LED Pulse Handling Capability

Hello folks,

I've got a bunch of RGB Leds that I've purchased, and I'm planning on pulse width modulating the colours while multiplexing between 5 of them. So each LED will have a time slice during which the colours are PWM'd for a couple of cycles before moving on to the next LED.

I've tested this on the breadboard, and it seems to work quite well. My question, however, relates to the forward current that the LEDs can handle. Each colour is rated at 20mA constant current, but since I am pulsing them, I should be able to run them at a higher rate.

I've attached a chart from the LED data sheet. In my display, I plan on having at most 5 LEDs on at the same time, so if I set each colour to full (ie 100% pwm), then my duty cycle is 1/5 (or 0.2) and according to the attached chart, I should be able to run them at a peak current of about 200mA if my pulse is on for 100uS or less. I'll likely play it safe and try closer to 100mA or lower for 100uS pulses, lowering the current until it no longer is bright enough for my purposes.

Am I interpreting this chart correctly?

It looks correct to me.

I am not exactly sure what you are saying here. On the chart if you follow the 100mA line to the 0.2 duty cycle trace and then down to the bottom you get a tp of 0.007S.

Since D=tp/T, T=tp/D, and F=D/tp so at 100mA you can run at a maximum duty cycle of 0.2 (tp=<7mS) at a minimum of 0.2/0.007=30Hz (T=35m)

You'll want to note that LEDs have an efficiency curve as well. While pulsing an LED will give you an average light value over time that average will not make it as efficient as the LED being run at the proper current. If this is a line powered application probably not a big deal as long as you have the heat dissipation handled, but if it's battery powered you could get dramatically less battery life pulsing an LED like that. Doesn't do great things for the life of the LED either because the diode junction is pulsed at a very high temperature for a very short period of time, this is far worse than the diode being run at it's rated average current over a long period of time. Since you're trying to pulse it near it's physical limits you're going to have to pay extra special attention to the current limiting, because even a slight blip in power at these ratings will instantly fry the diode.

Sceadwian, that is definately a good point. My application will be battery powered so I will have to check efficiency carefully. The datasheet doesnt seem to go into much detail about efficiency vs frequency.

Ubergeek; What I was saying is that if I take a Tp of 100uS (10kHz) the rated current is around 200mA for a 0.2 duty cycle. So, to play it safe, I can try and run them at half the rated current (or lower), so only 100mA at 10kHz and it should be a bit safer because I'm running it farther from the limits.

Frequency will not be mentioned in the data sheet as far as efficiency goes because it is unrelated. LED efficiency is related to one thing and one thing ONLY, the current going through it. Don't peak them so high it's horrible in every respect as I mentioned. Are you size constrained? You could use pulse stretchers to increase the duty cycle of each sub pulse to the diode. Do a google search for Mickey Mouse Logic. Here's the link it will take you to.
**broken link removed**
The circuits use only simple Schmitt triggered inverters capacitors diodes and resistors for a wide range of digital and logic solutions. Adding a pulse stretcher to 5 LEDS may be as simple as a couple of hex inverters and a handful of discrete components.