 in parallel? (Or 2 sets of 50 in series in parallel but that results in the same.) That is an odd arrangement for 240V countries which in which they probably just sell 100 in series which results in the same voltage. That would make them 2,4V each. Just don't go over that, done :wink:

Messing around with "looks to be the same brightness" is just to vague. Or eyes are terrible at spotting differences like that. And the voltage they directly blow at and the voltage they might blow at after some time are completely different :wink:

Doug101:
Wall wart is not a transformer, not heavy enough, there's a chip in there.

But did you measure it? :wink:

Doug101:
I'm wondering if maybe I should add a resistor in series to drop the voltage a bit?

Might be an option but you need to measure the current for that as well.

Doug101:
Or a zener?

And leave to rest of the voltage where?

Doug101:
Or maybe switch to 12v, and use a bcuk voltage reducer. And that will give me 12 to power the Arduino.

Why 12V? The Arduino is a 5V device (even less at lower speeds). Will work with 5V fine, only connect it to the 5V/Vcc pin or USB, no to Vin. Only don't simultaneously connect it to your computer (that's why doing it via the USB is a safe option).

[/quote]

The grey ring indicates that the bulb is nearing the end of it's life - all chrismas light bulbs slowly get that while in use. I have noticed wide variation in just how dark the grey ring is at the time of bulb failure, but no apparent pattern, other than that the longer the bulb has run, the darker the ring.

If they're starting out with no ring, but getting one within timescale of hours, you are driving them too hard, apply a lower voltage, or put more in the string.

In the US (120V mains), christmas light strings have one or more series strings of lights. The strings are generally either 35 or 50 lights long (3.5v or 2.5v bulbs).

Once christmas lights get to a certain age, they become more fragile, and will fail upon handling (these will typically have the grey ring at least faintly visible in most or all of the bulbs). Often these will fail without the shunt working, making the entire string go dark.

Always replace bulbs immediately when they fail. Spare bulbs are not readily available at sane prices; I've resorted to buying extra strings for bulbs. When the bulb fails, the shunt makes contact to bypass the failed filiment so the rest of the string stays lit - but this results in a higher voltage across the other bulbs, and the influence of applied voltage on bulb lifetime is huge.

(I use christmas lights for general illumination, and have since I got my own apartment after graduation nearly a decade ago. I replace the bulbs (have gotten really good at swapping the plastic bases), and would save the dead ones (sans plastic base) - when I threw it out had a 2.5" x 2.5" x 10" drawer full of burned out bulbs. I still have christmas lights. I no longer keep the burned out bulbs).

120V RMS, 169.7 peak. :wink: And don't forget, increasing voltage by 41% DOUBLES wattage,
delta P (Watts) = delta V squared.

Don't the bulbs have a voltage printed on the ?

If not , count how many in series across your 120vac supply.

The life of a bulb is inversely proportional to the applied voltage to the 4th or 5th power, so overvolting seriously decreases life.

Allan

If the string is working with 120V, pull-out one lamp. All of the bulbs in series will go out. If any stay-on they are in a separate series-string.

If you have a series string of 50, that's 120V/50 = 2.4V per bulb. The energy/wattage from 120V RMS is the same as 120VDC. (Don't worry about the peak AC voltage.)

So with 5VDC, two in series would be slightly over-voltage but probably OK.

DVDdoug:
If the string is working with 120V, pull-out one lamp. All of the bulbs in series will go out. If any stay-on they are in a separate series-string.

If you have a series string of 50, that's 120V/50 = 2.4V per bulb. The energy/wattage from 120V RMS is the same as 120VDC. (Don't worry about the peak AC voltage.)

So with 5VDC, two in series would be slightly over-voltage but probably OK.

It is string of 100. 2 strings of 50 bulbs in series. The 2 srings in series are in parallel with the mains. A 2.5v drop per bulb sounds right.

The thing is at 120 vac the bulbs are dim and on the yellow side. At 5 volts they look bight like they should. THis makes me think these blubs were desiged for a 220v market. Instead of having 2 voltages of bulbs one set for 120vac and the other for 220 vac they just use the same bulbs in both markets.

