Street Lights turning Purple

[rgarito]

All over America (at least), there are numerous reports of street lights turning purple.

The issue is caused by some sort of "manufacturing defect" of the LED bulbs that many cities have used to replace their old lamps.

Here's an example:
https://www.fox13news.com/news/failing-street-lights-in-tampa-emit-bright-purple-hue

A simple Google search reveals 1000's of hits.

Sounds like some LED manufacturer(s) gonna be replacing a lot of LEDs!!!   

[Whales]

Almost all commercial white LEDs are blue LEDs with phosphors ontop.  These street lights sound like they're losing their phosphor layers.

Temperature x hours too much for them?  Or maybe something more creative like the phosphors flaking off due to moisture and cycling?

[sleemanj]

I think I recently saw somebody look at this sort of failure where the phosphor detached... maybe it was BigClive.

Edit: ahh it was the lg tv thread below I was thinking of

[floobydust]

It's been attributed to phosphor failure. I thought it's the same problem affecting LCD TV backlights https://www.eevblog.com/forum/chat/why-do-backlight-leds-burn-out-and-go-blue/

The debacle of Detroit's streetlight failures- over 20,000 "prematurely dimming and burning out" cost almost $7M to replace them. They failed melting the lens.
Detroit Sues Lighting Manufacturer Over Failed Street Lights The City sued Leotek Electronics, got only $4M in settlement.

Point is it's very expensive labour-wise to swap out defective streetlights, about 1,500 a week were done and the City of Detroit contracted out all streetlight work, they didn't have their own crews.

[james_s]

My friend who works in the streetlighting industry was just talking about this the other day when I was hanging out with him. He said it is a particular brand and model of luminaire that is having this problem, he's been working on getting me a defective one to do a failure analysis on.

[jonpaul]

the cities purchase the cheapest Chinese junk.

Buy Européen or American not  China
j

[james_s]

IIRC it is an American company's products that are doing this, I think he said it was Cooper Lighting but I'll check.

[AVGresponding]

Regardless of the nationality of the manufacturer the LEDs themselves are likely made in China nowadays.

My main gripe with LED lights is that you have to replace the entire fitting when they go poof, rather than just a lamp or ballast. Any organisation looking at reducing their overall costs by installing low energy LED lighting is going to be in for a nasty surprise a few years from now.

[floobydust]

Cobra head LED light fixtures fail for many reasons - overheating, water ingress, cheap/overstressed components like the LED's, electrolytic capacitors etc. and thermal cycling, bad driver design, lightning/transients, loose bolts falling down and many more reasons.

For the purple/blue streetlights, "American Electric Lighting (AEL) a division of Acuity Brands Lighting products" seems to be just a middleman, a dealer of numerous lighting brands.
Their warranty is the usual 1 year, liability limited to the purchase price of the product. A few warranties went to 6 years on select components.
Our municipality's tax dollars pay for these failures but it's always hush hush and swept under the rug because it's very expensive to replace streetlights, labour-wise.

For Detroit, LEOTEK is part of Lite-On Group.
So is their melting LED "cheap" or driven at the datasheet maximum? You want the highest lumens number in order to make the sale, who cares about lifetime- by then you will have bought your Lambo and can blame someone else.

[SiliconWizard]

That's funny, especially when you think about the gigantic cost behind the replacement of all street lights by LED-based ones, allegedly to draw less power and/or require less maintenance.

But in hindsight, with real figures, have they really lowered cost, energy included - or will they ever during their lifetime - compared to sodium vapor lights? Actually, efficiency is on par, or was even in favor of sodium vapor until recently. As to lifetime, as we all know, this is just mostly a lie. LEDs as built and used in most street light are not made to last. Yes, I've also witnessed changes of brand new (like one-year old) street lights due to early failure. It seems common.

So, where are the real figures, and independent analysis of that? This is our tax money.

The two real "benefits" of LEDs for street lights are that they can be almost instantly turned on, while the sodium vapor ones take a few minutes of warm-up time. The other is that they have a much broader spectrum. Useful in particular for surveillance cameras.

[james_s]

I asked my friend, he said it's American Electric ATB0 and ATB2 luminaires that have been turning blue.

