Dimming of white LEDs over time

[kjoiner]

Hello,

I'm usually posting in the mechanical engineering groups but I've run into a problem that hopefully some others have seen before. We have a light kit that uses a small circuit board we purchase. We install 9 white LEDs on the board. After about a month, the LEDs dim and the light output become unsatisfactory for use with a camera system that needs the proper amout of light. Here are some spec:

LEDs - white, 2500mcd, 45° view angle
Circuit board - Series circuit of 3 LEDs with provision for a resistor to limit current draw. The circuit board places the 3 series circuits in parallel. That way it only takes 3 resistors to limit the current.

Power supply - Switching wall transformer type. 12.18 VDC output. UL approved.

The resistor is a 120 ohm, 2%, axial lead type, 1/4 watt rating

The resistor should limit the current per LED to 20mA. Usually LEDs can take 30mA. We should be well within the limits of the LED.

Here's what is strange:

The unit that came in from the field with dim light draws 22mA.

A new unit has much brighter light but only draws 19mA.

I worked out the current draw when designing the light kit with our electrical engineer and checked all calculations and also checked the electrical performance with a DMM after we built a prototype. Now we have dozens in the field.

I have also run the unit to measure the temperature rise in the housing and the rise is only about 6°F.

Does anyone have a lot of experience with the white LEDs and have some feedback on what chould be the problem? Is it a heat buildup problem or could it be as simple as a quality problem with the LEDs?

If you need any more information other than what I have listed I will be glad to provide it.

p.s. - The manufacturer of the LEDs says that the temperature should be held to 23 degrees C. This seems very restrictive since that's room temperature. Most components can go well above that and I'm only a few degrees F above ambient.

Thanks,

Kyle

 

[IRstuff]

I don't understand how you are controlling the input voltage. If the range is truly 12-18, you are potentially putting more than 50 mA in the LEDs with a high input voltage.

How are you removing heat? while the power dissipation is low, the lack of a good thermal path can still result in unacceptably high temperatures.



TTFN

 

[sreid]

I'd try some White LEDs from a known quality supplier, specifically, Agilent.

 

[kjoiner]

IRstuff,

The voltage coming from the power supply is 12.18 VDC not 12-18 VDC.

Regarding your comment about heat dissipation, that is one area I'm looking at. The circuit board is mounted on standoffs inside a housing. The housing is covered with a piece of textured lexan to serve as a light diffuser. There is a good bit of room inside the housing and the aluminum should conduct heat well and it's painted black so it should radiate any heat as well. The fact that the enclosure is covered with the lexan could be trapping some heat.

Sreid,

I'm looking at other suppliers right now. Cost is a factor (as always) so the $3.00 ones would be too expensive. I'm looking for something in the $1.25 to $1.50 range. Of course, cheap LEDs could cost more down the road so I should keep my options open.

Thanks for the quick response to both of you. I'll keep looking and pass along any information I gain. If you think of anything else, please post it.

Thanks,

Kyle

 

[IRstuff]

What's the environment? As you may be aware, the "white" LED is actually a fluorescence effect.

A coating is applied to normal LED, and its emission plus the stimulated emission from the LED makes the LED "white."

If the application is outside, there may be some potential for degrading the coating from solar UV.

TTFN

 

[kjoiner]

IRstuff,

We use the light kit for a video system. The camera is a USB type web cam. The system is in a compartment in a machine that is typically in an air conditioned space. There is some heat rise from other components in the compartment but no exposure to sunlight.

Kyle

 

[BrianG]

Hi kjoiner,

I am not surprised that the LEDs are producing less light with time. Since the earliest days of LEDs this has always been a problem, despite what manufacturer's would have you believe.
The mechanisms involved are quite complex and I'm no physicist, but as a practical electronics engineer I reckon it's still good practice to de-rate the device for temperature/power dissipation versus current.

Although newer LEDS are getting better all the time the design choice (for me) is still: a longer life at a lower, more even, brighntness or shorter life, high brightness!

