PhotonWrangler
Flashaholic
Ok, I see an MOV (RV1) next to the input inductor. I also see a big ol' transistor (Q5) in the upper-left area that looks a little, ah, stressed.
Led Traffic Lights Dying Already.
- Thread starter 3rd_shift
- Start date
Help Support Candle Power Flashlight Forum
gadget_lover
Flashaholic
[ QUOTE ]
elgarak said:
Well, I referred to the error mode I observe. I see the following: An array of 5(let's say) 0 LEDs. 25 on the left side do not give off any light (when observed during daylight), the right half is very bright. Ageing the LEDs with overheating would not result in this error. One should see varying brightness levels over the array. If one side of the array is hotter than the other side, one should see gradually changing brightness from one side to the other. Not the sharp edge of LEDs on and off I observe.
[/ QUOTE ]
While I understand your assertion, there are many failure modes that can occur from temperature;
A set of LED's in series will all go out when one overheats and goes open.
Frequent change of temperature can cause the boards to expand and contract, breaking traces and causing bad contact at connectors.
Overheating may also cause the driver / power supply electronics to fail in interesting ways.
Daniel
elgarak said:
Well, I referred to the error mode I observe. I see the following: An array of 5(let's say) 0 LEDs. 25 on the left side do not give off any light (when observed during daylight), the right half is very bright. Ageing the LEDs with overheating would not result in this error. One should see varying brightness levels over the array. If one side of the array is hotter than the other side, one should see gradually changing brightness from one side to the other. Not the sharp edge of LEDs on and off I observe.
[/ QUOTE ]
While I understand your assertion, there are many failure modes that can occur from temperature;
A set of LED's in series will all go out when one overheats and goes open.
Frequent change of temperature can cause the boards to expand and contract, breaking traces and causing bad contact at connectors.
Overheating may also cause the driver / power supply electronics to fail in interesting ways.
Daniel
PlayboyJoeShmoe
Flashaholic
I notice even more after this thread.
The Red signals around here are one of two types. Straight runs of leds, or a sunburst array.
I haven't seen any complete lines out on the straight runs, but individual leds (almost always around the outside edges) are out. Maybe two in a row sometimes. Same goes for the sunburst.
I actually don't notice the greens as much, since I'm boogying when I get one!
And a lot of intersections have yellow leds too.
The Red signals around here are one of two types. Straight runs of leds, or a sunburst array.
I haven't seen any complete lines out on the straight runs, but individual leds (almost always around the outside edges) are out. Maybe two in a row sometimes. Same goes for the sunburst.
I actually don't notice the greens as much, since I'm boogying when I get one!
And a lot of intersections have yellow leds too.
ViReN
Flashlight Enthusiast
3rd_shift .. The Electronics is THAT BAD ????
Oh Please.... Even the Cheap Chinese Telephones look much better (in electronic components & Layouts).... and they sure can handle more shocks & humidity.. than that bord picture....
To Me, It looks like, Its some where in old 70's PCB Board.. where they used to have that kind of placement.. (hand placed components???)...
Oh Look at the ceramic capacators... & the way resistors are placed !!! ....
It sure does not make any difference in the functionality.. but sure could impact the performance & Reliablity.....
If that is how (all) Traffic Signals have supplies... then there is NO USE of AIRBAGS & Seat belts... after all (if a light fails to show RED) who is gonna stop ??? Accidents are inevitable !!!....
I am really sorry to see that board.. wish i had never seen it... EVER !
3rd_shift , Just take a look at some good medical electronic equipment.... you will know what i am talking about...
If medical equipment is a matter of life and death.. WHY NOT TRAFFIC SIGNALS ???? Why such pathetic boards... ?????
When there is LOT of space.. is there any POINT in putting STANDING RESISTORS ????
WHATS THE USE OF DOUBLE SIDED PCB ????? what.. if by shock... 2 resistors get short... (which may drive the Transistors that DRIVE LED's !!!!!!!!!) and WHAT IF THAT LIGHT IS A GREEN ONE ? ... 2 GREEN LIGHTS... 2 FAST CARS... both see green... and THATS ALL IT TAKES DUDE !!!...
Please forgive me for using Caps... but.... i cant keep cool man... not at all.... ( i aint angry on any of CPF guys... please let this be known clearly... but angry... on... the way it has been designed)
I have an OLD u.p.s. (10 years back) ... and they have used floor resistors... NONE of em is standing... (though home computer u.p.s. is not LIFE Critical Application nor it is going to handle shocks(physical shocks .. wether / storm related)) ... I also understand that the resistor leads might be strong.. but they are NOT THAT strong... seen Bouncing Signals.. when there is a thunder storm... ???