The melting Leotek lights are another common problem, it isn't the LEDs in those but the optics. They get dirty and it absorbs heat and causes them to darken which of course absorbs more heat and eventually the lens melts completely. Apparently they're pushing a lot of optical power through a relatively small lens.

[james_s]

The two real "benefits" of LEDs for street lights are that they can be almost instantly turned on, while the sodium vapor ones take a few minutes of warm-up time. The other is that they have a much broader spectrum. Useful in particular for surveillance cameras.

The major benefits of LED is that they are about half the wattage of the HPS lamps they replace, and they produce nice white light. The latter is huge IMO, at least to me. Personally my favorite is still the old mercury vapor lights that were common up into the 80s. They are less efficient on paper but they do a better job of lighting things up than HPS, the orange light is absorbed by so much and just makes greenery look brown and gross. The spectral lines of a mercury lamp make plants practically glow.

[Ben321]

Almost all commercial white LEDs are blue LEDs with phosphors ontop.  These street lights sound like they're losing their phosphor layers.

Temperature x hours too much for them?  Or maybe something more creative like the phosphors flaking off due to moisture and cycling?

The phosphors are inside of the LED component itself. There's the blue gallium nitride blue emitter chip itself, which is then coated with a layer of yellow fluorescent material. This entire structure then is fully encapsulated inside the dome shaped piece of clear plastic. And by encapsulated I mean that liquid plastic is poured over the chip+phosphor structure that's sitting in a dome shaped mold so that the chip+phosphor structure is completely potted. There's absolutely no way for it to flake off. Overheating could burn out LED or even explode it, but the phosphor can't flake off while the LED is operational because it's physically sealed by the surrounding plastic. And of course if it burns out or blows up, the LED stops emitting all light.


I have a hypothesis about the ones turning purple or blue. I think that there's those are using tri-color non-phosphor LEDs. These are the same type that are used in RGB LED advertising signs that can display bright full color pictures on billboards. Except here they are being used not for image display, but rather for illumination. My guess is that the drivers for green LED chips and red LED chips are failing faster than the ones powering the blue LEDs, resulting in the blue color you see from them when they fail. While most LED streetlights use the normal white phosphor design, I think a few are using tri-color LEDs and these are the ones that turn blue when they fail.

[wraper]

The phosphors are inside of the LED component itself. There's the blue gallium nitride blue emitter chip itself, which is then coated with a layer of yellow fluorescent material. This entire structure then is fully encapsulated inside the dome shaped piece of clear plastic. And by encapsulated I mean that liquid plastic is poured over the chip+phosphor structure that's sitting in a dome shaped mold so that the chip+phosphor structure is completely potted. There's absolutely no way for it to flake off. Overheating could burn out LED or even explode it, but the phosphor can't flake off while the LED is operational because it's physically sealed by the surrounding plastic. And of course if it burns out or blows up, the LED stops emitting all light.

Have you ever seen an SMD LED? I have never seen any lighting solution with outdated through-hole LEDs.

[themadhippy]

Will the traffic lights be turning blue tomorrow?

[Zero999]

Almost all commercial white LEDs are blue LEDs with phosphors ontop.  These street lights sound like they're losing their phosphor layers.

Temperature x hours too much for them?  Or maybe something more creative like the phosphors flaking off due to moisture and cycling?

The phosphors are inside of the LED component itself. There's the blue gallium nitride blue emitter chip itself, which is then coated with a layer of yellow fluorescent material. This entire structure then is fully encapsulated inside the dome shaped piece of clear plastic. And by encapsulated I mean that liquid plastic is poured over the chip+phosphor structure that's sitting in a dome shaped mold so that the chip+phosphor structure is completely potted. There's absolutely no way for it to flake off. Overheating could burn out LED or even explode it, but the phosphor can't flake off while the LED is operational because it's physically sealed by the surrounding plastic. And of course if it burns out or blows up, the LED stops emitting all light.


I have a hypothesis about the ones turning purple or blue. I think that there's those are using tri-color non-phosphor LEDs. These are the same type that are used in RGB LED advertising signs that can display bright full color pictures on billboards. Except here they are being used not for image display, but rather for illumination. My guess is that the drivers for green LED chips and red LED chips are failing faster than the ones powering the blue LEDs, resulting in the blue color you see from them when they fail. While most LED streetlights use the normal white phosphor design, I think a few are using tri-color LEDs and these are the ones that turn blue when they fail.

