"They Don’t Make ‘Em Like They Used To" Wrong.
The explanation why the bulb still holds is very simple. By turning it off and on again, the tungsten contracts and expands. That is why light bulbs break after a certain amount of time. If you keep the bulb lit or turned off, it won't break. In this case, the bulb was lit with the exception of very few power outages.
Also mentioned in the article: "Katz also adds that the bulb’s age could be, in part, contributed to the fact that it hasn’t been turned off and on a whole lot -- a process which is more exhausting on a bulb than letting it run continuously (the filament needs to reheat itself, much like a car’s engine)."
+1 This is exactly how I understood it to work...
"has been proclaimed the “Eternal Light” by General Electric experts and physicists around the world."
Not so much; it's operating far under the level of light it could put out and at atrocious efficiency. As said in the article: "Modern bulbs, she explains, use thinner tungsten filaments that put out more light (40 to 200 watts) burn hotter, and are therefore taxed more rigorously than older bulbs like the Shelby"
With semiconductors, the rule of thumb is an increase of 10 Kelvin halves the lifetime of the part. Considering this bulb is running several hundred Kelvin below modern bulbs, it isn't surprising it lasts. Run a modern 60W lightbulb at 30Vac and it'll last just as long too, but put out so little light as to be near useless (like this bulb).
Another common example is the light bulb behind a typical doorbell button. For example, the one on my parent's house is a 12V bulb, with the voltage turned down so it only produces a glimmer of light, enough to locate the button at night. It also runs continuously. So far it's clocked up 40 years and will probably outlast the structure of the brick wall it is attached to.
You pretty much nailed it. I lived down the street from this lightbulb as a kid and never heard of it until it became a phenomenon on the web. It should be said that they intentionally run it dim in order to prevent burnout. Didn't read the full article, but I'm guessing they mentioned the lightbulb also has its own UPS to deal with power outages/surges.
Unless I'm forgetting my P-Chem (possible, it wasn't a favorite subject), the 10 kelvin/halving lifetime loosely holds true for most chemical reactions that occur out of extremes -- e.g. that relationship holds true for most reactions that naturally occur on our planet. I still find that to be one of nature's neat factoids.
If you're at all interested in this topic, I can't recommend highly enough a little book by Tony Kordyban, Hot Air Rises and Heat Sinks: Everything You Know About Cooling Electronics Is Wrong. (The sequel, More Hot Air, is good too). The author makes the case, in Chapter 30, that those numbers, which came from MIL-HDBK-217, were pulled out of somebody's hat at a time when semiconductor reliability rates were changing rapidly due to process improvements, and probably don't have much validity any more. When actual reliability rates were compared to prediction and measured temperatures, there was no correlation.
These books are great.
Refs:
[1] Tony Kordyban. Hot Air Rises and Heat Sinks. New York: ASME Press, 1998. ISBN 0-7918-0074-1
[2] ---. More Hot Air. New York: The American Society of Mechanical Engineers, 2005. ISBN 0-7918-0223-X
Not completely wrong, they actually do not make them like they used to. There is evidence that lightbulb manufacturers forged a coalition and agreed to artificially shorten then mean lifespan of their products, a classic example of planned obsolescence. For an entertaining walk through, watch the documentary The Lightbulb Conspiracy.
Apparently, its continuing with LED's as well, as I've just had the 3rd "burn out" yesterday. I'm too lazy to count the number of bulbs (~20) and compute my own personal failure rates and expected lifetimes, but my seat of the pants estimates are that the 10 year lifespans are a bit optimistic, given that all three died in under a year (three different manufactures too). The ones that died were in open fixtures, so heat shouldn't be a problem either.
The first couple I took apart to determine if the LEDs were the problem (they weren't, the power supplies are the weak link), this one I'm going to send back to Cree to see if they actually honor the 10 year warranty, on my 3 month old bulb.
I know personal experience doesn't exactly count as rigorous scientific study, but I do have one light in my house that I literally never turn off in my kitchen. No automatic switch or anything, it burns 24/7/365.
It burns out roughly every 3 years (26,000 hours, pretty damned good, right.) That bulb is subject to various other environmental factors that will reduce it's lifespan (vibration and high heat.)
I've forgotten my point now, but I think it boils down to, a single factor, such as power cycling, isn't likely going to be the only reason a bulb doesn't last as long. A lot of factors play into it, including both environment and manufacturing.
Can definitely confirm this effect. Had a regular bulb in a standing lamp that was pointing upward toward a corner of my home office, that I never turned off. It lasted five years of being on 24/7. The other bulbs used in it (the lamp has three light inserts) have also lasted incredibly long durations (typically 2 to 3 years).
I have seen numerous light bulbs in my house break while being lit, and not just when switching them on and off. I am not very convinced by your explanation.
This article is terrible, priceonomics I'm ashamed of you.
> raises questions as to whether it is a miracle of physics, or a sign that new bulbs are weaker. Its longevity still remains a mystery.
It's not mystery at all. All you have to do to make a bulb last forever is derate it. The lower the temperature the longer it lasts, (and the less efficient it is).
At the dull red glow this one is emitting it'll last forever.
