LED Power Help (Please!)

[anthony691]

Okay... I didn't know what I was doing, but I went ahead and wired up 2 superbright LEDs to the +5v DC on my computers power supply. The LED had two 3 volt batterys in it's former life as a flashlight. It quickly got hot and nearly burned me when I went to test it out. I havn't tested it yet w/ two, but I assume I will see the same result.

WHAT WENT WRONG? I am really at a loss (not too much of an electrician). Do I need a resistor or something? How big?

Thanks a ton,
Anthony691

[BobA]

Almost all LEDs need a limiting resistor.  Here is a link to a resistor calculator.  Seach on LED Calculator as this has been discussed before in this forum.

LED Calc

Your flashlight may have had a built in resistor or perhaps was designed to run both LEDs in series depending on the battery resistance to protect from burnout.  Without the proper resistor or circuit arrangement the LED will destroy itself.

It is much easier to calculate the required resistor when the LED is a known component rather then guessing at the requirements of one.   If your LEDs are white then you can guess at their voltage by comparing to other high output white leds.   Their max current is critical and you may have to experiment with larger value resistors reducing them as you determine if the LED is OK or burning up.


BobA

[anthony691]

Almost all LEDs need a limiting resistor.  Here is a link to a resistor calculator.  Seach on LED Calculator as this has been discussed before in this forum.

LED Calc

Your flashlight may have had a built in resistor or perhaps was designed to run both LEDs in series depending on the battery resistance to protect from burnout.  Without the proper resistor or circuit arrangement the LED will destroy itself.

It is much easier to calculate the required resistor when the LED is a known component rather then guessing at the requirements of one.   If your LEDs are white then you can guess at their voltage by comparing to other high output white leds.   Their max current is critical and you may have to experiment with larger value resistors reducing them as you determine if the LED is OK or burning up.


BobA

To clearify, they came from seperate flashlights with 2 3V watch battery things (CR2016).  There was indeed NO resister built in. This seems a little bit complicated... i will try out some resisters...

If the LED stays cool is it okay? can it damage itself w/o heating up?

[BobA]

A watch battery has a high internal resistance.   It cannot supply high current thus will not damage an LED even if connected without a resistor.

A white led has a forward voltage of 3.2 to 3.6 volts. Since you have stated that the LEDs are superbright start with a current to 30ma to be safe and you have a starting resistance of about 90 ohms.   If you want it brighter and it is not heating up go to a lower resistance.  If it heats up it will damage itself so do not lower the resistance too much.

You did not say what color so I am assuming white.  The calculation is different for different colored LEDs.

The one that got hot enough to burn you may have a limited life or limited output due to overheating damage.

BobA

[anthony691]

A watch battery has a high internal resistance.   It cannot supply high current thus will not damage an LED even if connected without a resistor.

A white led has a forward voltage of 3.2 to 3.6 volts. Since you have stated that the LEDs are superbright start with a current to 30ma to be safe and you have a starting resistance of about 90 ohms.   If you want it brighter and it is not heating up go to a lower resistance.  If it heats up it will damage itself so do not lower the resistance too much.

You did not say what color so I am assuming white.  The calculation is different for different colored LEDs.

The one that got hot enough to burn you may have a limited life or limited output due to overheating damage.

BobA

Thanks BobA
(it is white) And I am going to get another one to replace the burned one.

[grafixmonkey]

What I do for my LEDs is put it on a protoboard with a pot, start on max resistance (no less than 1 Kohm) and gradually decrease the resistance until you get what you want.

then leave it turned on for like 5 minutes, and make sure it isn't getting real hot.

Measure the pot's resistance and you're done.  then you don't have to worry about which LED, which manufacturer, etc. etc.

[anthony691]

What I do for my LEDs is put it on a protoboard with a pot, start on max resistance (no less than 1 Kohm) and gradually decrease the resistance until you get what you want.

then leave it turned on for like 5 minutes, and make sure it isn't getting real hot.

Measure the pot's resistance and you're done.  then you don't have to worry about which LED, which manufacturer, etc. etc.

PERFECT! AWESOME IDEA!