Why would the green and red LEDs fail first? If they were RGB then a good number of them would turn yellow, or cyan, rather than purple.

It's phosphor failure. The reason why some change purple is because violet, rather than blue LEDs are often used to excite the phosphor.

Cobra head LED light fixtures fail for many reasons - overheating, water ingress, cheap/overstressed components like the LED's, electrolytic capacitors etc. and thermal cycling, bad driver design, lightning/transients, loose bolts falling down and many more reasons.

For the purple/blue streetlights, "American Electric Lighting (AEL) a division of Acuity Brands Lighting products" seems to be just a middleman, a dealer of numerous lighting brands.
Their warranty is the usual 1 year, liability limited to the purchase price of the product. A few warranties went to 6 years on select components.
Our municipality's tax dollars pay for these failures but it's always hush hush and swept under the rug because it's very expensive to replace streetlights, labour-wise.

For Detroit, LEOTEK is part of Lite-On Group.
So is their melting LED "cheap" or driven at the datasheet maximum? You want the highest lumens number in order to make the sale, who cares about lifetime- by then you will have bought your Lambo and can blame someone else.

Laws vary. In some countries the manufacturer still has a statuary obligation to the customer, beyond a year, or whatever the warranty states. If a large number are failing too soon, they still might be able to get some money from the manufacturer to replace the failed luminaries.

[Gyro]

Will the traffic lights be turning blue tomorrow?

No, they use dedicated colour LEDs.

[james_s]

I have a hypothesis about the ones turning purple or blue. I think that there's those are using tri-color non-phosphor LEDs. These are the same type that are used in RGB LED advertising signs that can display bright full color pictures on billboards. Except here they are being used not for image display, but rather for illumination. My guess is that the drivers for green LED chips and red LED chips are failing faster than the ones powering the blue LEDs, resulting in the blue color you see from them when they fail. While most LED streetlights use the normal white phosphor design, I think a few are using tri-color LEDs and these are the ones that turn blue when they fail.

Your hypothesis is wrong. I have worked as a consultant in the street lighting industry, a close friend of mine has spent his career in it. I've seen many different examples of LED street lighting luminaires up close and absolutely every single one of them uses phosphor based white LEDs in surface mount packages. The lights that are turning blue are having the phosphor flake off, the same has happened with some LED backlit TVs.

[Ground_Loop]

I just drove into Atlanta airport and saw literally dozens of the street lights that had gone purple. Sorry no pics, but it looked to be around 25% had the problem.

[Stray Electron]

  They put a new street light in front of my house about a year ago and it was a brilliant, glaring white! About two weeks ago I came home and it had suddenly changed to pinkish purple color and wasn't nearly as bright. This week they came out and repaired or replaced it (I wasn't here and didn't see what they did).  IMO it didn't last very long! The old mercury vapor lights would run for years!

  I see LED signs and stop lights all over town that have large numbers of failed LEDs in them and I'm also noticing a lot of very new cars and trucks with LED lights that have failed.

  IMO the poor quality of the mass produced LED lights vastly exceeds any theoretical improvements in longevity or energy efficiency.

[SeanB]

Down the road from me is a street with LED lamps that were fitted around 2008, and, due to the usual competence of the council electrician contractors, they all have shorting caps, and the group relay was disconnected, and each was wired to the phase wiring of the overhead line. Now on 27/7/365 they are still sort of functional, as in they still emit some light at night, even though more than half the dies in the head are failed, and these obviously have failed led detection in the optical head, and a really good power supply. Lumen output is down by more than half, they used to be impossible to look at during the day, but now they are a lot dimmer.

The other ones in the street had to be replaced within a year, as they all failed, so the supplier, Beka, replaced them, and then the metro also put in a lot more additional fixtures to supplement the light, replacing the 400W HPS fixtures with an 80W fixture on the street side, and an 18W fixture on the pavement side.

Still a few 400W HPS and MH units left though, and the rear alley by me got a relamp with a new 80W MV fixture, as the ballast had rotted away. I kind of wanted that old Phillips lamp, but it fell to pieces when they touched it, so in went a Beka GRP fixture, complete with 80W Osram lamp. I did get the old shorting plug though, to use on my other 80W fixture instead of the photocontrol.

