Hey guys,

I'm planning on making an LED display that will be outside, so I expect temperature ranges from -40 to +40 Celsius.

I'm going to drive the LEDs with a lm317 configured as a constant current source which should be ok at that temperature. What I'm wondering is how reliable LEDs are at that temperature. From what I've read, most should be fine until -40°C, but I don't want to have the replace the LEDs. Does anyone know if there are there special types of LEDs that are better suited for low temperatures?

Also does anyone know if there are resistors and capacitors that are better suited for low temperatures (the resistor and capacitor used in the lm317 circuit to set the current) or should I just use normal ones.

Note: The electronics will be in IP65 housing so water should not be a problem, but humidity might be.

Thanks.

 

Why Don't you try military grade components and seal them in potting

 

I don't think you will have any problem, LEDs do generate a little internal heat as they light. A little insulation will go a long way. Are you wanting to light them continously?

I've never heard of them having a temperature issue (especially cold). But then, there is a lot of things I've never heard of.

The high power LEDs (1W or greater) get extremely warm.

As mentioned, mil speced components won't have the problem at all. Resistors are very rugged, and non-electrolytic caps should be also. The LM317 may have problems, just follow the data sheet.

 

This might be a hint at the answer to your question: Philips' Affinium LED modules LDM400, for refrigerated display cases are only rated to -22°F (-30°C).

 

I'm sure the LEDs and the LM317 will work at -40C. The problem is materials - some plastics get very brittle at low temperatures.

 

Hey,

Thanks for the suggestions. In case anyone is wondering, these are the parts I found that would work with the attached schematic for the constant current source.

I'm still deciding if I'm going to use the fuse or not. The idea is that if something happens to the regulator and it supplies too much current, or if one of the LEDs goes out and the remaining LEDs get too much current it should shut off instead of damaging the remaining LEDs. I'm not sure how effective this would be.


Part details:

LM317: MIL spec version that comes in TO-3 'can' package. Found on e-bay.

Capacitor: ceramic capacitors generally have -55°C ratings.
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=445-5303-ND

Resistor: Carbon Film resistors generally operate at -55°C
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=13H-ND

LED: Not all manufacturers indicate operating temp, of those that do Cree is the only one that specifies -40°C in the 5mm round package.
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=C535A-WJN-CU0V0231-ND

Fuse: Most companies don't specify operating temp range for PTC resettable fuses, Littlefuse is the exception. They indicate the actual characteristics of the part at up to -40°C.
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=F2000-ND

 

You're LEDs should be connected in series. You cannot reliably light parallel LEDs. Even if all three lit you could not guarantee equally divided currents, and therefore equal brilliance either.

 

Are the LEDs actually going to see -40'C? I expect the low temp will occur at night and the LEDs would be on at night? In that case the heat from the LM317 should stop the IC and LEDS getting that cold. As Bill_Marsden said a little insulation will go a long way.

 

Note that the LED forward voltages may increase significantly at low temperatures: you should make sure that your drive circuit will have sufficient voltage headroom to allow for that.

This pdf from Osram may be of interest - note that they do characterise products down to -40C. http://ledlight.osram-os.com/wp-content/uploads/2011/01/Thermal-Characteristics-of-LEDs-02-.pdf

 

Good point.

Well if he has already decided on using a LM317 in each remote light fitting he can wire the LM317 as a constant current source, which will always run the LEDs at constant current regardless of temperature. And it uses less resistors.

 

Hey, thanks for the suggestions.

-40°C is the worst case scenario I am designing for, it is pretty rare. The LEDs will be sandwiched between two pieces of acrylic and I will probably fill them in with some kind of clear silicon gel/glue. Any suggestions on a specific product to use? I believe this is similar to potting, but I need it to be clear.


If they are connected in series, the entire chain would go out if one of the LEDs stopped working. Right now, I am designing the current source to provide 10mA to each LED (there are actually 9 LEDs, I just drew 3 of them). I don't really care about how bright they are since they are accent lighting, my priority is longevity and reliability. So as it stands, if one of the LEDs stops working, the remaining 8 will have the current distributed among them. However, since they are rated at up to 20mA, it would take many failed LEDs before it became problem. I also have a fuse set up on the power supply so that if there is too much current, it shuts off. Given this, is it still a better idea to use a series configuration over parallel? Is there something I'm missing?

Finally, from what I understand, the LM317 configured as a constant current source is fine with up to 37V output. That should be more than enough, even with 9 LEDs in parallel. Should the input transformer have a voltage higher than what the output is expected to be or does it not matter?