*Pulls out Fluke and a pot*

[RandyT]

What I do for my LEDs is put it on a protoboard with a pot, start on max resistance (no less than 1 Kohm) and gradually decrease the resistance until you get what you want.

then leave it turned on for like 5 minutes, and make sure it isn't getting real hot.

Measure the pot's resistance and you're done.  then you don't have to worry about which LED, which manufacturer, etc. etc.

This is a good method if you don't have a calculator at hand or know what the specs of the LED are.  But one should really give the LED what it is rated for.

Most LEDs can last for up to 10 years of constant operation if run within it's proper specifications.  Out side of those specs, the life can be shortened quite dramatically.

If you give the white ones too much juice, they will gradually do a color shift, and lose brightness as well.

Just some things to watch out for.....

RandyT

[grafixmonkey]

This is true...  if I'm designing something that I want to last for a long time I use the specs.  That's usually also when I'm buying the component though, and have access to them.  I've got LEDs sitting around in tubs that I have no idea what they are, but I've seen them light up once or twice...  then I crank the resistor until they look like they're supposed to.

Maybe after you've got the LED going, measure the voltage across it, subtract that from your voltage source, and divide by your resistance.  If you get significantly more than 0.02 for normal or 0.03 for superbright, you might be suspicious of overdoing it.


I know I toasted some new wuperbright white LEDs using this method, just because I thought they were supposed to be as bright as one of those flashlight things.  They weren't.  

[IceCold]

I'm soon going to do the same thing.  Except I just ordered 2 of them from www.superbrightleds.com .  Shipping was quick, and they all seem to work.  Plus, then you know exactly what they're rated at.  I found out I'm going to need an 80 ohm resistor for each LED, if I wire each one to a separate 5v wire off of my PC's power supply.

I have a question about LED's though.  If I wired up 2 of them in series, would each of them get half of the voltage?  Like if I hooked up 2 of them to a 5v power supply, would each only get 2.5v?  I don't think that's true, but is it?

[TazMan]

Or you could be lazy like me   and just get em from these guys on the cheap (with a built-in resisitor) <SuperBrightLeds.com>

[BobA]

If wired in series to the 5V supply the voltage drop across each LED would be 2.5 V.  Depending on the forward volatage required for your LED this is an acceptable way to wire 2 leds if the resistor is sized accordingly.   The drop would be 2x the forward voltage and the current would be the same as a single LED. A single resistor would be wired in series with the 2 LEDs.

BobA

[grafixmonkey]

Agreed with BobA.  

But don't mistakenly think "2.5 + 2.5 = 5, so I don't need a resistor!".  Still stick a small ohmage resistor on there, like 5 ohm, maybe 10 or 15.
 

And since this is becoming an LED tutorial, I'll make sure to mention that each LED gets its own resistor!  No saving resistors by using one of them on a whole parallel bank of leds.  They'll easily burn out or at least light unpredictably.

[anthony691]

Well, I am done! I got it working nicely...

[RandyT]

And since this is becoming an LED tutorial, I'll make sure to mention that each LED gets its own resistor!  No saving resistors by using one of them on a whole parallel bank of leds.  They'll easily burn out or at least light unpredictably.

This isn't exactly true.  It isn't uncommon to use a single resistor to limit current to multiple LEDS wired in parallel.  The only thing you have to pay attention to is the wattage rating of the resistor and the combined power requirements of the LEDs.

I use single half-watt resistors to power hi-output LEDs in groups of 3 all the time without any issues.  The resistor gets a little warm, but that's what resistors do.  They dissipate power as heat.  If the resistor is HOT, then you are pushing your luck and probably will have problems down the road.  Just use one with a higher wattage rating.

That being said, there is nothing wrong with giving each LED it's own resistor, but sometimes circuit size and complexity won't allow for this.  If you do the math and design your circuits accordingly, you won't have problems.  There are loads of tools on the web to help.

RandyT

[grafixmonkey]

The problem with that is you're getting away with it only because of the precision of the manufacturing.  LEDs drop an almost constant voltage no matter what current passes through, so if you give all of the LEDs one resistor and put them all in parallel, you're driving them all at a certain voltage determined by the current the resistor passes in that configuration.  Since the voltage is set for multiple LEDs, there's really nothing to determine how that current distributes itself among them.  If one LED changes for some reason and begins dropping a slightly lower voltage than the others (gets too hot, overpowered, or something) then the other LEDs would dim and that one would suck up more than its share of current, and blow.  Also, if one LED in the bank just happens for some reason to disconnect, blow, fail, or begin dropping more voltage than the others, all the others will suck up its share of current - for a bank of under 5 LEDs, one of them blowing would cause all the others to blow as well.