Typically the LED is between 80W and 135W, replacing a 400W fixture, and for residential 18W or 36W, you place either one lamp assembly in, or two, and set the power supply accordingly.

[james_s]

  They put a new street light in front of my house about a year ago and it was a brilliant, glaring white! About two weeks ago I came home and it had suddenly changed to pinkish purple color and wasn't nearly as bright. This week they came out and repaired or replaced it (I wasn't here and didn't see what they did).  IMO it didn't last very long! The old mercury vapor lights would run for years!

  I see LED signs and stop lights all over town that have large numbers of failed LEDs in them and I'm also noticing a lot of very new cars and trucks with LED lights that have failed.

  IMO the poor quality of the mass produced LED lights vastly exceeds any theoretical improvements in longevity or energy efficiency.

It's really hard to beat the old mercury lamps in terms of longevity, at least the good quality American and European lamps. My grandparents house had a 175W mercury yard light on the front of the workshop in the back yard and it had been running dusk till dawn for around 35 years when the house was sold, still on the original lamp and it still looked about as bright as a new one. It had one of the GE Bonus Line lamps they made for a few years that had electrodes formulated to make white deposits instead of black.

That old mercury light consumed about 200W though with ballast losses, and a modern LED replacement would probably be around 30-50 watts for roughly the same light output so the LEDs do probably ultimately save money, especially when you consider the mercury lamps made more recently are mostly cheap crap. The last producer of good quality mercury lamps is/was probably Iwasaki in Japan.

[CCitizenTO]

IIRC it is an American company's products that are doing this, I think he said it was Cooper Lighting but I'll check.

Just because it's an American Company doesn't mean they aren't using cheap Chinese parts. Remember the cheaper they are to produce something the more profit they make...

[james_s]

Just because it's an American Company doesn't mean they aren't using cheap Chinese parts. Remember the cheaper they are to produce something the more profit they make...

Sure but almost everyone is using Chinese parts. Some Chinese parts are quite good, what matters is the specifics.

[floobydust]

I think the supporting design is also important, as far as not driving the LED's so hard that they have a short lifetime. Unfortunately this is commonplace, just to get a bigger lumens number and be cost-competitive. Then, the city's engineers use the exaggerated light numbers and design to their own cheapness. My city's LED streetlighting has so many dark areas where some CivE thought the cheap cobra head would cover many 100m, space them faaaaar apart, it's kind of pathetic. Easy to bust into cars in total darkness. If I were a woman walking at night, I would be terrified it's just creepy.

There is also contract manufacturing of LED's, rather than use something in a product line, specify your own to get advantage. But numbers can get exaggerated twice (by the LED component manufacturer and the product designer) and the result is short lifetime, a failed product.

[TimFox]

Recently in Chicago, I have noticed a problem with another common light, which I believe now uses LEDs.
There is a standard square electrical sign for "no left turn", which has a red circle and diagonal bar over a white right angle icon for "left turn".
I now see many of these where the red circle and diagonal is clearly visible, but the white "left turn" icon is barely visible.

[AlbertL]

I saw my first purple streetlight recently, in my neighborhood in Fairfax County, VA (about 8 mi. west of Washington, DC).

Some people are making good use of them: https://www.tampabay.com/photos/2021/07/05/tampa-photographer-uses-damaged-streetlights-to-create-art/
and want to keep them: https://www.change.org/p/manitoba-hydro-keep-the-purple-led-street-lights-in-winnipeg .

[AlbertL]

My friend who works in the streetlighting industry was just talking about this the other day when I was hanging out with him. He said it is a particular brand and model of luminaire that is having this problem, he's been working on getting me a defective one to do a failure analysis on.

Here's a report citing a conversation with a manufacturer's representative who confirmed that the problem is failure of the phosphor coating: https://schaumburg.novusagenda.com/AgendaPublic/CoverSheet.aspx?ItemID=16260&MeetingID=2566 

And another mention: https://streets.mn/2022/03/21/spring-2022-highway-news-roundup/ (second item on page)

[Ground_Loop]

I just drove into Atlanta airport and saw literally dozens of the street lights that had gone purple. Sorry no pics, but it looked to be around 25% had the problem.

With pics

[james_s]

That's going to be disastrous for the manufacture. They do look pretty cool though.

[SiliconWizard]

Ahahah!