Oh Let Quickbeam & C.J. Do some testing on these Lights too... I am sure .. they wont get any more than Z E R O Stars !!!
ViReN
Oh Please.... Even the Cheap Chinese Telephones look much better (in electronic components & Layouts).... and they sure can handle more shocks & humidity.. than that bord picture....
To Me, It looks like, Its some where in old 70's PCB Board.. where they used to have that kind of placement.. (hand placed components???)...
Oh Look at the ceramic capacators... & the way resistors are placed !!! ....
It sure does not make any difference in the functionality.. but sure could impact the performance & Reliablity.....
If that is how (all) Traffic Signals have supplies... then there is NO USE of AIRBAGS & Seat belts... after all (if a light fails to show RED) who is gonna stop ??? Accidents are inevitable !!!....
I am really sorry to see that board.. wish i had never seen it... EVER !
3rd_shift , Just take a look at some good medical electronic equipment.... you will know what i am talking about...
If medical equipment is a matter of life and death.. WHY NOT TRAFFIC SIGNALS ???? Why such pathetic boards... ?????
When there is LOT of space.. is there any POINT in putting STANDING RESISTORS ????
WHATS THE USE OF DOUBLE SIDED PCB ????? what.. if by shock... 2 resistors get short... (which may drive the Transistors that DRIVE LED's !!!!!!!!!) and WHAT IF THAT LIGHT IS A GREEN ONE ? ... 2 GREEN LIGHTS... 2 FAST CARS... both see green... and THATS ALL IT TAKES DUDE !!!...
Please forgive me for using Caps... but.... i cant keep cool man... not at all.... ( i aint angry on any of CPF guys... please let this be known clearly... but angry... on... the way it has been designed)
I have an OLD u.p.s. (10 years back) ... and they have used floor resistors... NONE of em is standing... (though home computer u.p.s. is not LIFE Critical Application nor it is going to handle shocks(physical shocks .. wether / storm related)) ... I also understand that the resistor leads might be strong.. but they are NOT THAT strong... seen Bouncing Signals.. when there is a thunder storm... ???
Oh Let Quickbeam & C.J. Do some testing on these Lights too... I am sure .. they wont get any more than Z E R O Stars !!!
ViReN
HarryN
Flashlight Enthusiast
There are a number of LED manufacturer's that produce LEDs with a wide range of lifetimes. The lifeimes we are used to looking at for monochromatic Luxeons like RED / Green (nominally 100K hours to 80 - 90% initial brightness) are quite deceivingly applied to 5mm LEDs. AFAIK, only really premium brand and quality 5mm LEDs come even close to 50 % of this performance. Most of the curves I have seen are more like 50% brightness after 5 - 10K hrs.
Especially with 5mm red LED die prices in the US $ 0.005 - 0.02 range, pressure to cut corners in product development and production and is quite high. Reliability is a common victim.
Another commonly applied cost cutting method is to substantially over drive the LEDs in the fixture to meet the required brightness spec, then blame the LED maker when product lifeime is less than specified. (this is commonly discussed at LED conferences)
Last but not least, humidity also degrades LEDs, along with the more commonly discussed heat. There is a famous case study of LED traffic light installations in Singapore, land of heat and humidity (similar to Houston). As it turned out, none of the suppliers that actually knew what they were doing would bid on the project because the lifetime specs could not be achieved. Guess who got the order ? Guess how long the LED traffic lights lasted ?
In California, the most used signal light is RED, which is of course key to our ultra efficient traffic control system. /ubbthreads/images/graemlins/icon15.gif Power consumption was the main reason for converting these to LED first.
Amber is hardly ever on, and our traffic control friends carefully time this signal to be on a random amount of time so that the traffic cameras can issue tickets in the most efficient manner possible. LED use in this light is not necessarily cost effective, but potentially reduces labor to change the bulb.
Especially with 5mm red LED die prices in the US $ 0.005 - 0.02 range, pressure to cut corners in product development and production and is quite high. Reliability is a common victim.
Another commonly applied cost cutting method is to substantially over drive the LEDs in the fixture to meet the required brightness spec, then blame the LED maker when product lifeime is less than specified. (this is commonly discussed at LED conferences)
Last but not least, humidity also degrades LEDs, along with the more commonly discussed heat. There is a famous case study of LED traffic light installations in Singapore, land of heat and humidity (similar to Houston). As it turned out, none of the suppliers that actually knew what they were doing would bid on the project because the lifetime specs could not be achieved. Guess who got the order ? Guess how long the LED traffic lights lasted ?
