Efficient way of driving 72 watts of LED awesomeness?

Thread Starter

andrewortman

Joined Jul 4, 2009
7
Hello there,

I'm enduring on a project, and it might be (it is) crazy. My goal is to make a superbright (stadium like) multicolor lighting for an outdoor concert my friends are having. I want to power 24 3w (3 x 1w) RGB LEDs individually from a microcomputer. I am utilizing PWM to dim each bulb. Most of the time, these things will only be running on 20-100mA current at the brightest inside. However, I want to be able to flash some of the bulbs to full power for a split second or move it outside for this concert and cover a very large area with this colored light. Actually, I just want to overdo it ;)

This thing will be very, very, very bright (which is what I want) - however my search for the right driver for each channel is starting to kill me.

I know I dont need all 350mA for something so bright it can burn my retinas in seconds, but I still want all 350mA for each channel. I have a large heatsink for all of the LEDS (a nice piece of half inch alluminum they are thermal pasted to) - so the heat from those aren't an issue right now.

I'm having problems selecting the right constant current driver IC for my circuit. I have a 5V, 150W power supply and 24 of these fun things:
http://www.dealextreme.com/details.dx/sku.4530

If I use a regular linear constant current source with my 5V supply - or carefully calculated resistor values - I'm going to get plenty of heat (about 2.5 watts per LED!) - which is extremely inefficient

Here's another problem: Money! I'm a college student, and my budget for this project is < USD$400

My goal was to make each LED driver < $3 in components, so I have some money left over to make a nice frame and PCB. I can do higher, but I may cringe.

Can anyone recommend a highly efficient way of driving these monsters? Does anyone know of an IC for what I need? I have already taken Circuits I, II, and taking Electronics I at my college in the first semester of my Junior year studying Computer Engineering. Unfortunately, thermodynamics isn't in my course load anytime soon.

Thanks for your help!
Andrew
 

rjenkins

Joined Nov 6, 2005
1,013
Hi,
those LEDs are at an incredibly good price, I'll be getting some of those myself!

A basic series resistor + transistor switch will be fine - the leds are rated for around 3.5V drop, so you will only be dropping the other 1.5V across the control circuit (and wiring).

This gives you a maximum heat output from the control circuit of .35 (the current) x 1.5 (voltage drop) = 0.525W per circuit at maximum output.

(Each actual LED will be dissipating something over twice as much heat as the switch circuit, but that is required power regardless of the drive circuit or method).

If you drive these with a PWM signal, the dissipation will be proportional to the duty cycle, so heat at 20mA = 30mW and at 100mA = 150mW

For 350mA with a drop of 1.4V (assuming 0.1V drop on a saturated transistor), the calculated resistor is 4 Ohms.

At these resistance levels, the wiring resistance may have significant effects - if the leds are remote from the controller, measure or look up the resistance for the length & type of cable you are using.

I'd start with 3.9 Ohm 1W, that should be pretty close - I can't imagine less than 0.1 Ohm somewhere in the wiring.

The PWM frequency should be at least 150Hz, preferably higher (up to 1KHz or so).

The transistors should be rated at least 500mA, preferably a bit higher.
If using bipolar parts, set the drive resistor to guarantee saturation at 500mA load.
If you use logic-level mosfets, you can direct drive them from a micro or CMOS latch, just put a 100K or 1M resistor between Gate & Source to prevent the gate floating if the input is disconnected.

Edit: Just noticed the Red elements have a lower voltage drop, typical voltage 2.75
For these, the control dissipation will be about 0.8W at full power, the calculated series resistor should be 6.14 Ohms
If you use the same 3.9 Ohm 1W in series with a 2.2 Ohm 1W, that would give you 6.1 Ohms; near enough.

So final values:
Red 3.9 + 2.2 Ohms
Green 3.9 Ohms
Blue 3.9 Ohms
 
Last edited:

rjenkins

Joined Nov 6, 2005
1,013
All the comments on the site say they work well, but may have incorrect terminal markings.

I hope they are good, I've ordered a batch!
 

Thread Starter

andrewortman

Joined Jul 4, 2009
7
Sorry for the late response.

Yea, these lights are awesome. The polarity is backwards, but they really do give out 3W of white light. They are comparable to the $12 3W leds ;) The only problem is that they ship from china, and it took 3 weeks to get to my house.

I think I may stick to the resistor method (with a NFET) for each channel. I tried putting together the circuit in the instructables site in the first post, but I kept getting bad current changes depending on the voltage drop. It's weird.

Any other suggestions greatly appreciated as I may be committing electronics suicide.

-Andrew
 
Top