[PlainName]

If these lights are because of the phospor failing, isn't the UV being blasted out? Wouldn't there be a health hazard at least to your eyes?

[Zero999]

Will the traffic lights be turning blue tomorrow?

No, they use dedicated colour LEDs.

Not for red and unlikely for bluish-green, but phosphor converted LEDs are sometimes used for amber, because LEDs in that region of the spectrum are notoriously inefficient. I wouldn't know for sure whether these are used in traffic lights, but it wouldn't surprise me.
https://www.lighting.philips.com/main/support/support/faqs/white-light-and-colour/how-is-amber-produced
https://phys.org/news/2009-07-yellow-gap-full-conversion-blue.html

[WatchfulEye]

If these lights are because of the phospor failing, isn't the UV being blasted out? Wouldn't there be a health hazard at least to your eyes?

There should be zero UV emission as the pump LEDs are blue and, as is typical for LEDs, have a very narrow emission spectrum which would not extend to UV.

While UV LEDs exist, they are substantially more expensive. They seem a poor choice for illumination, as you would need to phosphor convert the blue light as well as red/green, reducing efficiency and complicating phosphor design.

Blue light alone is unlikely to have major health effects at the intensities seen here. There is some concern over disruption to circadian rhythms, and there is a current fad about blue light reduction some of which is questionable at best.

At high intensities, any visible light can cause eye injury due to direct phototoxicity (prolonged saturation of photoreceptors results in biochemical imbalance which is toxic and can lead to cell injury). This can occur if you stare directly at a high intensity LED especially if it has collimating optics (eclipse viewing with an inadequate filter is a well recognised cause, as is deliberate misuse of laser pointers), but this is unlikely to occur by accident.

[SiliconWizard]

Agreed. There's very little UV emitted by LEDs made for lighting.

Beyond the spectrum, the amount of external ligthing at night is a real environmental problem. And a potential health issue at least for animals that cannot escape it.
It's not very hard to figure that billions of years of evolution with basically no light at night (just moonlight) have shaped organisms in a way that a sudden change with constant night lighting above the brightness of moonlight will cause problems. But we don't care, as we must do what we want. Ha!

[AVGresponding]

Oh come on, it's not like night flying insects and animals do anything useful like pollinate food crops... oh wait...

[Zero999]

Agreed. There's very little UV emitted by LEDs made for lighting.

Beyond the spectrum, the amount of external ligthing at night is a real environmental problem. And a potential health issue at least for animals that cannot escape it.
It's not very hard to figure that billions of years of evolution with basically no light at night (just moonlight) have shaped organisms in a way that a sudden change with constant night lighting above the brightness of moonlight will cause problems. But we don't care, as we must do what we want. Ha!

Yes, low pressure sodium was less of a problem for people and wildlife alike. I believe there are some areas where they've switched to longer wavelength LEDs, where the blue/white ones have been shown to be problematic to wildlife i.e. nesting turtles.

[PlainName]

Oh come on, it's not like night flying insects and animals do anything useful like pollinate food crops... oh wait...

Don't flowers close at night?

[Zero999]

Oh come on, it's not like night flying insects and animals do anything useful like pollinate food crops... oh wait...

Don't flowers close at night?

Some flowers do. It depends.

[AlbertL]

That's going to be disastrous for the manufacture. They do look pretty cool though.

Yes, especially because in some jurisdictions the manufacturer is also on the hook for the cost of the labor to replace the fixture.

[james_s]

My friend who works in the streetlighting industry was just talking about this the other day when I was hanging out with him. He said it is a particular brand and model of luminaire that is having this problem, he's been working on getting me a defective one to do a failure analysis on.

Here's a report citing a conversation with a manufacturer's representative who confirmed that the problem is failure of the phosphor coating: https://schaumburg.novusagenda.com/AgendaPublic/CoverSheet.aspx?ItemID=16260&MeetingID=2566 

And another mention: https://streets.mn/2022/03/21/spring-2022-highway-news-roundup/ (second item on page)

Just saw this. Whoever wrote the document keeps using the word "phosphorus" instead of the correct and completely different term "phosphor". 

[Zero999]

My friend who works in the streetlighting industry was just talking about this the other day when I was hanging out with him. He said it is a particular brand and model of luminaire that is having this problem, he's been working on getting me a defective one to do a failure analysis on.

