Gradual LED Failure

This is a educational sight on LEDs and they sell the brightest LEDs I have ever used. https://www.superbrightleds.com/led_info.htm I had a professor in a analog class tell us that the half life of an LED is 40 years. I have a design that uses a IR LED and detector from Radio Shack that has been operating for 27 years almost every day and the design still works. Radio Shack has terrible LEDs. Did you know that LEDs are static sensitive? See the sight very useful info and great LEDs but they are expensive compared to your average LED. I have had the super brights running off and on for about a year and they blind you if you stare at them. Maybe reading this site will help you figure out the problem.

Thank you for the replies. tcmtech, these are indeed "white" LEDs (a fact I didn't think to mention in my original post), and your phosphor explanation sounds like it may be what's happening here.

That's a good idea. I'll try to check the temperature. The resistor is soldered directly to the lead, so they are fairly close together, in sort of an epoxy casing that's made to fit into the socket.

Thats a good method.

Good informative answers. Leds esp hi brite ones are critical in regards to power & heat- there must be cooling- even with the perfect set-up & the makers saying 100k life- reality is at 50k output is halved usually. The exception to the rules is by pulsing the leds- where life & brite can be maintained.

Lots of good tips already, all should be read and understood fully.

I personally find that running such LEDs under the rated current (in your case dropping the voltage) is a good idea. You do not appear to be controlling the environment for the LEDs very well, so dimming is to be expected......sadly!

Also having a better quality system to monitor/supply the current is a must if the LEDs are to be on so much of the time.....there are many companies producing chips for this and easy to read documentation, TI for example (but there are many companies that can help) has a lot of info which is downloadable in .pdf format.....

Start here first:-

Texas Instrument LEDs

To my mind, you are treating the LEDs too simply. For a long life you need tight control of the temperature, current and voltage. Also do not drive them at the maximum "allowed" current, drop to around 80% or so.....less would be even better!!!

Buy LEDS that at 80% (or less) current supply the brightness that you are looking for.....or simply use more LEDs......at the lower current.

Also email (on CR4 email system) our good LED friend Campbell Lighting

He is a good CR4 LED lighting man for more and better help. Send him the weblink for this blog as well so he can read it all.......

Best of luck.

hi cube,

If my Math is correct, at 10 hours a day the LEDs should last ~11 years.

I have used white LEDs and, the life although estimated to be as you say 40000 hours, a much more realistic figure would be around 3 years, at 10 hours a day.

I think it is the spikes caused by constant turning on and off which may be to blame. If they are in a position where they may get battered by things being moved or people walking by, or perhaps the branches from a tree, this would or could diminish the life.

To test the possible 'spike' theory, you could run some LEDs on a UPS (Un-interruptible Power Supply) or a Surge Protector.

You may like to think about using thyristors in the layout as well. All helping to supply a power supply which is gentle of the LEDs and the thyristors can take care and help control temperatures. (I think)!

From experience of using LEDs, I would say the life of LEDs is shortened by a very rough 70%, when they are used for just a few hours a day, and so are possibly liable to receive a spike each time they are switched on, compared to LEDs used for 24/7.

have you checked on-line for the particular LEDs you have and the test conditions they went through?

I think though cannot recall precisely, that using a variable copacitor, which can turn the LEDs on over a longer period (Comparitively speaking), of maybe 100th second, or anything up to a tenth of a second perhaps?

This is just a theory of mine, but automatically turning the power on over a short period may help prevent spikes and thermal shock, though I realise thermal shock may not be such a problem.

If you have so many LEDs it would be relatively easy to isolate a bunch and try this slowly switched power supply and compare it to other groups that have not been touched?

-----------------------------------------------------------------------------

I am not an electronic engineer so I suggest the above as a possible remedy to any spikes, but I only know they should be before the LED/s, and apart from a rectifier I will leave that as a suggestion for you.

The details below are just a short part of a whole lot of detail on different types of LEDs and way to power them. They do seem to say thermal shock over time is a known reason for white LEDs losing brightness. I found it to be a great piece and you can follow it up if you think the same as the site has several different ways to stabilise and supply power to various types of LED. Check it out below.

http://www.arrowne.com/innovation/archive/print_10_2006.pdf

Small lighting LEDs such as those mentioned for camera applications
can be readily designed into a new product. On the other hand,
for sustained operation at light output levels close to those of
traditional incandescent lights, additional factors must be borne
in mind. For example, since the light output from an LED falls as
the internal temperature of the chip increases, large increases in
temperature will lead to visible variations in light output. These may
impair the visual effect of lighting products or systems using high
brightness LEDs. Excessive internal temperatures will also reduce
the lifetime of the LED, leading to early failure.
High brightness LED components tend to exhibit more thermally
aware design than traditional indicator LEDs. However, care must
be taken to ensure low thermal resistance paths away from the
device. This may comprise a thermal substrate such as IMS, as well
as efficient bonding to purpose-designed heatsinks or other large
metallic surfaces.
Driving a high brightness LED also demands a different approach,
compared to design for ordinary LEDs. For example, the standard
approach is to use a series resistor to limit the forward current
and thereby regulate light output. In a high brightness application
drawing several hundreds of milliamperes, the series resistor may be
required to dissipate several Watts. Power resistors are expensive,
and also bring additional overheads such as the need for hand
soldering and attachment to an additional heatsink. On the other
hand, adding extra LEDs to a given string, to reduce the current in
the resistor, makes the LED array more susceptible to variations in
supply voltage, causing larger current variations in the LEDs.
To allow a high brightness LED to be operated close to its maximum
capacity, despite variations in temperature, supply voltage, and also
manufacturing variations from device to device, a constant current
drive may be desirable. This requires extra design work to configure
a linear or switch-mode current source. This approach is naturally
more complex than a series resistor arrangement. However, a
current source can be built efficiently using SMT components and
allows lower LED forward voltage conditions, higher supply voltage
and reduced power dissipation. This typically allows the LED count
to be reduced, leading to lower costs, easier assembly and smaller
overall dimensions.

