This is a story about a successful system that nevertheless failed to make the cut. An experimental LED brightness adjustment is something [Mitxela] explored in a project for a high-precision clock; one that shows time down to the nearest millisecond, and won’t flicker or otherwise look weird when photographed with a high-speed camera. To pull this off means reinventing many things about a clock display, including how to handle brightness adjustment elegantly. Now, to be clear, the brightness adjustment idea described here is something that did not end up being used, but it’s interesting enough that [Mitxela] wrote it up and we’re very glad he did.
The idea was to have a smooth and seamless automatic brightness adjustment, ideally with no added components. Since LEDs can be used as light sensors, [Mitxela] saw an opportunity to use elements of the clock displays themselves as sensors. This is how it works: a charge in the p-n junction that makes up an LED will decay at a rate proportional to the amount of light hitting the junction. By measuring the speed of this decay, it’s therefore possible to tell how much light is hitting the LED. It’s effective and elegant, but there are a few practical issues to deal with.
The first failed idea was to employ as sensors the unused decimal points in the seven-segment LED modules, but that turned out to have issues. One was the common-cathode wiring of the display modules; this makes them very convenient to drive as displays, but made using the decimal point as a light sensor impractical. The other issue was that the built-in diffuser that makes the displays easier to read absorbs a lot of ambient light. A much better option was to use the LEDs in the colon separators between digits, since they’re independent. Naturally they still have to light up in addition to being used as sensors, but [Mitxela] made a successful prototype by performing the necessary measurements in between the LEDs being driven by PWM.
Despite how clever and efficient the solution was, in the end what sank it was the fact that the LEDs just don’t do a very good job of sensing ambient light for this purpose. The LEDs are simply too directional. Even after sanding away the top (lens) part of the LEDs, they still had a very narrow field of view. As [Mitxela] describes it, tilting the clock towards the ceiling could send it to full brightness, and the shadow of one’s head falling across the clock would plummet it into “night mode” dimness. In short, it responded to what was directly in front of it, rather than the ambient light level as a whole.
It’s a reminder that sometimes a solution simply won’t tick all the right boxes, and it can happen for unexpected reasons. Still, LEDs are versatile things. Not only can they sense light, but as the name implies they’re also diodes. As diodes can be used as temperature sensors that means LEDs can as well.