Nonfailing light-bulb-string story causes readers to think

-- EDN, 2/17/2000

Our Leading Edge item on how series-wired Christmas lights self-heal prompted some interesting reader responses suggesting alternative ways that a failing bulb does not cause an entire string of bulbs to fail ("Simple scheme saves string-light situation," EDN, Dec 23, 1999, pg 24). Apparently, some light strings use another mechanism, and we have some hands-on evidence to confirm this fact. This approach involves no software, no microprocessors, no logic gates, no nanosecond timing—just some passive light bulbs and clever materials engineering.

Check out what the following people have to say:

From Stacey Jarvin, signguy@islandnet.com:

With reference to your note in the Dec 23 edition about the series-string Christmas lights: I, too, became interested in how they work, and, unless yours have a different technology, I think you missed the point. If you break one of these open and carefully break the filament, you will not find that parallel shunt resistance. If you measure the resistance of one of the burnt-out ones, you find it is less than a good bulb, virtually zero, which doesn't fit with your theory.

The way I see it working is this: The two support wires have a coating on them as you suggest, but it is not resistive; it is insulating but thin. Around the two filament wires is wound a number of turns of uninsulated fine wire. As long as the string of bulbs is working, each bulb has 2 or 3V across it, and all is well. When a filament burns out, no current flows in the series string, there are no voltage drops, and the full 110V appears across that open bulb. This ruptures the thin insulation, and the two filament posts short out through the wire and spot weld the whole thing into a short.

Measurements seem to bear this theory out. Once in a while, the spot welding doesn't work, and, when the bulb goes, the whole string goes out, due to the insulating layer still working. A novice will "frantically search for that one bulb that's the culprit" assuming that there is only one burned out. This can become very frustrating because of another effect.

As each one burns out, a slightly higher voltage is placed across each of the other bulbs, reducing their life. As further ones burn out, this effect accelerates, and you can see a whole string go in a couple minutes—the later ones like flash bulbs. Until, of course, the resulting higher current fuses something in the circuit and breaks the circuit. Thus, there are many burnt-out bulbs in the string, and replacing one at a time is a hopeless task. And, if you do find the one that eventually stopped the chain reaction, the replacement lasts only a few seconds, unless you have replaced all of the others also.

Because five replacement bulbs are sold for about the same price as a whole new string, there is no real point in trying to repair one of these chain-reaction burnouts.

Moral: Replace your burnt bulbs as soon as possible—just like they tell you to in the instructions.Now, having experienced all this trouble getting a string working, I would like you to do an article on the factory that turns out these things, revealing how it does it so cheaply and with the whole string working!!!

From Jon Titus, Editor in Chief, Test & Measurement World:

I ran a few experiments on series-string lights, and I think that Stacey Jarvin may be correct. The shunt wire may "spot weld" itself to form a shunt across a bad bulb. Here's what I did:

I carefully broke the glass on three replacement bulbs so as to maintain the integrity of the filament. Prior to breaking the glass, I checked each bulb to ensure it worked properly in the string.

1. First bulb inserted in the string with intact but exposed filament. The string works fine. I expected the filament to go "poof," and then I would observe the results. A closer look showed that the filament had heated, but it was still intact. Small traces of white (oxidation?), but mainly blue discoloration in the middle section of the filament.

2. Second bulb inserted with intact but exposed filament. The string works fine again. The filament looks intact, but discolored as above. Used a toothpick to break the filament in the hot circuit while I observed any action on the shunt wire. As I broke the filament, I saw small sparks at the small shunt wire. Increased current through the shunt may have caused more "welding" across resistive connections.

3. Third bulb inserted with a purposely severed filament. As soon as I inserted this bulb in the string I observed the small sparks at the shunt wire. The bulb completed the circuit, and the bulbs on the string all lit.

So, I'd say the small wire does act like a shunt, getting welded due to the initial high potential. That's neat. It must take some interesting QC to ensure a uniform coating on the wire. Or perhaps the surface is just oxidized slightly.

An interesting effect was noticed in bulb #1. Even though exposed to air, the filament didn't heat up and break. Perhaps the filament started to overheat and oxidize, thus raising its resistance. As the resistance increases, the potential across the bulb increased enough to cause the shunt to weld itself in place. Thus the filament fails but does not necessarily break. In years past, I have looked at several series of dead lights with a magnifying glass to try to find a broken filament. In most cases, I could not find such a bulb. This effect may be the reason. As the bulb starts to fail, the shunt kicks in so quickly that the filament never has to break. Some clever engineering in a $3 string of cheap lights.

From Dan Strassberg, Senior Technical Editor, EDN:

There are some other interesting wrinkles to this. What happens if, when you plug in the string, multiple bulbs are missing from it or have open filaments but unwelded shunts? This situation could occur if the string had been subjected to mechanical shock while it was in storage between uses. And in an extreme but unlikely case, what happens if all of the shunts in the string are welded? Do you then rely on the branch-circuit fuse or circuit breaker for protection, or do one or more shunts act as fuses and open up?

—by Bill Schweber