 

[gr8electronics]

I think a couple of you have hit around it and this may not be it. But Phosphor, like used on CRT screens looses it brightness over time. The harder you drive it the quicker it looses its brightness... I think you may need to pulse modulate the LEDs instead of leaving them STEADY ON!

 

[kjoiner]

Hello,

I've done some testing and ended up going to another source for the LEDs. I'm using LEDs made by Nichia. They were able to provide detailed specs on the LEDs and also some recommendations for driving them. In our application, pulsing them is not really an option, so we have done 2 things:
1. Increase the resitors on the board so that the current is now only 10mA instead of 16mA.

2. Turn the lamp on only when the digital video system is recording. In this case, the LEDs run only for 30 seconds at a time.

The main problem with our previous brand of LEDs was that the voltage drop across the LEDs was not consistent. Based on specs I had at the time, I assumed the drop was about 3.4 V. It turns out that the drop varied quite a bit, some LEDs had only a 3.0 V drop. This was allowing too much current in the circuit. The new supplier, Nichia, has a much better tolerance on the voltage drop. Their light output is also much higher with less degredation over time.

Kyle

 

[9QCAC]

LED's work better if powered using a constant current source, and life time of the LED is longer when using pulse modulation even for a full brightness. ie apply a 95% PWM square wave instead of a DC current. Make sure you do not exceed the maximum RMS current in the LED. (Use a True RMS Ammeter)

 

[courtside]

I work extensively with LEDs and have learned this the heard way. You've done the right thing in going with Nichia - who pioneered blue and white LEDs - and have the best thru-hole lamp on the market.

Plug-in wall adapters are not usually regulated. If you are arranging 9 LEDs into 3 groups of 3, your load should be pretty close to 60mA. If the adapter is rated for say, 500mA then that is the voltage at which it delivers 12VDC, otherwise you could be delivering anywhere up to 18V into your fixture - depending on the quality of the power supply.

Another thing to try is installing a varistor or TVS somewhere on your PCB as well. This will help protect the LEDs.

If you're using a regulated power supply, then this post is moot, but I thought I'd try to share... being my first post and all

I also buy Nichia LEDs in substantial volume and could sell you finished circuit boards and/or diodes themselves. Let me know.

 

[kjoiner]

Courtside,

Thank you for the response. We are having much better luck using the Nichia LEDs than what we used from the previous supplier. They were very helpful with technical advice and also tested the LEDs from the previous supplier - those LEDs were nowhere close to the rated specs and had quite a bit of variation. The Nichia LEDs substantially better in quality and output.

We increased the resistors to cut down on the current draw and this has worked well. The wall transformer that supplies power to the LEDs is regulated and UL listed which is a requirement for our product line.

Like you mentioned, you sometimes have to learn the hard way but all posts I've recieved have helped a great deal in making sure the real world conditions are considered and I think we have a good lighting system now.

Thanks,

Kyle

 

[IRstuff]

According to some accounts there's over a factor of 10 difference in lifetime between different manufacturers for LEDs operated under identical conditions.

TTFN

 

[nbucska]

You may extend the lifetime somewhat by using a phototransistor for current control so at the beginning
you may use the minimal current which is just enough
to give the required illumination.


<nbucska@pcperipherals DOT com> subj: eng-tips
read FAQ240-1032

 

[Warpspeed]

As IRstuff has already mentioned, white LEDs are different to all other LEDs because they use a fluorescent material to generate the white light. This material can as I understand it degrade significantly with time, and the light output fall.

White light contains the entire visible spectrum, whereas a normal led is monochromatic with one color only.

If this equipment is to be used with a colour camera, white LEDs will obviously be required. But if a black and white camera is being used, infrared LEDs will work much better as a source of illumination.

For many industrial applications black and white is quite adequate.

 

[zappedagain]

I've heard (second hand) that the loss for white LEDs follows a logarithmic curve over time. Can you burn-in the LEDs first to get them past the 'knee of the curve'? Then the variation won't be as drastic. From the original post I see this would take some redesign so you end up with a light bright enough after then burn-in.