Resistor size for LEDs

[chimaera]

I am not that experienced with electronics, so this may be a stupid question. I have an application where I have used LEDs to light a surface. My source voltage is 11.1 v and I am driving the LEDs at 20 mah ~3.6 v with a 1/2 watt 390 ohm resistor on each LED. I decided to use a single 5 watt 390 ohm resistor to power the whole line of 6-10 LEDs so I wouldn't have to place a single resistors on each LED. What I got was a significant dimming of my LEDs. I'm assuming 5 watts is too large, but should that dim the LEDs? Any help would be appreciated. Do I have to match the watts with the number of LEDs I will be using?

[Rob]

Welcome to the boards.

The wattage rating of an LED is going to be driven by the amount of power you're asking it to dissipate. For the simple series case,

P = (V_S - V_F) * I_D, which is just the current through the resistor times the voltage drop across it.

So for V_S = 11.1V, V_F= 3.6V, and I_D = 20 mA, I get
P = 0.15W, or 150 mW. So a 1/4W resistor would suit this circuit fine.

Now, when you combine LEDs in parallel from a single resistor, you just add the currents. But, that also changes the resistor value you need.

R = (V_S - V_F) / I_D, so as you increase I_D, you'll be wanting a smaller value resistor in ohms.

For the 10 LED case,
V_S = 11.1V, V_F= 3.6V, and I_D = 200 mA, I get
R = 39 ohms.

This is made a lot simpler by using my calculator:
http://led.linear1.org/1led.wiz

Just treat 10 LEDs in parallel as one with 10x the current rating.

[chimaera]

Thanks for the quick reply.

Basically I want a light rail that is a constant voltage that can have varying numbers of LEDs (with the same forward voltage) plugged in at any given time. I understand the current limiting issue with LEDs, but is there a way to make this work without a resistor at each socket? I have been looking at a small switching power supply that will give me my target voltage of ~2.2 v at 1A output. This would mean I could still overdrive my LEDs right?  Is there a way I can do this without current matching the rail with a min/max number of LEDs that keeps the circuit in a zone? I hope I'm making sense.

http://www.dimensionengineering.com/DE-SWADJ.htm



Thanks again!

[Rob]

I'm not seeing an obvious solution given those constraints.

You probably know that current versus voltage is a constant for a given temperature--you can regulate the voltage supply to arbitrary standards, and what will still happen is that as the junction heats, you'll increaase the current drawn by the LED, which in turn produces more heat, and you have thermal runaway.

It might be inelegant, but at least a resistor per socket would be cheap.

[justDIY]

if you increase your source voltage, so that you can wire your leds in series, then you can use a constant current regulator, such as the lm317 (or newer kin), which will supply the correct current regardless of whether 1 led is connected or 10 leds are connected.  All that is required is the regulator IC and one resistor, so it's not too complicated.

[chimaera]

Thanks again guys. This has been very helpful.

One last question if I may.  If I run my LEDs for say 30 minutes and they are not heating up, is it safe to assume that I am safe and not overdriving them?  I have fried LEDs before and as I recall it happened fairly quickly and they got very warm/hot. I tried running just one LED on my tiny switched power supply and if anything It may have been ever so slightly warmer than room temp. Do you think this is okay? I know this is not "proper" electronics design, but I just need it to work until I can revise my circuit and do it properly.

[Rob]

For 5mm LEDs and their 20mA cousins in other assorted sizes, my rule of thumb is that proper operating current should not make them hot to the touch. So I think you got the right idea. Double check with an ammeter for extra insurance.

[chimaera]

Thanks Rob.