In California, the most used signal light is RED, which is of course key to our ultra efficient traffic control system. /ubbthreads/images/graemlins/icon15.gif Power consumption was the main reason for converting these to LED first.
Amber is hardly ever on, and our traffic control friends carefully time this signal to be on a random amount of time so that the traffic cameras can issue tickets in the most efficient manner possible. LED use in this light is not necessarily cost effective, but potentially reduces labor to change the bulb.
jtr1962
Flashaholic
[ QUOTE ]
HarryN said:
AFAIK, only really premium brand and quality 5mm LEDs come even close to 50 % of this performance. Most of the curves I have seen are more like 50% brightness after 5 - 10K hrs.
[/ QUOTE ]
Are you 100% sure of this? I know that 5mm whites driven to spec last about 5,000 to 10,000 hours, and smaller wavelength colored LEDs (UV, blue, cyan) also last less than 100,000 hours (but much more than 10,000 except for UV). Red, orange, amber, and green should last at least 100,000 hours. The same tests which were performed on 5mm whites also used 5mm reds as a control. They even stated that it had been well established that reds lasted at least 100,000 hours.
I'm wondering exactly how cutting corners can affect lifetime. The usual degradation mechanisms are blackening of the epoxy (not the main factor in long wavelength LEDs), phosphor degradation (again not applicable except in white), and degradation of the die itself. Since many types of semiconductors are mass produced at low costs (i.e. diodes, transistors), and these all last decades if driven to spec, I don't see how low-cost mass production can affect die lifetimes. This leaves us with only two other failure mechanisms-substandard wire bonds and humidity entering the package due to substandard encapsulation. I'm not counting heat since I said the LEDs were driven to spec. Both of these will certainly shorten the life on average, but not through gradual dimming. Rather, we would have catastrophic failure of a greater than usual number within the first few hundred hours of life (termed infant mortality). The remainder that survive will last 100,000 hours because even cutting corners with packaging, you will need a majority of good units to make production worthwhile. Or put another way, if you cut corners too much, so many of your LEDs will fail in the first few hundred hours than you will put yourself out of business.
The only difference between low cost and higher cost LEDs will be a higher than normal infant mortality rate. Provided you "burn-in" finished products for a few hundred hours, then you should catch most of the failures before they go out in the field. Evidently, some traffic signal manufacturers are using cheap LEDs and not doing this. Nothing necessarily wrong with cheap LEDs, BTW, provided you accept a higher than usual infant mortality in return for paying less. I'm perfectly confident the cheap white LEDs I buy from HK for $0.25 each will last as long as $1.50 Nichias. I'll just get a few more failures in the first few hundred hours. Truth be told, given the infant mortality rate of most types of electronics, I would probably have to use tens of thousands of LEDs to notice one infant mortality failure even with the cheaper LEDs. Of course, when you're dealing with thousands of traffic lights each with 100+ LEDs you're going to have some failures.
HarryN said:
AFAIK, only really premium brand and quality 5mm LEDs come even close to 50 % of this performance. Most of the curves I have seen are more like 50% brightness after 5 - 10K hrs.
[/ QUOTE ]
Are you 100% sure of this? I know that 5mm whites driven to spec last about 5,000 to 10,000 hours, and smaller wavelength colored LEDs (UV, blue, cyan) also last less than 100,000 hours (but much more than 10,000 except for UV). Red, orange, amber, and green should last at least 100,000 hours. The same tests which were performed on 5mm whites also used 5mm reds as a control. They even stated that it had been well established that reds lasted at least 100,000 hours.
I'm wondering exactly how cutting corners can affect lifetime. The usual degradation mechanisms are blackening of the epoxy (not the main factor in long wavelength LEDs), phosphor degradation (again not applicable except in white), and degradation of the die itself. Since many types of semiconductors are mass produced at low costs (i.e. diodes, transistors), and these all last decades if driven to spec, I don't see how low-cost mass production can affect die lifetimes. This leaves us with only two other failure mechanisms-substandard wire bonds and humidity entering the package due to substandard encapsulation. I'm not counting heat since I said the LEDs were driven to spec. Both of these will certainly shorten the life on average, but not through gradual dimming. Rather, we would have catastrophic failure of a greater than usual number within the first few hundred hours of life (termed infant mortality). The remainder that survive will last 100,000 hours because even cutting corners with packaging, you will need a majority of good units to make production worthwhile. Or put another way, if you cut corners too much, so many of your LEDs will fail in the first few hundred hours than you will put yourself out of business.
