LEDs in series and forward voltages.

If you put the LEDs in series and the voltage across each LED is 3V, and the power supply is fixed and regulated, you won't have a problem. fixed voltage == fixed current.

Yes and that fixed current will be too much or too little and in real life nothing is regulated 100% so there is no such thing as fixed voltage and so no such thing as fixed current. And it all changes with temperature.

so in real life:-

YOU ALWAYS NEED A RESISTOR OR SOMETHING TO LIMIT THE CURRENT

Lol that's quite a run on sentence, there.

A typical 7XXX 3-terminal regulator does a pretty damn good job.
What I was basically trying to say there is that the equation he mentioned works most of the time, but breaks down when you get towards the supply voltage. He will have to experiment with the resistance, which will be pretty low.

You don't even need 20mA of current through the LED, it can be lower and even somewhat higher. I still say you don't HAVE TO put a resistor, but you don't even need 20mA either, so a small one won't hurt the performance much.

Not having a resistor is only dangerous if the supply voltage might change. With three terminal regulators like the 7812 (+12V), the chances of deregulation are low. They build a decent amount of protection into those things. However, a human "oops" could burn out the LEDs since the dynamic resistance of the LED decreases fairly rapidly with forward voltage.

Assuming the V-I curve of his specific LEDs say that at 3V, the current is 20mA, then the resistance of the LED is 150ohms at that point.

12V / (4*150ohms) = 20mA as well, obviously.

Again, I agree, a small resistor won't hurt. Without any other information, i'd try around 50 ohms.

Scot,
I don't think you have quite grasped the concept of resistors and non liner devices. Operating an LED on the knee of it's characteristics and using that as your current limit is at best hairy and will produce a non repeatable design.

Not having a resistor is only dangerous if the supply voltage might change.

Simply NO, there are other things that change. The knee characteristic of a LED varies from device to device, manufacturer to manufacturer, LED technology to LED technology and it varies with temperature.

So you will end up with a circuit that will only work with the components you played about with at the temperature you played with them. So it could work fine at 20C but then it gets hot and you are operating at 35C and bang you burn out your LED. Look up thermal runaway.

Again, I agree, a small resistor won't hurt.

The smaller the resistor is the less effective it will be in curbing the variations in the LED characteristics. The smaller the resistor the less useful it is in regulating the current. When the forward voltage drop of the LEDs approaches that of the power supply then it is time to think of other current limiting measures like a constant current drive. It is not a time for thinking of abandoning all forms of current control. This is why high power LEDs are not usefully controlled with a series resistor.

Sorry to stamp on you so hard but it is important that beginners don't go on to think that current limiting components on LED systems are something of an option. It is also important that you don't think that as well.

Yeah, those are some things I wasn't thinking about.

The "iffyness" of the specific situation is why I said he'd have to do some testing to get the right current. I did not consider a constant current source for this application, however. Seems overkill, though not a bad idea.

A large resistor will hurt the performance of the LEDs (if thats even much of a concern to him, I don't know)

Basically, my posts were supposed to by hypothetical in nature to show him why his formula doesn't work. Especially to show that LEDs do not follow Ohm's law (though nonlinear, they are approximately linear. Linear does not mean V = I * R, it could mean V = k(I*R) where k is some constant) I can see how my posts would mislead someone now.

Thanks for livening up my day, my current job is boring.

I'm really frickin' tired this morning.

Many thanks for all your replies. However a question still remains. When working out the resistor to limit the curent do I use the upper or lower limit of the forward voltage in the equation? The datasheet gives the forward voltage as anywhere between 3 volts and 3.6 volts. If I use 3 volts, since the supply voltage is 12 volts and the LED current 20mA I get ((12-(3 LEDs x 3 volts))/0.02=150 Ohms. But if the forward voltage is taken to be 3.6 volts I get ((12-(3 LEDs x 3.6 volts))/0.02=60 Ohms.

Use 150 ohms, it's a standard value. It's safer to use the smaller number for forward voltage because if you assume 3.6V and your LEDs happen to all be 3.0V at 20mA then the following might happen:

60ohm resistor chosen, assuming 3.6V across each LED.

Iled = Vres / Rres
Vres = 3.0V = 12V - 3*3.0V
Rres = 60ohms

Iled = 3.0V / 60ohms = 50mA

You're screwed in this "worse-case". In reality it can be worse because of what was talked about before.

Ignoring all this crap me and Mike have been throwing at each other, if you're using the 3 LEDs, you're pretty safe with 150 ohms. It becomes difficult with 4 LEDs, and you won't be able to get 20mA most likely. It would have to be less.