[justDIY]

if you figure your body temp is around 98 deg F so anything that feels "warm" should be over that temperature ... running LEDs "warm" means you're going to get maybe half or less of the rated hours for colored leds, and significantly less rated hours for white leds (check the data sheets for thermal derating and lumen maintanence vs temperature) ... in my opinion, this is A-OK ... I dont need a design to last 100k hours - for the majority of leds out there, the listed specs assume the junction will remain at or below 25C or 77F ... that is pretty unrealistic, so keep a big grain of salt handy.

the flipside ... if you're building these on a workbench, you'll be having ample ventilation for them, and they are still warm, then IMHO, they're being driven too hard to be enclosed in something... once they trap heat inside their enclosure, things can go south quickly.

it all boils down to your application - if you want a short but sweet life, drive those puppies to the edge of their ratings ... if you want a safe, long life from them, run at a conservative number ... for most leds half the rated power still produces the majority of the candela/lumen rating... that's a subjective opinion from my observation, I dont really know if its true or not I'm just saying, running a blue led at 10ma and another at 20 and another at 30... sure I can see a difference between 10 and 30, but the one at 30 is also nice and warm, the one at 10 is still too bright to stare at for a long time.

[chimaera]

Okay confession time. Here is a link to what I am actually doing with this project. I can guarantee airflow is not going to be a problem. Scroll down to post #8. This one is actually actually wired with individual resistors, but I am working on a more flexible design that will allow color changing. If I can get 500 hours or more I'm good.

http://www.rcgroups.com/forums/showthread.php?t=438925

[justDIY]

awesome

well, application is everything ... I'd aim for running those leds at 30ma (or even 50ma if they're quality leds), they don't need to last 1000's of hours, and like you said, they'll have plenty of forced air cooling.

go for a 1/4w or even smaller resistor, it'll also be getting plenty of cooling, save some weight there ... it terms of leds and voltages ... if your voltage is 11v, you can run any combination of leds in series, as long as the voltage adds up to less than 11

figure rule of thumb, blue, green and whites = 3v, reds, yellows and oranges = 2v ... so 11v could get you blue green red red with 1 volt left over ... use the LED calculator, punch in 11 for your battery, 10 for your LED, and figure on 30mA for the current, it'll tell you what resistor to use.

of course, this spits in the face of sensible LED design somewhat, but if you arrive at or close to 11v, you can try it without a resistor, since your batteries and wires all have a little resistance in them, and you don't need the leds to last forever.

[Rob]

Okay confession time. Here is a link to what I am actually doing with this project. I can guarantee airflow is not going to be a problem. Scroll down to post #8. This one is actually actually wired with individual resistors, but I am working on a more flexible design that will allow color changing. If I can get 500 hours or more I'm good.

http://www.rcgroups.com/forums/showthread.php?t=438925

Very nice, I dig the tag-team posting as well. If you would like to post some of those pics in the LED project showcase forums, that would be awesome.

[chimaera]

awesome

well, application is everything ... I'd aim for running those leds at 30ma (or even 50ma if they're quality leds), they don't need to last 1000's of hours, and like you said, they'll have plenty of forced air cooling.

go for a 1/4w or even smaller resistor, it'll also be getting plenty of cooling, save some weight there ... it terms of leds and voltages ... if your voltage is 11v, you can run any combination of leds in series, as long as the voltage adds up to less than 11

figure rule of thumb, blue, green and whites = 3v, reds, yellows and oranges = 2v ... so 11v could get you blue green red red with 1 volt left over ... use the LED calculator, punch in 11 for your battery, 10 for your LED, and figure on 30mA for the current, it'll tell you what resistor to use.

of course, this spits in the face of sensible LED design somewhat, but if you arrive at or close to 11v, you can try it without a resistor, since your batteries and wires all have a little resistance in them, and you don't need the leds to last forever.

Thanks Guys I appreciate the help. I checked the current and I get 15-16 mA on the rail driving the 2.2 volt LEDs (Red, yellow, orange, etc) and about 31 mA driving the 3.6 volt LEDS (blue, white, green, etc). I think this will work for my purposes. 30 is a little more than I would like, but as you said 1,000 hours is more than enough. The orange do not like this setup for some reason, but Red seems to be just fine. I haven't tried yellow yet, but I will. Again thanks for the help.