Here's a report citing a conversation with a manufacturer's representative who confirmed that the problem is failure of the phosphor coating: https://schaumburg.novusagenda.com/AgendaPublic/CoverSheet.aspx?ItemID=16260&MeetingID=2566 

And another mention: https://streets.mn/2022/03/21/spring-2022-highway-news-roundup/ (second item on page)

Just saw this. Whoever wrote the document keeps using the word "phosphorus" instead of the correct and completely different term "phosphor". 

I also noticed that, but couldn't be bothered to mention it here. It's quite common for people who are unfamiliar with LEDs and lighting.

[james_s]

I also noticed that, but couldn't be bothered to mention it here. It's quite common for people who are unfamiliar with LEDs and lighting.

It drives me nuts, I remember noticing it in a show about how fluorescent tubes are made too. It's such a rookie mistake that immediately tells me the person writing doesn't know what they're talking about. It's like silicone and silicon, similar sound, big difference.

[Sal Ammoniac]

The debacle of Detroit's streetlight failures- over 20,000 "prematurely dimming and burning out" cost almost $7M to replace them. They failed melting the lens.
Detroit Sues Lighting Manufacturer Over Failed Street Lights The City sued Leotek Electronics, got only $4M in settlement.

Point is it's very expensive labour-wise to swap out defective streetlights, about 1,500 a week were done and the City of Detroit contracted out all streetlight work, they didn't have their own crews.

Fun historical fact: Before the turn of the 20th century Detroit had 122 175-foot tall Moonlight Towers to light up 22 square miles of the city center. Austin TX still has some of these towers.

[PlainName]

Wow! It works really well in that photo! Almost as good as daylight.

[fourfathom]

Will the traffic lights be turning blue tomorrow?

No, they use dedicated colour LEDs.

Not for red and unlikely for bluish-green, but phosphor converted LEDs are sometimes used for amber, because LEDs in that region of the spectrum are notoriously inefficient. I wouldn't know for sure whether these are used in traffic lights, but it wouldn't surprise me.
https://www.lighting.philips.com/main/support/support/faqs/white-light-and-colour/how-is-amber-produced
https://phys.org/news/2009-07-yellow-gap-full-conversion-blue.html

C'mon guys!  That's a Jimi Hendrix quote:

The traffic lights, they turn blue tomorrow
And shine their emptiness down on my bed
The tiny island sags downstream
'Cause the life that lived is dead
And the wind screams, Mary

From The Wind Cries Mary

[Ben321]

The phosphors are inside of the LED component itself. There's the blue gallium nitride blue emitter chip itself, which is then coated with a layer of yellow fluorescent material. This entire structure then is fully encapsulated inside the dome shaped piece of clear plastic. And by encapsulated I mean that liquid plastic is poured over the chip+phosphor structure that's sitting in a dome shaped mold so that the chip+phosphor structure is completely potted. There's absolutely no way for it to flake off. Overheating could burn out LED or even explode it, but the phosphor can't flake off while the LED is operational because it's physically sealed by the surrounding plastic. And of course if it burns out or blows up, the LED stops emitting all light.

Have you ever seen an SMD LED? I have never seen any lighting solution with outdated through-hole LEDs.

Don't those SMD LEDs have a clear plastic layer over it though, so that it protects the inside of the LED? Also I've seen some cheap dollar-store LED flashlights with like 10 low power through-hole LEDs, rather than a single high power SMD LED.

[Ben321]

Almost all commercial white LEDs are blue LEDs with phosphors ontop.  These street lights sound like they're losing their phosphor layers.

Temperature x hours too much for them?  Or maybe something more creative like the phosphors flaking off due to moisture and cycling?

The phosphors are inside of the LED component itself. There's the blue gallium nitride blue emitter chip itself, which is then coated with a layer of yellow fluorescent material. This entire structure then is fully encapsulated inside the dome shaped piece of clear plastic. And by encapsulated I mean that liquid plastic is poured over the chip+phosphor structure that's sitting in a dome shaped mold so that the chip+phosphor structure is completely potted. There's absolutely no way for it to flake off. Overheating could burn out LED or even explode it, but the phosphor can't flake off while the LED is operational because it's physically sealed by the surrounding plastic. And of course if it burns out or blows up, the LED stops emitting all light.