The only difference between low cost and higher cost LEDs will be a higher than normal infant mortality rate. Provided you "burn-in" finished products for a few hundred hours, then you should catch most of the failures before they go out in the field. Evidently, some traffic signal manufacturers are using cheap LEDs and not doing this. Nothing necessarily wrong with cheap LEDs, BTW, provided you accept a higher than usual infant mortality in return for paying less. I'm perfectly confident the cheap white LEDs I buy from HK for $0.25 each will last as long as $1.50 Nichias. I'll just get a few more failures in the first few hundred hours. Truth be told, given the infant mortality rate of most types of electronics, I would probably have to use tens of thousands of LEDs to notice one infant mortality failure even with the cheaper LEDs. Of course, when you're dealing with thousands of traffic lights each with 100+ LEDs you're going to have some failures.
jtr1962
Flashaholic
I also wish to add that the control board pictured earlier leaves a lot to be desired. Those filter caps are too small to do any good, and it looks like the transistors just turn on strings of LEDs. What regulates the current-a simple resistor? /ubbthreads/images/graemlins/frown.gif If I had to hazard a guess, I'd say many of the failures are not the LEDs themselves, but something on the circuit board causing a string to go out.
For what it's worth, so far I've only seen a few bad LEDs on NYC's crosswalk signs, and these seem to be most the orange-red ones rather than the white ones. I haven't seen any traffic signals with noticable dark areas yet. Maybe these things really will last 25 years if designed correctly. I hope so, because I think traffic signals are a testbed for the next thing to be replaced by LEDs once efficiency exceeds that of sodium vapor lamps-street lights. I certainly look forward to the day we have nice white LED street lighting in place of those hard to see under sodium vapor lights.
For what it's worth, so far I've only seen a few bad LEDs on NYC's crosswalk signs, and these seem to be most the orange-red ones rather than the white ones. I haven't seen any traffic signals with noticable dark areas yet. Maybe these things really will last 25 years if designed correctly. I hope so, because I think traffic signals are a testbed for the next thing to be replaced by LEDs once efficiency exceeds that of sodium vapor lamps-street lights. I certainly look forward to the day we have nice white LED street lighting in place of those hard to see under sodium vapor lights.
HarryN
Flashlight Enthusiast
There are some fundamental differences between a common Si transistor based device, and an LED. For one, a Si device is built in nearly a perfect material - single crystal Si. The equivalent material does not exist for LEDs. This very low cost Si material, with exceptional crystal properties, is at the heart of high reliability devices.
For seconds, Si devices have been around a long time, the third tier, lower reliability players are all out of business. Also, early on, IBM and the car companies really pushed Si device reliability very hard.
An LED is grown in thin, much less than perfect crystaline layers. ONE, of the many factors that affect LED lifetime is just how well this part of the process is performed. The IP related to this is a closely held secret, and only companies willing to invest in serious R and D programs have done the groundwork to have this IP. There are still quite a few fly by night operators in the LED business.
You can see the effect of this IP investment if you follow a Lumileds, Cree, or Nichia data sheet for 12 - 18 months. The "same" LED gets brighter, specs tighten, Vf decreases, etc.
I would have to do some digging to find the paper, but Dr. Heng Liu, formerly of AXT, presented at a conference approx 18 months ago on this very subject, comparing blue, 5mm LEDs driven at spec conditions, and comparing "output" vs "time". There is a pretty dramatic difference between the shape of the curves from the tier 1 vs other suppliers.
Part of the lifetime spec question is based on when you call a "end of life". I believe that Lumileds specs EOL when they reach approx 90% of initial brightness. Many others use 50 % of initial brightness.
AFAIK, LEDs usually either work or not, and are not really subject to a lot of burn in testing.
For seconds, Si devices have been around a long time, the third tier, lower reliability players are all out of business. Also, early on, IBM and the car companies really pushed Si device reliability very hard.
An LED is grown in thin, much less than perfect crystaline layers. ONE, of the many factors that affect LED lifetime is just how well this part of the process is performed. The IP related to this is a closely held secret, and only companies willing to invest in serious R and D programs have done the groundwork to have this IP. There are still quite a few fly by night operators in the LED business.
You can see the effect of this IP investment if you follow a Lumileds, Cree, or Nichia data sheet for 12 - 18 months. The "same" LED gets brighter, specs tighten, Vf decreases, etc.
