Help with 3 watt RGB LED

Thread Starter


Joined Sep 21, 2011
Hi! This is my first post.

I recently bought some high intensity RGB LEDs from this site. They are rated at 1W per color:

They are common anode (though they are mis-labelled as common catode, someone says it in the comments) so I soldered three resistors (47 ohm to red, 33 ohm to green and 33 ohm to blue, 2W resistors) to the cathodes. Connected 5V to anode, and ground to the three resistors. And voila, it works! They are very bright and nice.

The problem is that the resistors get very hot. After a minute you almost can't touch them. It scares me to use the LEDs continuosly...

The led itself doesn't get very hot, and I was planning to buy some heatsinks here:

But now, how am i supposed to cool the resistors?

The LED specifications, according to Dealextreme, are:

Red: 2.5V ~ 3.0V, 350mA
Green: 3.2V ~ 3.8V, 350mA
Blue: 3.2V ~ 3.8V, 350mA

I used 47 ohm for red and 33 ohm for green and blue, so it would give less than the 350 mA maximum for each color, so I wouldn't burn the LED.

With the resistors I used, it gives:

Red: (5V - 2.5V) / 47 ohm = 53 mA
Green: (5V - 3.2V) / 33 ohm = 53 mA
Blue: (5V - 3.2V) / 33 ohm = 54 mA

I'm well under the maximum of 350 mA... what am I doing wrong?

So my question is: is there a better way of turning on the LEDs, so nothing gets hot? Or am I using wrong resistors? They're rated at 2W so I thought they wouldn't get hot.

I'm going to drive the LEDs with a ULN2803 (darlington transistors) controlled by an AVR microcontroller (ATMEGA328P in a breadboard) and do PWM to control the color.

So that's all! Thanks in advance


Joined Jul 17, 2007
Well, the Vf of an LED will change with the amount of current flowing through it. If your LED has a Vf of 3.2v @ 350mA, it might only be 2.8v at 50mA. That would make your resistors dissipate a good bit more power than you were expecting.

Let's just say 2.8v / 33 Ohm; that's 84.848...mA, and 237mW power dissipation. Are you sure the resistors are rated for 2W? That's about 8.4x as much as it should be dissipating; so the resistors should really be pretty cool.

The ULN2003's will get mighty toasty if you have more than 1 channel pulling more than 300mA.
Have a look at this datasheet:
Take a look at figure 15, the graph on the upper left corner of page 11. With three outputs sinking 350mA, you can't be operating them at more than 50% duty cycle or the IC will overheat. In reality, you would want to stay lower than that, because that chart is MAXIMUM ratings, and you always want to stay well below maximums.

If you want a more robust driver, look at the ULN2064A or ULN2065A. They are more expensive, but they will sink up to 1.5A per channel.

You'll have another problem with Darlingtons, and that's the high saturation voltage. Take a look at Figure 17, bottom left of page 11. At 350mA, Vsat might vary anywhere from ~1.3v to ~1.7v. It will vary over temperature as well as from IC to IC. Makes trying to figure out current limiting resistors pretty tough.

You'd be better off using some logic-level N-ch power MOSFETs.
Last edited:


Joined Dec 20, 2007
Resistors are made to survive high heat.
2W in a 2W resistor makes it very very hot.
1W in a 2W resistor makes it hot.
0.5W in a 2W resistor makes it very warm.

A 47 ohm 5% resistor could be 44.7 ohms.
Your 5V supply could be 5.25V.
Measure them.

Thread Starter


Joined Sep 21, 2011
I'm totally stupid. No, really.

I have a power supply recycled from a router that gives 5 and 12V. I was connecting the LED to the 12V !!!! :p:p:D:D:D I should have labeled the outputs...

Fortunately I'm measuring 250 mA, wich is under the maximum of 350mA. I was wondering how bright the LED was at only about 80 mA :D

Of course this explains the heat.

Now I'm connecting to 5V and the LED is dimmer but still quite bright. It measures 65 mA on each channel and of course the resistors are a lot cooler.

Red: Vf = 5 - 0.0661A * 46.3 ohm = 1.93V
Green: Vf = 5 - 0.0667A * 33.1 ohm = 2.792V
Blue: Vf = 5 - 0.0666A * 33.2 ohm = 2.788V

A lot of thanks for the tips!

I'll use some TC4422 MOSFET drivers that I have around here...