04 December 2018

Studying a Hallowe'en LED flasher

A friend kindly gave us some flashing-light stickers for Hallowe'en, and the natural thing to do with them once the festivities were over was to dissect them. So we did this, and found that each contained a square circuit board about 25mm on a side. When you press a little button in the middle, three coloured LEDs — red, green, blue — flash rapidly for about twenty seconds.

It looked like they were flashing round in a circle, but too quickly for me to be sure of the sequence. To investigate, I set up the flasher approximately vertically, held in a lump of Blu-tack:

Then I turned off the lights, pressed the flasher's button, and took some pictures with a ¼-second exposure while rotating the camera anti-clockwise about a vertical axis. This had the same effect as moving the flasher from left to right across the field of view, but was easier to do. The resulting photo is effectively a graph of when each LED is on, with time on the horizontal axis.

Because the three LEDs were in a triangle, I had to do this three times, once each for each pair of LEDs being aligned vertically one above the other. It took a few tries to get a good trace of the lights in the three orientations, but it did work:

Comparing red (top) and green (bottom) traces (ignore blue traces as the blue LED is off to the left):

The red LED turns on pretty much at the same time as the green one turns off. Then when the red LED turns off, there is a period when neither the green nor the red is on. (Presumably the blue one is one during that time, but we can't tell that from this trace because of the offset of the blue LED.)

Comparing green (top) and blue (bottom) traces (ignore red traces as the red LED is off to the left):

The green LED turns on as the blue one turns off. When the green turns off, neither blue nor green is on. (Presumably the red is on during this time.)

Comparing blue (top) and red (bottom) traces (ignore green traces as the green LED is off to the left):

The blue LED turns on as the red one turns off. When blue turns off, neither is on, but presumably green is.

Putting this all together, we can be fairly sure that the sequence is green, red, blue, green, red, blue, etc. I.e., the cycle is anti-clockwise. Also, there is a consistent measurement of very close to four cycles of the pattern during the ¼-second exposure, so we can estimate that the flashing is happening at about 16Hz.

Extra study

Thanks to Oliver Nash offering the use of his iPhone with its 240fps video capture ability, I was able to confirm the above. YouTube has the actual video we took, cropped to the section of relevance, but the following GIF recreation gives a pretty good idea of what the flasher looks like at ⅛ speed:

The sequence matches what I got originally, and the estimate of 16Hz is supported by the video too — counting cycles gives an estimate of just over 14Hz.

Future work

Oliver also pointed out that there would be interesting things you could do if you mounted one of the flashers on a motor. If you got everything lined up properly, then at various speeds of rotation you would get different effects. I think you ought to be able to get all three colours mixing, giving the effect of a white light. I initially thought you might be able to mix pairs of colours, but now I'm not so sure. I'm unlikely to get round to this, but it would be cool!

Comments are closed.