Repairing The Blue Canary In The Outlet By The Light Switch


[Vince] and his wife are big fans of [They Might be Giants], so when they were perusing their local Target one evening and stumbled upon a blue canary nightlight, they bought it immediately. While the nightlight was easy for his toddler to use, the LEDs inside started to dim after about a month, and eventually they started flickering like mad as you can see in the video below. A battery swap didn’t remedy the problem, and instead of returning it, [Vince] decided to try fixing it himself.

After poring over the device’s simple circuit, he couldn’t figure out any reason why the nightlight would start behaving like it did. He did notice that a resistor was left out of the device, likely as a cost-cutting measure, so he added one in before replacing both of the nightlight’s LEDs.

With his simple tweak, the nightlight was better than new, saving him from what would likely be a string of annoying merchandise exchanges.


11 thoughts on “Repairing The Blue Canary In The Outlet By The Light Switch

  1. Nice fix.
    If Vince wants more diffuse leds he should file them down. You can file off most of the focusing dome without coming anywhere near the actual electrodes and ruining the device. If you leave it rough filed/sanded it’s fairly diffuse, or you can polish the surface back up and get a none focused, none diffuse, illumination.

  2. “Weird. How could this ever result in flashing? I know having LEDs in parallel like that is a bad idea.”

    I know why… The resistor is too small. the LED’s have been running in overcurrent and when one damaged before the other it started to cause the semiconductor to do strange things… This made the flicker.

    Also it’s a GOOD IDEA to run led’s that way if you know what you are doing. 99.9978% of all china engineers at small toy companies do NOT know what they are doing. Series or parallel led wiring is just fine.

    There was more than 20ma running through each LED. this is out of spec for most blue LED’s. Plus I will also bet you those were china garbage led’s that typically have a very short lifespan because they are seconds or even thirds that have never been tested, burned in, or failed QC.

  3. The resistor most likely had nothing to do with it (and cost less than a cent in volume) – one of the LEDs was bad (and wired in series?), have seen exactly the same effect first hand

  4. Tom, LEDs flash when exposed to high temperatures. Think of it as the P/N junction breaks down and basically the light you see is arcing. I’ve seen it several times when I’ve overheated during soldering. It starts off slowly and then happens more frequently

    I did the math on this… I would expect those LEDs to be standard blue LEDs with a 3.7v forward bias and a 0.02A maximum power requirement. Together their max would be 40mA total or 0.04A.

    From the circuit, 4.5V with 3.7v chopped off due to Forward Bias from the LEDs leaves .8V of current carying capabilities. .8V dropped over a 30 ohm resistor leaves 0.026A or 26mA So we’re lower then the total requirement by a single LED….

    However.. If a power surge happened, or if one of those LEDs flickered JUST ONCE, the other LED would be overdriven to 26mA which is well over the standard of 0.20mA for illumination style LEDs. This leads to overheating and then eventually flickering and/or premature death.

    Caution High Ohmage…. The 510 Ohm resistors which were replaced in this circuit are entirely too high of ohms. According to my calculation you’re driving SuperbrightLEDs at 1mA instead of 20mA

    If anyone would like to clarify on flickering… I’d like to hear it.

    1. There isn’t really any hard limit on how much current you can put through a LED junction. The limitation is keeping the junction temperature below roughly 125 degrees C.

      That’s why it’s possible to use the stroboscopic effect to drive LEDs at higher than normal current at a duty cycle that allows them to cool off, to make them appear brighter.

      So if you can keep the LEDs cold, you can push any amount of current you want, up until electromigration starts to ruin the junction. Conversely, if you want to install your LEDs in a sauna, you have to drop the current significantly to stop them from burning out.

      26 mA through a standard LED really won’t do anything to it. It’ll just burn brighter.

    2. Is the ohmage being high going to damage the LEDs? Or do you just mean you won’t be getting peak brightness?

      I didn’t really worry about peak brightness; it is a nighlight after all. In fact I think it runs a little too bright.

      It’s sort of frustrating. I have an EE degree, and while I had to take multiple semesters of semiconductors courses, I don’t think we covered LEDs at all. I learned more about LEDs in an intro quantum physics class I took than anything from the EE department.

  5. For the sake of general information: white LEDs fade over time.

    A white LED is actually a blue LED coated in a phosphor that absorbs blue light and emits yellow. In additive color mixing, yellow is red plus green, and adding blue to yellow produces white.

    Thing is, the phosphors break down over time, partially due to heat, and partially due to bleaching from the blue LED light.

    LEDs also degrade over time, for several reasons. Electromigration pushes defects and impurities into the PN junction, and can actually move the dopant atoms around. Heating the chip makes the problems worse.

    In most LEDs, the chip uses the lead frame as a heatsink, and blue LEDs need a bigger heatsink than red or green ones. Thermal stress can break down the bond between the chip and the lead frame, making it just that much easier for the chip to overheat.

    The solutions to all those problems are expensive, and if you aren’t too concerned about ethics, you can save a lot of money using lower quality silicon wafers, lower quality doping, lead frames that are too small to provide adequate heat sinking, and cheap phosphors. You’ll get something that lights up just fine the first few times, but starts to decay in a matter of weeks.

    The moral is: cheap LEDs you buy from some off-name vendor in China may not give you the same service lifetime as devices from vendors like Cree and Dialight.

  6. Had this happen to me, in my case it was a load of cheap noname LEDs I made into a table lamp.

    Flickering, whole sections turning on and off, etc. In the end it got scrapped because there was no way to get to them once soldered and glued in place.

    Its well worth testing a sample if you intend to buy a lot of them, a good test is to drive Imax*1.5 mA through them for 20 minutes on 1 hour off for 48 hours and measure the “before” and “after” brightness.
    If less than 3% power outout drop they are likely to be fine for general use.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.