Change of Plans

LED sensors have been (for the time being) ruled out of this project. I simply cannot get any of my LEDs to work properly as emitters and receivers. Most of my LEDs were poor sensors in the lightbox; the only set that was really "good" was the set of four RadioShack LEDs. Unfortunately, though, these LEDs were unable to sense the light that identical LEDs emitted. This was likely a result of the phosphor that is coated on the little diode itself (for more information, see the "White LEDs" section of the Introduction to LEDs page).

Why Don't LEDs Work?

That is a great question, perhaps one that I can't entirely answer. Based on two recent publications - "Very Low-Cost Sensing and Communication Using Bidirectional LEDs" (Dietz et al, 2003) and "An LED-only BRDF Measurement Device" (Ben-Ezra et al, 2008) - using LEDs as sensors has been confirmed to work. So why have I not gotten better results from my LEDs?

I am fairly certain that my methods of sensing are correct. I have duplicated the timer-based approach to LED sensing used by Dietz et al, which seems to produce moderately accurate results on a few of the LEDs I've tested (more information on the Timer-based Sensing page). In terms of direct voltage measurement, I've also reproduced the techniques used by others to measure voltage (see the Voltage-based Sensing page for more information). This gave particularly excellent results for the RadioShack LEDs.

Looking back to the Introduction to LEDs page, we can see the astonishing progress of LEDs in just the last 50 years. We started with crude, expensive LEDs that had an extremely low specific output (< 10 mcd)in the red range. Soon after we developed green LEDs, and (much later), blue. Today, we have LEDs that use significantly less power than early LEDs, cost only a few cents to manufacture, cover almost every part of the visible spectrum, and have extremely high outputs (> 20,000 mcd). They have thousands of applications, including traffic lights, automotive interior and exterior lighting, flashlights, architectural lighting, and a plethora of other uses. Due to the rising cost of power and increasing interest in being environmentally friendly, LEDs are starting to find more and more uses. On the development and manufacturing side, they are continually being modified to use less power and produce more light. Manufacturers are developing new processes to decrease production costs.

What I'm saying here is that LEDs are constantly changing, and the recent rise in demand for these cheap, efficient, durable, and flexible light sources is likely driving industry progress. It is possible that new technology and new manufacturing methods have rendered "new" LEDs to be especially poor sensors. As of late, little study of LEDs as sensors has been done; the most recent was from Ben-Ezra et al in 2008, with the LED-only BRDF measurement device that inspired this project. Though I have been unable to determine when any of my tested LEDs were manufactured, I feel that this may be contributing to the problems I've been having with some of the LEDs I've been buying. The best sensors of the bunch, from RadioShack, were likely old stock.

The first instance of LED sensing occurred in the 1971 by Forrest Mims III, designing a sunlight photometer using LEDs. I'd like to get some vintage diodes in the lab and test them before I end my research, just to see how they different from new diodes (both physically and in terms of sensing abilities). I picked up a set of two Western Electric LEDs (seen here) that probably date back to 1975 or so. They were used in telephones, replacing incandescent bulbs and allowing the whole phone to be powered from only the phone jack. They were among the first LEDs that were mass-produced in the US (a few years behind the Monsanto MV1 and MV2). When these LEDs arrive (likely next week), I'll be sure to post test samples for a set of LEDs that have likely not been used as sensors by anybody.