I have a hypothesis about the ones turning purple or blue. I think that there's those are using tri-color non-phosphor LEDs. These are the same type that are used in RGB LED advertising signs that can display bright full color pictures on billboards. Except here they are being used not for image display, but rather for illumination. My guess is that the drivers for green LED chips and red LED chips are failing faster than the ones powering the blue LEDs, resulting in the blue color you see from them when they fail. While most LED streetlights use the normal white phosphor design, I think a few are using tri-color LEDs and these are the ones that turn blue when they fail.

Why would the green and red LEDs fail first? If they were RGB then a good number of them would turn yellow, or cyan, rather than purple.

It's phosphor failure. The reason why some change purple is because violet, rather than blue LEDs are often used to excite the phosphor.

Cobra head LED light fixtures fail for many reasons - overheating, water ingress, cheap/overstressed components like the LED's, electrolytic capacitors etc. and thermal cycling, bad driver design, lightning/transients, loose bolts falling down and many more reasons.

For the purple/blue streetlights, "American Electric Lighting (AEL) a division of Acuity Brands Lighting products" seems to be just a middleman, a dealer of numerous lighting brands.
Their warranty is the usual 1 year, liability limited to the purchase price of the product. A few warranties went to 6 years on select components.
Our municipality's tax dollars pay for these failures but it's always hush hush and swept under the rug because it's very expensive to replace streetlights, labour-wise.

For Detroit, LEOTEK is part of Lite-On Group.
So is their melting LED "cheap" or driven at the datasheet maximum? You want the highest lumens number in order to make the sale, who cares about lifetime- by then you will have bought your Lambo and can blame someone else.

Laws vary. In some countries the manufacturer still has a statuary obligation to the customer, beyond a year, or whatever the warranty states. If a large number are failing too soon, they still might be able to get some money from the manufacturer to replace the failed luminaries.

Regarding the LEDs turning color, it could be cheaply made LED driers. Typically for the 20mA through hole LEDs, red LEDs have a voltage drop of about 2 volts, while green LEDs have a slightly higher voltage drop (close to 3 volts), and blue have even higher (about 4 to 5 volts).  So if the LED drivers were using a simple resistor, and all of the drivers of different colors were using the same resistor value. The lower voltage drop of the red LEDs would cause them to draw more current through the resistor, causing them to fail first. Obviously these aren't low power 20mA LEDs being used, but the idea scales up to the higher power LEDs anyway, because the higher power red green and blue LEDs still use the same semiconductor materials (so now the voltages across the devices are higher, but by the same ratios). I could easily see a cheap Chinese company doing something like using simple resistors (not constant current LED drivers) and even using the same value of resistor for all 3 colors, just to cut costs, even if it shortens the usable life of the product. So I'm guessing that the LED streetlights you see failing were all made by some Chinese company who made them on the cheap, with shoddy design (specifically shoddy in the way I described above).

[Ben321]

I just drove into Atlanta airport and saw literally dozens of the street lights that had gone purple. Sorry no pics, but it looked to be around 25% had the problem.

With pics

(Attachment Link)
(Attachment Link)

Some look blue and others look purple. Is that just the result of how the camera captured the light, and they all actually had the same color (either blue or purple)? Or is that actually what they looked like (some being blue and others purple)? And I can't imagine all of them failing together such that an entire stretch of roadway had these defective LEDs. I also have never personally seen LED streetlights fail this way. The ones I've seen fail blink at a fixed rate like a strobe light (at about 2 to 4 flashes per second), but their color stays the same as when they were fully operational.

[PlainName]

all of them failing together such that an entire stretch of roadway had these defective LEDs

They will buy and build in  bulk, so if it's a batch problem it would likely affect adjacent lamps.

[james_s]

Regarding the LEDs turning color, it could be cheaply made LED driers. Typically for the 20mA through hole LEDs, red LEDs have a voltage drop of about 2 volts, while green LEDs have a slightly higher voltage drop (close to 3 volts), and blue have even higher (about 4 to 5 volts).  So if the LED drivers were using a simple resistor, and all of the drivers of different colors were using the same resistor value. The lower voltage drop of the red LEDs would cause them to draw more current through the resistor, causing them to fail first. Obviously these aren't low power 20mA LEDs being used, but the idea scales up to the higher power LEDs anyway, because the higher power red green and blue LEDs still use the same semiconductor materials (so now the voltages across the devices are higher, but by the same ratios). I could easily see a cheap Chinese company doing something like using simple resistors (not constant current LED drivers) and even using the same value of resistor for all 3 colors, just to cut costs, even if it shortens the usable life of the product. So I'm guessing that the LED streetlights you see failing were all made by some Chinese company who made them on the cheap, with shoddy design (specifically shoddy in the way I described above).

