Have you ever seen an LED-based headlamp that has selectable brightness? Sometimes the brightness change is from turning on more or fewer LEDs, but sometimes the individual LEDs seem to change brightness. How does this work? Usually when you reduce the voltage across an LED, it just goes out. We looked into this question using a Vernier Light Sensor and a LabQuest. You could do this with any visible light sensor and a computer or calculator interface.
The first clue was an observation: When dimmed, and when moved rapidly across the eye’s field of vision, the LED leaves a flash trail. When at full brightness, the eye just perceives a solid streak. Could the headlamp be built so that the LEDs are pulsed? This suggests an experiment: Measure the light intensity as a function of time for a dimmed LED light.
What would be a reasonable data-collection rate and duration? Well, we know from looking at the light that it looks more or less steady to the eye, so it must be doing something very quickly. A television refreshes the screen at about 30 Hz, and that sometimes appears to flicker. The LED doesn’t visibly flicker in standard use, so it must flash faster than that.
As a starting guess, then, how about taking data at 5000 Hz for a tenth of a second? If we need to change that later, we can. Here’s what we saw, after adjusting the time displayed to 0.02 seconds:
Aha! On high, the LED stays on; but on the mid-level setting, the LED is going on and off at some 400 Hz! What happens if we switch down the brightness a little more? Same parameters, new graph:
Here we see that the brightness is about the same when the LED is on, but that the LED is just not on as much of the time. In technical terms, the duty cycle is lower. When the LED is off, it is not consuming energy from the battery. We get less light, but the battery will last longer.
So, this headlamp manages to have a lowpower, battery-saving mode by flashing the LED on and off. Said another way, the device uses pulse-width modulation to alter the power. It would be hard to see or understand this behavior without data-collection tools being used to extend our senses.
Does the average power used by the headlamp scale with the duty cycle of the LED, or is there some significant overhead to perform the modulation?
How does the pulsing sleep light on some computers work?