Clock’s White LEDs Meet an Un-Timely Demise
Bad LEDs don’t die. They just fade away and sometimes exceedingly quickly, like this LED nightlight example. Observant reader Clive Mitchell sent in this rather amusing example of a large display LED clock gone all wrong: 
“This white LED digital clock displaying departure times at the Glasgow Central Station is clearly suffering from LED fade in a large scale manner. The clock, installed in 2005 and subsequently junked, was one of the first uses of low-quality, Chinese white LEDs. Up to this point, the traditional Gallium Arsenide LED had been considered almost indestructible. So it was a real kick in the teeth to a lot of companies when unreliable LEDs hit the market. I, too, got my fingers burned when I used a load for a TV production. Fortunately there were enough in the application to compensate for the ones that failed.”
Low-quality LEDs are something that some manufacturers today admit is problematic, in large part because the design flaws are not evident until after thousands of hours of testing.” So what’s a design engineer to do? A good adage to follow is, “If it seems like too good of a deal, it probably is.” But since all inferior products don’t necessarily carry a low price tag, Georg Bogner, Director, Visible LED Engineering at OSRAM, stresses that engineers should be demanding their vendors to, ‘Show me the data.’ Poor color stability and a reduced lifetime are two key hallmarks of an inferior LED design. So at minimum, engineers should request data on luminous intensity over time and color coordinates over time. And they should try to get the data over a temperature range as well.” Of course, the mere act of supplying a printed sheet of “test data” isn’t a guarantee that any actual testing has been done. “Getting this data is expensive,” says Bogner. “An engineer needs to make sure a potential vendor is in it for the long haul — that they have a serious intention to design a good product and have made the appropriate investment in R&D.” Chart: Luminous intensity over time:View image Chart: Color coordinates over time View image
5 Comments
-
June 27, 2007
i had a similar experience recently with a 12-v 48-led lamp i had converted for reading lamp use. it worked fine for several hours and then about nine leds went out and a bank of four began to flicker on and off.
subsequent checking discovered that the so-called “12v 500mA” power supply i had used was putting out a bit over 15v when loaded by only one lamp [about a 250mA load, by itself.]
check the sign for a bad power supply running above the rated voltage that the LEDs are expecting!
+af -
June 27, 2007
first, the two image links point to the same “intensity versus time” graph….
second, i recently had a bunch of LEDs in a 48-LED lamp go out. i had adapted two as reading lamps to replace the 150-w lamp my wife was using to read by. the 48-LED lamp was rated at 12v and about 300mA, so the roughly 3.5w seemed a nice improvement.
after several hours’ operation five or so LEDs went dark and a bank of four more began to flicker constantly.
i checked the power supply i had used: rated at 12vDC and 500mA. with one lamp turned on, it was putting out a bit in excess of 15vDC.
check the power supplies in that sign before you write off LEDs, Chinese or not. -
January 03, 2008
The Color Coordinates link has a typo. Use the “Luminous Intensity” URL, but change that portion to “Color Coordinates”, and the other chart shows up.
-
March 12, 2008
I had this kind of thing, but from overheating the LED’s.
If you solder LED’s at too high a temperature, they tend to work for a few weeks, then start to flicker and die. What’s happening is that the die bond wire flakes off the surface of the chip, because of the soldering heat stress followed by lots of heat & cool cycles. The problem is that the temperature you need to solder it at, with lead-free solder, is actually higher than the temperature the LED will tolerate.
So you need to leave the standoffs on (never solder the LED flat to the board), use a good iron at the right temperature, and never leave the iron on the joint for longer than 3 seconds. It slows the operation down, but the LED’s should survive. -
June 05, 2009
“with lead-free solder, is actually higher than the temperature the LED will tolerate.”
Great.
Imagine all those poor little 0805 SMT LED’s getting pan fried for the sake of ROHS.