The point is it's an unstable configuration.  If anything changes the whole thing becomes more unstable, or just blows right away.  With one resistor on each LED, if one LED blows or short circuits then nothing bad happens to the rest.

Being limited for board space or wiring time isn't even a good reason to put LEDs in parallel, because there are bank resistors you can buy that have a common pin and a whole set of other pins that internally have one resistor between the common and each other pin, so you can wire up a bunch of LEDs with one component, same amount of soldering, same amount of wiring, couple more rows of board space, and have a proper circuit.

[RandyT]

The problem with that is you're getting away with it only because of the precision of the manufacturing.

And this is a problem because.....?  You have to place a certain amount of faith in the manufacturing process and quality control from the factory.  The fact is, if you have out-of-spec parts, you will have a problem no matter what.  Will you burn out more than one LED if you have an out-of-spec component in a group of 3?  Maybe, but that depends on just how far out of spec that one actually is.  

Should you give each LED it's own resistor?  Sure, if you can.  But not doing so isn't nearly the catastrophy waiting to happen that you are making it out to be.  I have about 30 LED lighting units that don't follow your rule, which have been "in the field" for about 4 years now.  To date, there hasn't been a single problem.

LEDs drop an almost constant voltage no matter what current passes through, so if you give all of the LEDs one resistor and put them all in parallel, you're driving them all at a certain voltage determined by the current the resistor passes in that configuration.  Since the voltage is set for multiple LEDs, there's really nothing to determine how that current distributes itself among them.  If one LED changes for some reason and begins dropping a slightly lower voltage than the others (gets too hot, overpowered, or something) then the other LEDs would dim and that one would suck up more than its share of current, and blow.  Also, if one LED in the bank just happens for some reason to disconnect, blow, fail, or begin dropping more voltage than the others, all the others will suck up its share of current - for a bank of under 5 LEDs, one of them blowing would cause all the others to blow as well.

The point is it's an unstable configuration.  If anything changes the whole thing becomes more unstable, or just blows right away.  With one resistor on each LED, if one LED blows or short circuits then nothing bad happens to the rest.

I understand how it works.  But it should take about 10 years for any of this to be an issue in practical application of a properly designed circuit with parts that aren't faulty to begin with.  If you are the type that likes to poke around circuits with metal objects while the power is on, you might want to give each LED it's own regulator too .

BTW, this type of balanced circuit is used in lots of places more troublesome than lighting a few LEDs.  I don't know how many times I have replaced dead parts only to find out that a different dead part in the same circuit kills the one I just replaced.  And they aren't as simple to find as an unlit LED nor as inexpensive.

Being limited for board space or wiring time isn't even a good reason to put LEDs in parallel, because there are bank resistors you can buy that have a common pin and a whole set of other pins that internally have one resistor between the common and each other pin, so you can wire up a bunch of LEDs with one component, same amount of soldering, same amount of wiring, couple more rows of board space, and have a proper circuit.

You might want to tell that to the folks that make these.  I might be mistaken, but I am pretty sure that the LEDs in those products are handled in banks, rather than individually.

The resistor arrays you refer to are very handy and a good choice.  But they aren't very cheap.  You also can't get these from Radio Shack, so tack on a few bucks for shipping.  But if you have them, and your project is already on perf-board, by all means use them.  

RandyT

[grafixmonkey]

And this is a problem because.....?  You have to place a certain amount of faith in the manufacturing process and quality control from the factory.

Well, every EE professor I've had and every circuit theory both ideal and realistic I've seen say don't do it, but I looked up some LED datasheets at Fairchild and it appears that some (not all) LEDs with no resistors inside seem to have some ohmic qualities to them anyway, about 1.2v difference or so when you go from "off" to max spec current.  That would make your balancing work.  