I would have to do some digging to find the paper, but Dr. Heng Liu, formerly of AXT, presented at a conference approx 18 months ago on this very subject, comparing blue, 5mm LEDs driven at spec conditions, and comparing "output" vs "time". There is a pretty dramatic difference between the shape of the curves from the tier 1 vs other suppliers.
Part of the lifetime spec question is based on when you call a "end of life". I believe that Lumileds specs EOL when they reach approx 90% of initial brightness. Many others use 50 % of initial brightness.
AFAIK, LEDs usually either work or not, and are not really subject to a lot of burn in testing.
NewBie
*Retired*
70% for LumiLEDs Luxeons.
DS25 (1W datasheet)
Lumileds projects that Luxeon products will deliver on average 70%lumen maintenance at 50,000 hours of operation.This performance is based on independent test data, Lumileds historical data from tests run on similar material systems,and internal Luxeon reliability testing.This projection is based on constant current 350 mA operation with junction
temperature maintained at or below 90 °C.
http://www.lumileds.com/pdfs/protected/DS25.PDF
DS45 Datasheet (Luxeon III)
Lumileds projects that Luxeon III products will deliver, on average,70%lumen maintenance at 50,000 hours of operation at a 700mA forward current or
50% lumen maintenance at 20,000 hours of operation at a 1000 mA forward current.
This performance is based on independent test data, Lumileds historical data from tests run on similar material systems,and internal Luxeon reliability testing.This projection is based on constant current operation with junction temperature maintained at or below 90 °C.
http://www.lumileds.com/pdfs/DS45.PDF
DS40 (Luxeon V)
...and lifetimes greater than 50X conventional flashlight
sources over 500 hours compared to the typical 20 hour life for an incandescent bulb.
Heatsink Temperature 35C Current 700mA Average Lumen maintainance after 500 hours - 90%
Heatsink Temperature 75C Current 700mA Average Lumen maintainance after 500 hours - 75%
Heatsink Temperature 85C Current 600mA Average Lumen maintainance after 500 hours - 65%
http://www.lumileds.com/pdfs/protected/DS40.PDF
<font color="red"> (pay close attention to the most recent change in the Lux V datasheet here) </font>
This is a great read for anyone wishing to understand the issues involved and how they affect the Luxeon:
http://www.lumileds.com/pdfs/RD25.PDF
DS25 (1W datasheet)
Lumileds projects that Luxeon products will deliver on average 70%lumen maintenance at 50,000 hours of operation.This performance is based on independent test data, Lumileds historical data from tests run on similar material systems,and internal Luxeon reliability testing.This projection is based on constant current 350 mA operation with junction
temperature maintained at or below 90 °C.
http://www.lumileds.com/pdfs/protected/DS25.PDF
DS45 Datasheet (Luxeon III)
Lumileds projects that Luxeon III products will deliver, on average,70%lumen maintenance at 50,000 hours of operation at a 700mA forward current or
50% lumen maintenance at 20,000 hours of operation at a 1000 mA forward current.
This performance is based on independent test data, Lumileds historical data from tests run on similar material systems,and internal Luxeon reliability testing.This projection is based on constant current operation with junction temperature maintained at or below 90 °C.
http://www.lumileds.com/pdfs/DS45.PDF
DS40 (Luxeon V)
...and lifetimes greater than 50X conventional flashlight
sources over 500 hours compared to the typical 20 hour life for an incandescent bulb.
Heatsink Temperature 35C Current 700mA Average Lumen maintainance after 500 hours - 90%
Heatsink Temperature 75C Current 700mA Average Lumen maintainance after 500 hours - 75%
Heatsink Temperature 85C Current 600mA Average Lumen maintainance after 500 hours - 65%
http://www.lumileds.com/pdfs/protected/DS40.PDF
<font color="red"> (pay close attention to the most recent change in the Lux V datasheet here) </font>
This is a great read for anyone wishing to understand the issues involved and how they affect the Luxeon:
http://www.lumileds.com/pdfs/RD25.PDF
gadget_lover
Flashaholic
What HarryN said makes a lot of sense. Not only does it match my scketchy memory of semiconductor construction but I have visited with HarryN and believe he has the credentials and experience to back up his statements.