Actually I'm not that bothered if each LED gets less than 20mA. I already have a 12 volt power supply so I'd like to use if possible, but I've already built a test board with sets of four LEDs in series, so it seems I have three options.

First off, I could keep the power supply I've already got, but re-design/re-build the boards to include resistors.

The second option is to get a different power supply (15 volt laptop psu?) and some resistors to my boards.

The third option would be to keep both the boards and power supply, and find some way of limiting the current.

I'd like to go with either the second or preferably the third option. Its easy enough to add a few resistors to my boards though I'd rather not have to redsign them from scratch, which is what the first option would entail.

If I wanted to keep my power supply, and keep my LEDs in sets of four what is the maximum current I could reasonably expect to give them? I'm guessing that running them at less than their maximum current will give them an easier/longer life.

keep it hovering around 20mA, use a potentiometer if you have one to adjust the series resistance you need, then find the resistor closest to that as you can.

I'm guessing that running them at less than their maximum current will give them an easier/longer life.

Yes I normally run LEDs at 10mA as the extra 10mA seems only to contribute to glare rather than to actual brightness.
Any of those three options would do although I would suspect that options one and two are the simpler.

As to your original question as to what limit you go for, when designing something go for the half way situation for the design but check that it is within limits at either extreme.
So calculate your resistor value at 3.3V and check that when using that resistor the current is not exceeded or too low at 3.0V and 3.6V.

LEDs are current devices, not voltage. This seems to confuse a lot of people.

In theory if you connected 3.4 voltages to an LED with a Vf for 3.3, you'd fry the LED. The only reason it works at all is in practice there's internal resistances in everything and your power supply can't provide unlimited current to maintain 3.4 volts. But relying on imperfections in your components to make everything work is beyond poor engineering.

Please, check your facts before sprouting what's amounting to myths.

If Vs is below, or equal to Vf, you DO NOT need a resistor! Contrary to what most seem to think, there IS internal resistance in the LED after Vf, about 17ohms in our everyday 25mA devices.

Another fact about the LED junction is that it does not demonstrate the characteristics of a Si junction, so there is little chance of 'thermal runaway' or 'current hogging' so many are saying like it's the gospel.

So if you're right, I'm not an idiot?

I agree with your arguments.

I'm just awfully tired of hearing so-called experts tell everyone that the resistor is such a life saver. I mean it's such a simple thing to prove, or disprove:

Take a power-supply with an accurate V and A meter and hook it directly to a little LED, with big heavy wires so external R is minimized. Crank it until you see 20mA, now push the V another 10% higher, the current goes up, probably to around 40mA - which coincidentally shows that there IS internal resistance involved.

We'll get a warmer LED, for sure, but its not the armageddon they all claim it to be! No flames, no explosion, and no avalanche current-sucking catastrophe either!

As most experimenters who worked with multi-watt LEDs know, heat is the #1 enemy of the LED, and as long as the heat can be dissipated, things will be, er, cool.

Besides, we're all talking Green the same breath we say LED, but burning up the power savings in resistors? What are we thinking?

Sadly the internet is full of idiots like you two.

I taught Physics in a University for 21 years and I now design electronic circuits so I do know a thing or two about solid state design and designing reliable circuits.
It is all too easy to throw something together that works but not every time.

Besides, we're all talking Green the same breath we say LED, but burning up the power savings in resistors

Never herd anything so stupid on this forum. Pity you missed your vocation, four years ago you could have been President.

Funny thread :wink:

By the way, as you talked about constant-current sources I remembered there was an easy way to build one (I know they're not that expensive, but I like to build stuff myself...).
The wise old man in my electronics store told me about it, when I was buying an LM317T (adjustable Voltage regulator) and some TL431 (adjustable shunt regulator), but I don't remember which one was the one capable of doing so...
Diving through the datasheets of both of them I keep reading things like "constant current sink" and "current regulator" with their application diagrams, but unfortunately my capability of understanding datasheets is still very poor... which btw for me (as a german) is like reading Shakespeare in original when I was in school: spend a year reading it over and over again and you start to realize :wink:

Well, i would be very pleased to have someone of you point me in the right direction :slight_smile:

Well, i would be very pleased to have someone of you point me in the right direction :slight_smile:

try this:-

^^^

Ha-ha, oh wow.

Oh wow, indeed.

I remembered there was an easy way to build one (I know they're not that expensive, but I like to build stuff myself...)

The one I use is with the LM317L (TO92 size) and a resistor:

A value of 68R will give you 18mA, 100R for 12mA. Good from 5mA to 100mA.

Hmmm.... some interesting ideas, but I think I'll re-design my LED arrays to have 3 in series with a resistor to limit the current. Many thanks for everyones help with this.