So, for situations where you are sure you have LEDs that are all the same part number from the same manufacturer, have specifically chosen LEDs which have shallower I-V characteristic slopes and are almost ohmic devices, and have calculated that given their characteristics they will work in this configuration, sure go ahead.  But that's not most of the people here.

Will you burn out more than one LED if you have an out-of-spec component in a group of 3?  Maybe, but that depends on just how far out of spec that one actually is.

The tolerance for some of those LEDs is paper thin when they're powered by a voltage source instead of a current source, as the original poster found out.  If the voltage over some of the LEDs at Fairchild changes by 0.1v and current is available, the LED will draw 20mA more!  In a bank of 3, if one LED in the bank dropped 2v at 30mA instead of 2.06, it would reduce the current through the other two to 25 mA, giving that one LED 40 mA to deal with.  That's a 0.06 volt tolerance for error.  Whereas with a resistor per LED, if one LED drops anywhere between 1.7v and 3v instead of 2.06, that LED would draw between 20 and 33 mA and operate just fine, and any other LEDs would be completely uneffected by what happens to that one.  That's a huge safety step compared to 0.06v.

Furthermore, if you try putting a green superbright LED from Fairchild in parallel with a red superbright LED from Fairchild, and calculate the resistor necessary to run both at 30 mA assuming they'll drop about 2 to 2.5 volts, you will find that a green superbright LED actually draws 2 mA at the voltage where the red one draws 30.  The green LED would be off, and the red one would run at 58mA and burn out.  And this is different colors from the same manufacturer, same series (Super Red superbright LED vs. Super Green superbright LED).  With each LED on a separate resistor, even if the resistors were the same ohmage both LEDs would work even if you neglected to look up their voltages.  Considering a lot of people here are guessing at the voltage their LEDs are based on whether they're white or colors and superbright or not, because they don't have any datasheets, none of those lights had better end up in parallel.  I once burned out an expensive jumbo-size LED by trying to put it in parallel with the same color, smaller package LEDs.

But it should take about 10 years for any of this to be an issue in practical application of a properly designed circuit with parts that aren't faulty to begin with.

This is the BYOAC forum, right?  

If you are the type that likes to poke around circuits with metal objects while the power is on,

This is the BYOAC forum, right?  

The resistor arrays you refer to are very handy and a good choice.  But they aren't very cheap.  You also can't get these from Radio Shack, so tack on a few bucks for shipping.

I'm speaking of these things:  (see picture)   I'm pretty sure they're not expensive, but not knowing exactly what they're called I couldn't check on Newark to see for sure.  I'm pretty certain the part I pulled one off of would not have had anything expensive on it.  

And if you really really want to save on wiring effort or price, just get lights that have the resistor inside.

[RandyT]

All LEDS in my units were of the same make and model.....
 
The resistor arrays themselves really aren't that expensive (~$.30), but will require shipping and possibly minimum orders unless you have a very well stocked electronics store nearby.

You also cannot "mix and match" LEDs if using the resistor arrays as the resistors are all of a common value.   That is unless all your LEDs have the same specs.

So anyone wanting to use multiple different LED's should stick to individual resistors of the proper value based on the LED used.

Will you buy that?

BTW, if everyone everywhere adhered strictly to what some instructor states can and can't be done, we probably wouldn't have half of the "cool" innovations brought about by mankind in the last 100 years.

And since this is "BYOAC" as you say, you might have just wanted to use a simple illuminated "push-on, push-off" switch for the shifted inputs circuit in the other thread you started, instead of the fancy soft-touch design that probably scared the crap out of everyone who isn't an EE (j/k)

RandyT

[grafixmonkey]

but it was so much fun to build!      

I know you know what you're talking about - I just didn't want everyone putting LEDs in their cab to be like "Hey cool I only have to buy one resistor" and end up with 30 LEDs wired in parallel across the front of their cab, and a blackened, smoking 1/4 watt resistor powering the half that didn't burn out.    

After staring at those datasheets for a while, I could sort of see how they might even out their voltages automatically, since they're not "ideal" diodes, and do have a small slope off of vertical...  but those things are so easy to blow.  I fried one trying to set it up to take current/voltage measurements for some real world data, because I accidentally plugged it into the wrong side of my pot.  Dumb mistake.        Friend turned 21 last night though, so I have an excuse for being tired.