Daniel
Daniel
jtr1962
Flashaholic
I don't doubt HarryN's credentials. I'm just saying that this is the first I've heard of this difference in lifetime between different LED suppliers. And how does one know which chips end up in the LEDs you buy anyhow? For all I know the cheap HK LEDs may be using chips from Cree or Nichia. I'm planning on doing some long term testing of the various colors I have. I already know UV dims by quite a bit even after a few hundred hours, but this is a problem common to all UV LEDs as far as I know, or else they would be used in white LEDs instead of blue. Anyway, I'm finding this all very disturbing, especially if the LEDs in traffic lights don't last the claimed 100,000 hours.
gadget_lover
Flashaholic
I see your point, jtr1962. There is no way to know if you are getting top quality products in ANY category these days. With so much outsourcing of manufacturing, only a purchase directly from the manufacturer will provide any degree of assurance. Another thread quoted buying 1000mah battery packs only to find that they had 600mah cells inside.
I'm not sure why people worry about traffic signals as something extra critical. Ours don't have emergency power, and we live right through it. The traffic signals in the US are failsafed in a way. If the red does not light up, you are still supposed to stop because you don't have a green light.
I thought that blue was used to complement the yellow phosphor. I culd be wrong. They might have chosen a yellow phosphor to suplement the blue.
Daniel
I'm not sure why people worry about traffic signals as something extra critical. Ours don't have emergency power, and we live right through it. The traffic signals in the US are failsafed in a way. If the red does not light up, you are still supposed to stop because you don't have a green light.
I thought that blue was used to complement the yellow phosphor. I culd be wrong. They might have chosen a yellow phosphor to suplement the blue.
Daniel
jtr1962
Flashaholic
Actually the yellow phosphor is to complement the blue LED (combined they appear "white" to us even though it's not a perfect blackbody white).
HarryN
Flashlight Enthusiast
[ QUOTE ]
jtr1962 said:
I don't doubt HarryN's credentials. I'm just saying that this is the first I've heard of this difference in lifetime between different LED suppliers. And how does one know which chips end up in the LEDs you buy anyhow? For all I know the cheap HK LEDs may be using chips from Cree or Nichia. I'm planning on doing some long term testing of the various colors I have. I already know UV dims by quite a bit even after a few hundred hours, but this is a problem common to all UV LEDs as far as I know, or else they would be used in white LEDs instead of blue. Anyway, I'm finding this all very disturbing, especially if the LEDs in traffic lights don't last the claimed 100,000 hours.
[/ QUOTE ]
Hi jtr. Actually, my career has shifted over time away from my chem eng roots to being more qualitative than in the past - now I just know enough to be dangerous. The great thing about having Newbie in the forums is he is always there to fix our errors with real details.
/ubbthreads/images/graemlins/smile.gif He seems to know this stuff in and out.
Gadget is right, you are going to need to really understand your source all the way back to the die provider if you want to really be in control of your results. There are a few firms that are "buyer / packagers", that have pretty tight specs on their supply chain - Agilent is one of them, and there are a few others. (yes, believe it or not, Agilent typically buys die that meet their specs and has them packaged- usually not Lumileds die, even though they own 50 % of the firm.)
PM me if you would like to reach some people who have done similar testing.
jtr1962 said:
I don't doubt HarryN's credentials. I'm just saying that this is the first I've heard of this difference in lifetime between different LED suppliers. And how does one know which chips end up in the LEDs you buy anyhow? For all I know the cheap HK LEDs may be using chips from Cree or Nichia. I'm planning on doing some long term testing of the various colors I have. I already know UV dims by quite a bit even after a few hundred hours, but this is a problem common to all UV LEDs as far as I know, or else they would be used in white LEDs instead of blue. Anyway, I'm finding this all very disturbing, especially if the LEDs in traffic lights don't last the claimed 100,000 hours.
[/ QUOTE ]
Hi jtr. Actually, my career has shifted over time away from my chem eng roots to being more qualitative than in the past - now I just know enough to be dangerous. The great thing about having Newbie in the forums is he is always there to fix our errors with real details.
/ubbthreads/images/graemlins/smile.gif He seems to know this stuff in and out.
Gadget is right, you are going to need to really understand your source all the way back to the die provider if you want to really be in control of your results. There are a few firms that are "buyer / packagers", that have pretty tight specs on their supply chain - Agilent is one of them, and there are a few others. (yes, believe it or not, Agilent typically buys die that meet their specs and has them packaged- usually not Lumileds die, even though they own 50 % of the firm.)
PM me if you would like to reach some people who have done similar testing.
