Hi everyone,

I am a circuits newbie, currently trying to build a circuit to drive several LEDs at a constant current. I have seen several sites suggest using a current mirror, but as far as I can see this can only be used when you want to drive all the LEDs at the same current. I want to be able to drive each individual LED stably at a different current, and be able to choose this current as precisely as possible. I was just wondering if anyone could suggest the most efficient way of doing this?
I would really appreciate any pointers in the right direction!

 

Well, "most efficiently" will involve a switching-type supply that involves an inductor to keep the current flow relatively constant, and a method to constantly sense/adjust the flow of current.

What is your application? Do you have some specifics in mind?
What are the specifications for your LEDs? (Vf @ current)

 

Why would be the first question I think .... but that would depend on what you mean by efficiant?

If you mean electricaly efficiant then you need to be loooking at some sort of chopping supply for each LED, because that will not disipate much heat.

If you mean efficiant as in small number of componants and simplicity of build then a linear regulator and current limmit resistors is probably the way to go.

An LM317 and a single resistor will give you a current limit, look at the data sheet.
http://pdf1.alldatasheet.com/datasheet-pdf/view/11662/ONSEMI/LM317.html
You could also go with a regulated voltage feeding all the LED's in paralell with current limmiting resistors .... you could then dim them with Pulse Width Modulation as opposed to altering the supply charictoristics.

If you were to use a voltage regulator set up as a current limit for each LED you could turn that on and off with its sensing pin thus using it to manage the on current as well as switching the LED with PWM to achieve dimming.
And, if you want the whole thing to be analouge, replace the curent sensing resistor with some other feedback circuit, if you search ther are loads of alternativs, and you will be able to build a voltage controlled current source.

Its all about what is goint to work for you, there are so many diferent ways to go you will have to flesh out your request considderably to get a qualified answer.

Apart from managing brightness what are your requirements, low power, low cost, simplicity?
The list go's on......
Hope this gives you somewhere to strat looking.
Al

 

Sorry Sgt .... posts crossed.
By the way if you have realy big LED's a 350T is pin compatable with the 317 and good for 4A with a heat sink
Al

 

Thanks for your quick responses!

To answer some of your questions, I have currently built 6 separate circuits with a power resistor, high brightness LED (Luxeon Rebel Color range, Vf can vary from approx 2.5 - 3.5 V) and a DC power source so that they can be tested for brightness at different currents up to roughly 400mA. I will be testing the circuits to see which current for which LED will give the required luminosity for that particular colour.

So when I find out which currents I need, I will need to build a circuit that will allow me to combine these 6 circuits into one that requires less components. Since I am pretty new to building circuits I think I should probably go with the simplest option first, possibly the one with least components as eventually I will make an array of several of these combined circuits.

So then each LED will be running at the specified current (which should be as precise as possible). To be able to reduce and increase the current of the circuit so that the currents of the different LEDs remain in the same ratio (same spectrum, lower amplitude) might be difficult, but I like your idea of switching with PWM, as this could be used to dim the lights instead of having to lower or raise the current.

As to what sort of efficiency I am looking for, I think I am possibly looking for low power (I have a limited DC voltage supply), but extremely high efficiency is not the ultimate goal. As long as each LED is running at the correct current that should be enough as the first step. A constant current for a group of LEDs that run on the same current is not too difficult to find on the web:

http://www.instructables.com/id/Power-LED-s---simplest-light-with-constant-current/

But if I had to do this for every different colour LED that required a different current it would make for a rather bulky circuit, so really first I am just looking for a method that has less components, but still has a way of regulating the current so that it will not destroy the LEDs if I do something stupid which is very likely.
Hopefully that makes some sort of sense!

 

The instructible you linked to has a constant current regulator that consists of only 4 components; two resistors, a transistor and a MOSFET.

It's a linear regulator, and it will not be very efficient; a good deal of power will be consumed in the MOSFET and the current sense resistor; the greater the difference in the supply voltage vs the sum of the Vf's of the LEDs, the worse the power efficiency will be.

Without knowing more details about your power supply, it's going to be pretty difficult to make suggestions.

For testing, a circuit similar to the Instructibles schematic could be used; but instead of the transistor an opamp with adjustable feedback path would be needed.

 

Currently I am using a DC power supply (I think it has the range 0-35V) to test the simple circuits (single power resister and LED), but AC power supply could be used with a recifier and buck converter if necessary, and if this is not overcomplicating matters!

 

I would have jumped here sooner, but I've been flat on my back with the flu.

Current mirrors depend on the transistors being the same temperature, so they are not a good option. You might be able to find an array that would do it, but I wouldn't use them.

While inefficient, you can still use transistors. Power transistors are designed to be heatsinked, which is what you would need to do. Something like this would work...

If this is an experiment, and you have a variable DC power supply you could go with this...

We have had people pop their LEDs using the current feature of the power supply. If you try using this feature short the power supply, set the current, connect the LED, then remove the short. The power supply capacitors will surge the LED otherwise, with bad results.

 

Thanks Bill!
The second circuit is exactly what I have now on a heat sink. The first looks like a good way of combining the different LEDs onto one board, but will the LEDs on each branch have the same current? I am going to only be running one LED on each branch each with a different current (may be extended to more later in project), so I'm just wondering how you would alter the current in each branch to be the specified current for that colour? I'm sure it won't be as easy as sticking a resistor in somewhere, apologies for my lack of circuit knowledge! Would I be right to assume the transistors are MOSFETs and the diodes on the left are just normal diodes (not zener)?

 

The transistors are old bipolar junction types, something like the TIP100/TIP101/TIP102 series. They are pretty common, but the parts are noncritical. Two transistors in this configuration are called Darlingtons, it is common enough they sell them prepacked that way.

One of the main advantages of this design is most heat is dissipated by the transistors, though the 0.75Ω needs to be at least 1W. You could put qty 6 4.7Ω ¼W resistors in parallel, add a 7th resistor, 18Ω ¼W, and you have the same part (or you can buy the part as described).

CR1-3 are just plain jane diodes, very noncritical.

The transistors, as stated, will get very hot. Heat sinking is mandatory.