Thanks again for the comments!

 

Yea, stuff the entire circuit into an insulated box and if it has almost -0- current draw sneak a small resistor in there off the power input to serve as a minimal heater.

 

With R1 as 13 ohms your I source will drive a nominal 96 mA into the LEDs. Even with 3 in parallel (which occasionally will divide the current equally between all 3) you have 96 / 3 = 32 mA into a device rated for 25 mA max.

You do not specify the input beyond "transformer input" but a transformers output AC, so you need to rectify it, and if you want a true DC you need a lot more cap then the 0.1uF you have.

I would dump all that. You are lighting a LED and all you need to do is limit it's current. That isn't a precision operation and doesn't need an IC and resistors, just the resistor should do.

Here:

This way you can run lots of LEDs. Pick just about any low voltage output transformer. Except for the very lowest of output voltages you'll need blocking diodes, as LEDs make very poor diodes in the reverse direction. The rectifier diodes can be just about anything too, something like 1N400x. Two strings of LEDS will balance the transformers current, not necessary for just a few LEDs but something to do for when using the transformer up to it's max power rating.

The resistor's value is always a trick as you should look at it to see if you got it right. One way to compute it is to try : 1/2 * Vsec / Iled as a starting point, where Vsec is the RMS output pf the transformer, and Iled is the current you want in the LED.

Don't worry over getting MIL grade components, just look for a temperature rating to cover your range.

 

...Any suggestions on a specific product to use? I believe this is similar to potting, but I need it to be clear.

The rounded end of the LED is a lens, so if you cover it you make the light pattern it emits.

"Potting" is just epoxy and expensive. "Acrylic caulk" is inexpensive and good stuff, comes in clear and lots of colors too. Get it at any hardware store or home center. Once I waterproofed a replacement fluorescent lamp ballast with "Great Stuff" but I digress.

Finally, from what I understand, the LM317 configured as a constant current source is fine with up to 37V output. That should be more than enough, even with 9 LEDs in parallel. Should the input transformer have a voltage higher than what the output is expected to be or does it not matter?

The 317 output voltage will be around 3 or 4 volts, just the LED forward voltage. The 317 needs a few (3-4? its on the data sheet) volts minimum to run so you are looking for a "transformer" with about a 10 V output. And getting that as a DC if you insist on using the 317.

 

Hey, thanks for the suggestions.
...If they are connected in series, the entire chain would go out if one of the LEDs stopped working... Right now, I am designing the current source to provide 10mA to each LED (there are actually 9 LEDs, I just drew 3 of them). I don't really care about how bright they are since they are accent lighting, my priority is longevity and reliability. So as it stands, if one of the LEDs stops working, the remaining 8 will have the current distributed among them. However, since they are rated at up to 20mA, it would take many failed LEDs before it became problem. I also have a fuse set up on the power supply so that if there is too much current, it shuts off. Given this, is it still a better idea to use a series configuration over parallel? Is there something I'm missing?

Yes, I think there is something you may be missing. Filament lamps almost exclusively fail open-circuit, but this is less usual with LEDs. More commonly, they simply become inefficient and emit less light. Short-circuit failure is not unknown, so any LEDs operating in a parallel group would need substantial individual ballast resistors to avoid the possibility of one shorted device causing total failure.

Open-circuit failure is not impossible, of course, and possibly your severe temperature environment makes it more likely. A series chain can be protected from this by placing a Zener diode in parallel with each LED, with the Zener voltage chosen to be just above the LED forward voltage at the lowest working temperature.

 

@ ErnieM: I would not recommend un-smoothed half-wave rectified operation for LEDs. The flicker (at 25Hz or 30Hz depending on mains frequency) will be very noticeable, and the effective current through the LED will be more difficult to calculate.

 

Ernie, your schematic in post 13 will not work. Check it again.

 

Add a 100µF capacitor, as was suggested in post #16. Post #17 also is valid.

I'd go with a full wave bridge and make a conventional power supply. In the long run it is simpler.

Or you could use a wall wart. All it is is a power supply built next to the wall outlet.

 

Ernie, your schematic in post 13 will not work. Check it again.

The basic concept is wonderful, its just the expression that has a minor glitch. Happens to all of us.

By the way, there are LEDs out there with internal Zener diodes so if the LED opens, the voltage changes to that of the Zener and the series string continues to operate.

 

Ernie, your schematic in post 13 will not work. Check it again.

Heh, my bad.

Adjuster: I have several sets of AC driven Christmas lights that don't flicker. You can see some flicker if you shake them fast, so as the doctor said; "don't do that!"