NewBie
*Retired*
[ QUOTE ]
jtr1962 said:
I don't doubt HarryN's credentials. I'm just saying that this is the first I've heard of this difference in lifetime between different LED suppliers. And how does one know which chips end up in the LEDs you buy anyhow? For all I know the cheap HK LEDs may be using chips from Cree or Nichia. I'm planning on doing some long term testing of the various colors I have. I already know UV dims by quite a bit even after a few hundred hours, but this is a problem common to all UV LEDs as far as I know, or else they would be used in white LEDs instead of blue. Anyway, I'm finding this all very disturbing, especially if the LEDs in traffic lights don't last the claimed 100,000 hours.
[/ QUOTE ]
Dunno who HK companies are utilizing. Some Asian LED makers are in fact using CREE die, as well as Toyoda Gosei, the CREE die are distributed and rep'd by Sumitomo, the last purchase was for 160,000,000.00 dollars of LED die (the tiny chip inside the LED). There are quite a number of island based die makers too though....
In fact, Toyoda Gosei has even made die for LumiLEDs.
Some of the big factors in LED life in recent years will be in the packaging. Die attach adhesive, and how well it resists light (especially UV) (some are actually soldering down the die themselves, where Luxeons have a die that is floating on I think it was 16 solder balls that rest on a ESD or Zener diode, dunno if they solder their zener to the heatsink though, unlike others who do), in whites, the life of the phosphor (phosphors also have a lifetime), the filler material often used to make the body from (as in 5mm)- as most by know know the epoxy yellows with time/temperature/light exposure, some fill with versions of non-yellowing silicon (for example, GE's X-14 LED silicone) (such as newer Nichia's, CREE XLamps, LumiLEDs Luxeon, and OSRAM's Dragon) some have an acrylic dome (seems those'd yellow too with time) to cover the silicon gel, others use no dome, or even a glass lens. Additionally, you'll find some even in a metal can with a glass lens, and Microsemi was pursuing a metal leaded, and all glass package (no goop, gunk, plastic).
I've querried Toyoda Gosei (who makes a LED with a UV die and RGB phosphor, which doesn't rapidly degrade like everyone said it would), Nichia, and CREE, none of them have ever heard of the UV attacking the DIE itself, unsure where that rumor came from.
In fact, you'll find 20,000 hour UV LEDs here:
20,000 hour UV LEDs
Oh, there is a huge difference between LED suppliers, as they utilize different packaging, and I have some white LEDs from an asian source that keeled over an barely glow anymore, even ran within their specs. The quality of the die itself, can't tell you much about that.
The reason blue is used instead of UV is the YAG phosphor is matched for the blue. The folks like Toyoda Gosei, who have been making a UV based White LED utilize a special version of RGB phosphor. The color rendering is much higher than the blue die-yellow phophor, and pictures taken where those are utilized as a light source look *much* more natural.
You can read more here:
http://www.microlamps-tg.com/
http://www.microlamps-tg.com/html/Prod_white_compare.htm
Datasheet:
True White Hi
Catalog:
http://www.microlamps-tg.com/html/Cat04_01.htm
As far as the 100,000 hours, alot of it has to do with the skill and experience of the companies that designed the traffic light. This is one area where buying from the cheapest bidder can result some very unhappy campers. It's these companies that will give LEDs a bad rap. Just like the millions of CFL bulbs made by GE and Sylvannia that were only lasting a few months for many folks (yeah, other brands had the issue, but those two companies really showed how incompentent they have become).
You can even underdrive the LEDs and use more of them, to extend their lifetime....been there, done that.
jtr1962 said:
I don't doubt HarryN's credentials. I'm just saying that this is the first I've heard of this difference in lifetime between different LED suppliers. And how does one know which chips end up in the LEDs you buy anyhow? For all I know the cheap HK LEDs may be using chips from Cree or Nichia. I'm planning on doing some long term testing of the various colors I have. I already know UV dims by quite a bit even after a few hundred hours, but this is a problem common to all UV LEDs as far as I know, or else they would be used in white LEDs instead of blue. Anyway, I'm finding this all very disturbing, especially if the LEDs in traffic lights don't last the claimed 100,000 hours.
[/ QUOTE ]
Dunno who HK companies are utilizing. Some Asian LED makers are in fact using CREE die, as well as Toyoda Gosei, the CREE die are distributed and rep'd by Sumitomo, the last purchase was for 160,000,000.00 dollars of LED die (the tiny chip inside the LED). There are quite a number of island based die makers too though....
In fact, Toyoda Gosei has even made die for LumiLEDs.