Stop guessing, we already know the answers. The failing lights were made by American Electric, it's a reputable brand. The drivers are not using a simple resistor and they're not using separate RGB LEDs, and they're not using through-hole LEDs, they're surface mount white power LEDs with a proper electronic driver and the phosphor layer is flaking off. This is already known, there is no need to speculate or make things up.

[MrMobodies]

Joke: Does that mean that they will illuminate thieves that have been sprinkled with smart water.

[Haenk]

https://www.businessinsider.com/led-city-streetlights-turning-purple-broken-tech-danger-2022-11?utm_source=digg

purple lights made it into mainstream media...

[Ranayna]

LED streetlights are weird...
I don't mind them as much in a car, at least after they have lost a bit of their initial brightness. When new they can be annoying point sources
My town was switches from CCFL tubes to LED during the last two years. The older ones are noticeably dimmer already. Is this degradation linear? If so, i wonder how long they will last.

But as a pedestrian, especially if there are tree branches under the light, the shadows the LEDs cast are, well, weird. All shadows are multiplied, since each LED throws it's own shadow, and if the branches are moving in the wind, the effect on the ground can be almost nauseating.

[AlbertL]

The manufacturer's web site offers this timely warning: https://www.acuitybrands.com/resources/trending-topics/low-quality-products

[james_s]

LED streetlights are weird...
I don't mind them as much in a car, at least after they have lost a bit of their initial brightness. When new they can be annoying point sources
My town was switches from CCFL tubes to LED during the last two years. The older ones are noticeably dimmer already. Is this degradation linear? If so, i wonder how long they will last.

But as a pedestrian, especially if there are tree branches under the light, the shadows the LEDs cast are, well, weird. All shadows are multiplied, since each LED throws it's own shadow, and if the branches are moving in the wind, the effect on the ground can be almost nauseating.

I'm not sure what the depreciation curve for LEDs looks like but I bet the manufactures have it available. I know that with fluorescent and metal halide lamps they depreciate pretty rapidly during the first 100 hours and then settle into a long slow decline. When they hit 70% of rated lumens they are considered end of life. IIRC rated lumens is the value after the first 100 hours.

I've found that LED street lights vary widely, some of them are almost point sources and have a lot of glare, others are much nicer. I like the white light much better than orange HPS but I don't like it as well as mercury vapor.

[AlbertL]

I saw one in July of last year in my area of Fairfax County, Virginia, and until recently hadn't seen any more.  But in the past month or so, I've seen a few.  Perhaps my electric utility bought most of their lights late in the run of bad units, so the failures are just beginning, or maybe the bad batch comprises a small percentage of their fleet.  I'm certainly not seeing the clusters of failures like some other parts of the country are.

[AlbertL]

Apparently there's a market for the purple lights: https://acuitybrands.my.site.com/s/article/Purchasing-Fixtures-with-Blue-or-Purple-Color-Temperature

[David Hess]

This past winter I stopped at a rest stop in Pennsylvania that had the failing LED street lights.  I saw a bunch of them in Kansas also.

[SiliconWizard]

What's the cost of replacing and recycling all these failed lights compared to having kept the original equipment even if it drew maybe a bit more power? A bit curious.

[PlainName]

Should be $0 since they are faulty and should be repaired under warranty. And a few doing this is small compared to the many that are Just Working.

[Zero999]

To be pedantic, they're not purple, but spectral violet.

I don't see how it would draw more power. The problem is the eyes are less sensitive to violet, which means they're not bright enough and pose a road safey hazard.

It's definitely a warranty claim. I wonder if the manufacture could send a tech to clean the boards and reapply the phosphor? That might be cheaper than replacing them all.

[AVGresponding]

I think he's referring to the old discharge lamps drawing more power, but being much more reliable and maintainable.