One of my email correspondents, Paul, has built a baker’s dozen of the Joule thiefs with LED light strings, but all were white LEDs. I have a few strings of Xmas tree LED light strings, but all of them are multicolored, and are the type that run off a battery pack of four AA cells. I already knew when I bought these that they were a poor choice, because a co-worker had tried one and found that the batteries only last a few days to a week. Naturally, there are twenty LEDs in parallel and if you run them at ten milliamps apiece, they draw 200 mA and eat up a battery in no time flat. But I bought them knowing that I would abandon the batteries and use them on something else, like a power supply or solar panel. I also was curious as to how they managed to get the different colored LEDs to light up without the lower voltage ones hogging all the current from the higher voltage ones.
Paul’s string of LEDs has them also in parallel, but since they’re all white, there is no voltage mismatch, and they should all light about equally. To my eyes, my string looked like the blue LEDs were somewhat dimmer than the others, which would be expected since they are higher forward voltage than the other colors. This string uses 3mm LEDs, and I don’t see any bulges in the shrink tubing covering the wiring, which in my mind should be logical: resistors can be used to even out the LED brightness by reducing the current to the lower voltage LEDs. But the makers seem to be too cheap to add the resistors, so I’m not sure why the red and orange LEDs don’t hog all of the current. Maybe they’re special LEDs with a resistor inside of the case. I opened the battery box up and there is a single resistor inside, it’s low value but I don’t remember what the value is.
I poked around in the battery box and found the two contacts that powered the LEDs. I clipped them onto a Joule Thief that had been modified. I added a 100k logarithmic taper pot in series with the existing 1k resistor, so I could vary the brightness of the string. The transistor was a BC337-40, which is high gain and really requires a higher resistance than 1k. I powered the JT from a power supply set at 1.5V, and it drew about 200 mA with the pot set to its minimum. The LEDs were bright. I then adjusted the pot to give about 100 mA of supply current. Most of the LEDs were still bright but the blue ones were noticeably dimmer than the other colors. I increased the pot and the blue LEDs dimmed out completely, but the other colors were still lit.
This was what I expected, since I could not find any balancing resistors as I said above. I adjusted the pot back to a supply current of about 100 mA, and the LEDs looked okay, even though they were not all the same brightness. I removed the power and measured the resistance of the pot and resistor and they came to about 2000 ohms, so I unsoldered the wires, 1k and pot and put a 2k resistor in there. When I restored power , the LEDs were lit up the same as previously, and the supply current was about 130 mA (the power supply’s 1 amp analog meter is low resolution).