Some of the big factors in LED life in recent years will be in the packaging. Die attach adhesive, and how well it resists light (especially UV) (some are actually soldering down the die themselves, where Luxeons have a die that is floating on I think it was 16 solder balls that rest on a ESD or Zener diode, dunno if they solder their zener to the heatsink though, unlike others who do), in whites, the life of the phosphor (phosphors also have a lifetime), the filler material often used to make the body from (as in 5mm)- as most by know know the epoxy yellows with time/temperature/light exposure, some fill with versions of non-yellowing silicon (for example, GE's X-14 LED silicone) (such as newer Nichia's, CREE XLamps, LumiLEDs Luxeon, and OSRAM's Dragon) some have an acrylic dome (seems those'd yellow too with time) to cover the silicon gel, others use no dome, or even a glass lens. Additionally, you'll find some even in a metal can with a glass lens, and Microsemi was pursuing a metal leaded, and all glass package (no goop, gunk, plastic).
I've querried Toyoda Gosei (who makes a LED with a UV die and RGB phosphor, which doesn't rapidly degrade like everyone said it would), Nichia, and CREE, none of them have ever heard of the UV attacking the DIE itself, unsure where that rumor came from.
In fact, you'll find 20,000 hour UV LEDs here:
20,000 hour UV LEDs
Oh, there is a huge difference between LED suppliers, as they utilize different packaging, and I have some white LEDs from an asian source that keeled over an barely glow anymore, even ran within their specs. The quality of the die itself, can't tell you much about that.
The reason blue is used instead of UV is the YAG phosphor is matched for the blue. The folks like Toyoda Gosei, who have been making a UV based White LED utilize a special version of RGB phosphor. The color rendering is much higher than the blue die-yellow phophor, and pictures taken where those are utilized as a light source look *much* more natural.
You can read more here:
http://www.microlamps-tg.com/
http://www.microlamps-tg.com/html/Prod_white_compare.htm
Datasheet:
True White Hi
Catalog:
http://www.microlamps-tg.com/html/Cat04_01.htm
As far as the 100,000 hours, alot of it has to do with the skill and experience of the companies that designed the traffic light. This is one area where buying from the cheapest bidder can result some very unhappy campers. It's these companies that will give LEDs a bad rap. Just like the millions of CFL bulbs made by GE and Sylvannia that were only lasting a few months for many folks (yeah, other brands had the issue, but those two companies really showed how incompentent they have become).
You can even underdrive the LEDs and use more of them, to extend their lifetime....been there, done that.
Lightingguy321
Enlightened
- Joined
- Oct 5, 2006
- Messages
- 654
12" GELcores from GE have 192 LEDs in them. I can't recall what ecolux 12" modules have in them numbers wise
RODALCO
Newly Enlightened
In New Zealand we had initially the orange LED clusters failing. ( about mid 2005 ). Now I see that some green and red LED clusters are affected as well.
I got hold of a couple of broken red and green ones via someone I know in the traffic signal department.
There are different arrays of LED's in use, 230 Volts direct with a fancy electronic supply which drives the LED's at their max current of 20 mA's each. in parrallel lots of 4, in series strings.
Some lamps are arranged in a nice circular star pattern while others are leds in lines behind freznell lenzes.
Typically one LED fails and shorts out the other three hence 4 LED's are out on the board.
By replacing the one LED, the string works again properly.
I use a red and green as growing lights for my mango's and plant cuttings.
Will take some photo's soon and post. Also from some failed ones I'm aware of in my area.
I got hold of a couple of broken red and green ones via someone I know in the traffic signal department.
There are different arrays of LED's in use, 230 Volts direct with a fancy electronic supply which drives the LED's at their max current of 20 mA's each. in parrallel lots of 4, in series strings.
Some lamps are arranged in a nice circular star pattern while others are leds in lines behind freznell lenzes.
Typically one LED fails and shorts out the other three hence 4 LED's are out on the board.
By replacing the one LED, the string works again properly.
I use a red and green as growing lights for my mango's and plant cuttings.
Will take some photo's soon and post. Also from some failed ones I'm aware of in my area.
RODALCO
Newly Enlightened
http://farm1.static.flickr.com/55/131301311_b0b65e12ab_o.jpg
Orange LED string with 4 LED's failing at Mc Leod Rd te Atatu, Auckland, NZ.
Orange LED string with 4 LED's failing at Mc Leod Rd te Atatu, Auckland, NZ.
greenlight
Flashlight Enthusiast
I see mostly greenlight failures, but I'm usually moving too quickly to get a picture.