> Instead, they began to collectively engage in planned obsolescence. To achieve this the companies agreed to limit the life expectancy of light bulbs at 1,000 hours
I expected better from priceonomics than to repeat urban legends.
You can trade off long life and high efficiency. All that agreement did was decide that 1000 hours was a good trade off point, and standardize the luminous efficiency of incandescent bulbs so all incandescent bulbs of a specified wattage will be approximately the same brightness.
What really is ridiculous is the supposed agreement to limit bulbs to 1000 hours. Seems like GE and Sylvania weren't in on it, because before congress basically outlawed the common incandescent, I was buying long-life bulbs by those companies that were rated to last 20,000 hours. The only reason people's experience of light bulbs was that they only lasted 500-1000 hours was because they kept buying the 49-cent a bulb Wal-Mart crap.
Just a note that a site moderator has responded to this reply in their comments:
"First of all, we understand the correlation between low output and longevity, and acknowledge that the Centennial Light produces a very low output. Even so, out of hundreds of thousands of low wattage bulbs produced from 1880-1910, it's the only bulb of its kind that has burned for 113-years. Just a handful of bulbs on Earth are older than 60. We think that qualifies as mysterious on some level -- even when scientifically dissected.
Secondly, I appreciate you clarifying the complexity of the Phoebus cartel's actions (admittedly, it's a bigger picture than what we presented) -- but there exists compelling evidence that the cartel exercised planned obsolescence (we understand planned obsolescence to be when the "lifespan of a product is rendered artificially short by design"). Pre-1900, bulbs had lifespans of 1,200-2,500 hours; the cartel set bulbs at 1,000 hours -- even when GE (a member of the cartel) had access to modern tungsten filament technology that could've increased the output of bulbs without so aggressively compromising lifespan. While, as you said, the cartel's motivation was to standardize the haphazard market, the byproduct of this was an overly-assertive cap on longevity."
A crude bulb with a thick filament will last longer, as the current per unit area goes down. A thicker filament would of course drastically increase the life.
And reduce the voltage by 10%, an incandescent bulb will last 10x longer. With Tungsten the usual failure mode is excited evaporated material cooling and ending up on the cool bulb wall, this is the "blackening" you see on the bottoms of bulbs.
Tungsten is a pretty amazing material, but will wear down over time.
Is the Centennial Light actually superior to the classic light bulb?
The article states that the filament is 8 times thicker -- using 64 times more material -- and only runs at 4 Watts. So to match the standard 100 Watt bulb, you'd need 25 Centennial Lights, for a total of 1600 times more material.
Meanwhile, it's burned for 113 years, about 990 times the standard bulb's 1000 hour lifespan. 990 times better using 1600 times more material; should I be impressed?
Especially if you don't switch it on and off ... they've got that bulb on a UPS now so that it doesn't experience cooling/heating cycles which are what kill (or at least finally kill) most incandescent bulbs.
As an aside, it's my experience that what kills most LED bulbs is transients that fry their control circuits. Has anyone else noticed this?
What tends to kill incandescents is the non-uniform thinning of the tungsten wire due to tungsten evaporating, which causes the thinner parts to heat up more, combined with tungsten's positive temperature coefficient, which causes a current rush on power-up when the filament is cold, which causes the weak spots to heat up particularly rapidly and potentially higher that they would if the surrounding wire was hot and thus had higher resistance (a cold tungsten filament has about 1/15th the resistance compared to when it's glowing hot).
Transients certainly are one problem for LEDs if the power filtering is bad, another is (lack of) cooling - while LEDs produce much less heat, they also need to stay much, much cooler: A tungsten filament is fine at 2400 °C, an LED die should not get much hotter than 100 °C, preferably less than that, and most fixtures are not particularly good at getting rid of heat.
The power consumption of a bulb is not a measure of its brightness - due to its low temperature, this bulb's black body spectrum is shifted much further towards the red end of the optical range than a common 100 W bulb, which means that a larger proportion of its output is in the infrared, which is invisible, and much less of it is in the green, which is where human eyes are most sensitive. So, you'd need a lot more than 25 of those to achieve the same brightness (which still would not give you the same light color, though).
The only reason why you can somewhat compare modern incandescent bulbs by power is because the manufacturers have standardized on specific operating temperatures (the "cartel") - however, that only works for a given type of bulb, not between a "normal" one and a halogen one, for example, much less for the centennial light.
Still, all that indeed makes for even less reason to be impressed ...
30 comments
[ 3.1 ms ] story [ 82.6 ms ] threadhttps://en.wikipedia.org/wiki/Oxford_Electric_Bell
Also mentioned in the article: "Katz also adds that the bulb’s age could be, in part, contributed to the fact that it hasn’t been turned off and on a whole lot -- a process which is more exhausting on a bulb than letting it run continuously (the filament needs to reheat itself, much like a car’s engine)."
What experts and physicists?
With semiconductors, the rule of thumb is an increase of 10 Kelvin halves the lifetime of the part. Considering this bulb is running several hundred Kelvin below modern bulbs, it isn't surprising it lasts. Run a modern 60W lightbulb at 30Vac and it'll last just as long too, but put out so little light as to be near useless (like this bulb).