---------------------------------------------------------------------------------

Good luck and please let me how you get on, OK?

Thanks Andy for that good word.

When LED's for lighting purposes first started to circulate, most were overpowered, thereby causing some of these premature failure problems.

A basic understanding of LED operation will help in choosing the proper LED.

Red and yellow/orange 3-5 millimeter low power type run on 2 volts each, while white, green, and Blue run on 3 volts each.

Care must be taken to use the proper MA transformer, as both undersizing, and oversizing will shorten life.

As was mentioned, heat is another factor in short life, as well as misuse of capacitors.

We have a 2 watt/60pc white Led them has a IC chip that is holding up well in Chicken house environment, and it is fully dimmable.

I have just received a request from quite a large customer to design a LED type soft neon application that must work in a country that has extremely high summer heat.

My first inclination is to use UV stabilized PC tubing, with a built-in air moving design to maintain tolerable temperature levels.

Anyone else have any ideas?, like you Andy??

This customer claims the enclosed regular soft neon runs too hot, and doesn't hold up well.

Donald

Hi Campbell Lighting we have missed having you around, still very busy?

Sounds to me like you already have a good idea on that problem you mention with the cooling air, my only extra that I can think of is a) filter the air first, you don't want dust build up inside the tube, b) have a temperature sensor where the air exits and drive the fan so as to maintain that temperature or less, which means that the first ones to be cooled will be cooler, but that is a problem that you can live with I am sure!

It just struck me that maybe the air should not exit, but just circulate and be cooled at some point - think about it!!

Keep in touch.

Hello All,

Andy just reminded me of a test set-up we used for an old sensor project. The salient points are that we used 'tube-in-a-tube' cooling and as the breadboard was used both indoor and out, UV Stabilized PVC was the choice for the external jacket of the cooling media. Second, there was the issue of "the first ones to be cooled will be cooler". Long, linear strings of small diameter tubing create a lot of back pressure. Consider that the effective cross-section of the 'jacket' is reduced by the inner tube diameter. What we ended up with was a manifold arrangement that was similar to an automotive type radiator. This was constructed in staggered 'halves' to allow assembly. By using the 'radiator' concept, we were able to significantly reduce input to output delta t.

Also, the manifold arrangement reduced the back pressure from our original linear string. As (my enfeebled mind) recollection serves, we used the (12Vdc) water pump from one of my colleague's motor homes (his sink has never worked the same, since) and started with plain water as the coolant. This got us into the ballpark. The next generation used automotive coolant (ethylene glycol) and was the most thermally effective. The final iteration used a glycol derivative that was not as toxic, slimey or hard to manage when spilled. As the pump in question was a 'demand' type, I don't think the system ever ran at over 6 PSI.

As for the spacers required to keep our tubes concentric, we ended up with the center splines from a lot of Lord shaft to shaft couplings. NOT particularly elegant, but then when you are in the Mojave Desert, you make use of what is in ready supply (and is open to be 'dog-robbed').

I'm tossing these 'notes' out in the hope that one or more might bear the slightest resemblance to the application at hand. Maybe, Cube Wave, you could post a few photos??

Please keep in touch on this one and let us know of developments/ultimate solution.

Warmest Regards,

GLB

Wow, this thread really took off over the weekend! Lots of helpful information here. I can elaborate on the setup, and I'll see if I can come up with some photos.

The application is an exhibit display case. The inside is not easily accessed, and it's full of delicate objects, hence the need for low-maintenance lighting. The LED/resistor assembly comes already together from the manufacturer, as do the sockets and wiring (and they provide the power supplies). (I am talking to the manufacturer about this dimming issue, too. I imagine they are not the original manufacturer of the LEDs themselves, and they have so far resisted giving me the full specifications of the actual LEDs, so perhaps something is up.)

I did manage to get a temperature probe inside the epoxy casing during operation and found a highest reading of 52 degrees Celsius very close to the resistor (versus an ambient temperature of 21 degrees Celsius). The casing is in two parts: one fully encloses the LED (except for the leads) and then nests in another which contains the resistor. Then this whole assembly goes in the socket.

Unfortunately, these LEDs are intended not as indicators, but as illuminators of other objects, so brightness is critical. Voltage spikes would be the easy thing to fix, but it seems as though things may be thornier than this if heat is the big issue. You've given me quite a few good suggestions and directions to go in. I'll update with more developments as I have them. (I'd prefer fiberoptics for an application like this, actually, but switching at this point would be impractical at best.)

Thanks!