The bulbs work fine at 5.5 vdc. They blow at around 8 volt. The black ring in the glass makes me think they are blowing from a voltage spike and not over voltage. I would also think they would all blow at about the same time if it were from over voltage. Since only one or two blow at a time I'm thinking spike.

As I side note I know these strings of lights are made by chineese prisioners. If they install 7,000 bulbs per day they receive a full ration of food. Less than 4,000 and they go hungry.

Doug101:
Or would a better approach be to use a 9v or 12 v power supply with a 12 to 5 volt regulator.

What do you mean by 12 to 5 V regulator? If you have a 12 V DC supply, the regulator could take in 12V and maybe regulate 10V DC maybe.

Also, were those christmas tree lights ever operated for long periods while supplied with whatever they were originally packaged up with? Or these are mail-order ones where you need to provide your own power supply?

If the lights are known to work for a long time with whatever they're supposed to be driven with, then probably makes sense to follow that same powering method. Otherwise, it could be a matter of trial and error, like what you're doing now ....... ie. it seems that you're getting better reliability with driving with a lower voltage. So that might be key here.

Otherwise, might need to take a look at what the input voltage waveform (at the input to the lights, relative to ground) looks like on a measurement instrument ...... to see if everything appears as expected.

To put the expected life of an incandescent lamp into perspective, the life changes rapidly with changes in voltage.

In general the life changes with the 12th power of the voltage ratio.

So if a bulb is rated for 5V and you are applying 8 V the expected life would be (5/8)^12 or 0.3% or the original expected life.

The same is true for reduced voltages (but increasing the life)

JohnRob:
To put the expected life of an incandescent lamp into perspective, the life changes rapidly with changes in voltage.

In general the life changes with the 12th power of the voltage ratio.

So if a bulb is rated for 5V and you are applying 8 V the expected life would be (5/8)^12 or 0.3% or the original expected life.

The same is true for reduced voltages (but increasing the life)

And the life of a mini-Christmas light is.... not rated.
And for the voltage... Not stated.

I just tired one of the lights with my varible power supply.
At 10 volts it's nice a and bright. (Almost too bright.)
5 looks about right.
3 volts looks a bit anemic.

LEDs are cheap and very reliable....

Allan

allanhurst:
LEDs are cheap and very reliable....

Allan

Not sure what your point is? LEDs just like incandesent blubs burn out too from over voltage.

I was using a LED in a circuit with a solenoid and have burned several out while prototyping.

For my project and this applciaon I specifily selected incandescents blubs based on their properities.

Not sure what your point is? LEDs just like incandesent blubs burn out too from over voltage.

Quite. That's why you always drive them with a current-limited supply.

Allan

In the old days there were safety fuses (wire fuses). In theory they blow before the components. Not always, in real life.

Led with solenoid?
Back-emf?

Sounds like you didn't observe the LED's absolute max reverse voltage of 5volt.
Leo..

allanhurst:
Quite. That's why you always drive them with a current-limited supply.

Allan

Please explain how that works? If a current-limited supply is used, when I need 50 LEDs lit won't the current be limited?

Doug101:
The bulbs work fine at 5.5 vdc. They blow at around 8 volt.

But from earlier discussions they look like they will get 3V per bulb in normal use, that is 3V RMS which is a peak voltage of 4.2V. You would be better off just running them at 3V.

The black ring in the glass makes me think they are blowing from a voltage spike and not over voltage.

There is no correlation and you can not draw that conclusion. In fact an incandescent bulb is much more resilient to voltage spikes than just about any other component.

I would also think they would all blow at about the same time if it were from over voltage.

That would be wrong. Any component driven over the design rating will fail at random. The more over stressed they are the more often is a failure, but they do not fail at the same time unless they are well over the rating. You are in the grey zone which is not enough to trigger instant death but too high for reliable working.

The black ring in the glass makes me think they are blowing from a voltage spike and not over voltage.

In fact the very opposite. The black ring implies that the filament has partially evapourated due to over-temperature for quite a long period , which thins and weakens it.

The normal method of current limiting a LED is with a series resistor, whose value is calculated so as to limit the current to the LED's maximum current rating or below.

Allan