Unless I'm forgetting my P-Chem (possible, it wasn't a favorite subject), the 10 kelvin/halving lifetime loosely holds true for most chemical reactions that occur out of extremes -- e.g. that relationship holds true for most reactions that naturally occur on our planet. I still find that to be one of nature's neat factoids.
These books are great.
Refs:
[1] Tony Kordyban. Hot Air Rises and Heat Sinks. New York: ASME Press, 1998. ISBN 0-7918-0074-1
[2] ---. More Hot Air. New York: The American Society of Mechanical Engineers, 2005. ISBN 0-7918-0223-X
I submit Rechtgefühleinsamkeit as a candidate.
The first couple I took apart to determine if the LEDs were the problem (they weren't, the power supplies are the weak link), this one I'm going to send back to Cree to see if they actually honor the 10 year warranty, on my 3 month old bulb.
It burns out roughly every 3 years (26,000 hours, pretty damned good, right.) That bulb is subject to various other environmental factors that will reduce it's lifespan (vibration and high heat.)
I've forgotten my point now, but I think it boils down to, a single factor, such as power cycling, isn't likely going to be the only reason a bulb doesn't last as long. A lot of factors play into it, including both environment and manufacturing.
The idea is: Turn it on, metal expands (thermal expansion) Turn it off, metal contracts
This is the repeatedly applied load, hence eventual material failure.
Also, the hotter you run the bulb, the more tungsten evaporates.
http://www.hugtherhino.com/2013/04/planned-obsolescence-dark...
> raises questions as to whether it is a miracle of physics, or a sign that new bulbs are weaker. Its longevity still remains a mystery.
It's not mystery at all. All you have to do to make a bulb last forever is derate it. The lower the temperature the longer it lasts, (and the less efficient it is).
At the dull red glow this one is emitting it'll last forever.
> Instead, they began to collectively engage in planned obsolescence. To achieve this the companies agreed to limit the life expectancy of light bulbs at 1,000 hours
I expected better from priceonomics than to repeat urban legends.
You can trade off long life and high efficiency. All that agreement did was decide that 1000 hours was a good trade off point, and standardize the luminous efficiency of incandescent bulbs so all incandescent bulbs of a specified wattage will be approximately the same brightness.
You can read about it here: https://en.wikipedia.org/wiki/Incandescent_light_bulb#Light_... - it even mentions this very bulb.
Note that lifetime goes by the power of 16!!! so it's very easy to derate a bulb and make it last forever.
"First of all, we understand the correlation between low output and longevity, and acknowledge that the Centennial Light produces a very low output. Even so, out of hundreds of thousands of low wattage bulbs produced from 1880-1910, it's the only bulb of its kind that has burned for 113-years. Just a handful of bulbs on Earth are older than 60. We think that qualifies as mysterious on some level -- even when scientifically dissected.
Secondly, I appreciate you clarifying the complexity of the Phoebus cartel's actions (admittedly, it's a bigger picture than what we presented) -- but there exists compelling evidence that the cartel exercised planned obsolescence (we understand planned obsolescence to be when the "lifespan of a product is rendered artificially short by design"). Pre-1900, bulbs had lifespans of 1,200-2,500 hours; the cartel set bulbs at 1,000 hours -- even when GE (a member of the cartel) had access to modern tungsten filament technology that could've increased the output of bulbs without so aggressively compromising lifespan. While, as you said, the cartel's motivation was to standardize the haphazard market, the byproduct of this was an overly-assertive cap on longevity."
A crude bulb with a thick filament will last longer, as the current per unit area goes down. A thicker filament would of course drastically increase the life.
And reduce the voltage by 10%, an incandescent bulb will last 10x longer. With Tungsten the usual failure mode is excited evaporated material cooling and ending up on the cool bulb wall, this is the "blackening" you see on the bottoms of bulbs.
Tungsten is a pretty amazing material, but will wear down over time.
The article states that the filament is 8 times thicker -- using 64 times more material -- and only runs at 4 Watts. So to match the standard 100 Watt bulb, you'd need 25 Centennial Lights, for a total of 1600 times more material.
Meanwhile, it's burned for 113 years, about 990 times the standard bulb's 1000 hour lifespan. 990 times better using 1600 times more material; should I be impressed?
Not really. Run a 100 watt lightbulb dimmed down to the brightness of a 1 watt bulb and it will also likely run for 100 years.
As an aside, it's my experience that what kills most LED bulbs is transients that fry their control circuits. Has anyone else noticed this?
Transients certainly are one problem for LEDs if the power filtering is bad, another is (lack of) cooling - while LEDs produce much less heat, they also need to stay much, much cooler: A tungsten filament is fine at 2400 °C, an LED die should not get much hotter than 100 °C, preferably less than that, and most fixtures are not particularly good at getting rid of heat.
The only reason why you can somewhat compare modern incandescent bulbs by power is because the manufacturers have standardized on specific operating temperatures (the "cartel") - however, that only works for a given type of bulb, not between a "normal" one and a halogen one, for example, much less for the centennial light.
Still, all that indeed makes for even less reason to be impressed ...