LED Christmas Lights

and How to Fix Them


A Ciphers By Ritter Page

Terry Ritter

2007 May 24


Also see: Christmas Lights and How to Fix Them
Also see: Troubleshooting Christmas Mini-Lights
Also see: Christmas Mini-Lights FAQ


Introduction

[Front Porch]

Christmas lights are the strings of little electric light bulbs often used on Christmas trees or to decorate homes or yards. A wide range of strings are available, but this article is about the new "LED" bulb strings. "LED" stands for Light-Emitting Diode, and the colors they produce are both stronger and purer than dye coated incandescent lamps. The Philips LED strings I discuss here cost about 10 times as much per bulb as the cheap incandescent strings, but:

Many people find the intensity and purity of the colors in LED strings simply more enjoyable than incandescent lights. LED strings are not just a more costly modern version of the same old thing, but instead something better and more worthwhile. Someday they will be cheaper.

The ideal from the retail point of view might be for consumers to buy a few strings, use them for a few weeks, then throw them out. However, since LED strings are expensive, most consumers probably will want to keep them from year to year. But as with almost any new technology, there are some surprises!


LEDs

[LED Bulbs Disassembled]

Light-Emitting Diodes are not like incandescent lamps, and they do not behave like resistors that heat up. Accordingly, they cannot be used in the same way as flashlight bulbs or mini-bulbs. An incandescent flashlight simply connects each side of the battery to the bulb which then lights up automatically. If we connect an LED in the same way, it is likely to burn our almost immediately.

As an incandescent lamp filament heats up, it increases in resistance and reduces the current draw. Having a filament that automatically varies in resistance tends to reduce sensitivity to battery voltage. But LEDs have no filament and typically require some sort of external current-limiting device, whether a resistor or a more complex driver.

LEDs have been used in equipment for decades, and generally have had a resistor or on-chip current source for each and every LED lamp or numeric segment (in displays, the current typically is multiplexed between digits, one digit at a time). An easier alternative for portable situations is to use batteries which simply cannot source more current than the LEDs can handle. The power source limitations of little AAA or watch batteries can be used advantageously in LED flashlights. And, if we can find a bunch of LEDs which have nearly the same forward voltage, we can put them in parallel, drive them with a somewhat larger battery and get a lot more light, still without using discrete resistors. In essence, for some applications, the current-limiting we need is built into the battery.

Diode Technical Limits

Light-Emitting Diodes are a form of semiconductor technology. A diode is a device with two terminals that generally allows electrical current to flow in only one direction, after a certain voltage has been reached. Although the diode directional properties are the reason most diodes exist, they are almost a side-effect for LEDs, which produce light only when current flows.

LED Color

In an ordinary light bulb, electrical energy heats up a tiny metal filament so hot that it glows. The heat itself makes the filament emit light, and the hotter the filament, the greater and whiter the light and the shorter the lamp life. The heat produces a wide spectrum of different colors that we can call "white," but our cameras show us that incandescent light is more red whereas natural outside light is more blue.

We can put colored filters around incandescent lights to change their apparent light color. The filters, typically of dyed translucent plastic or coating, transmit a subset range of light frequencies that we see as a particular color. Since the filters reject other light frequencies, only a part of the total gets through, so the result is dimmer. And a light filter can only select light frequencies to pass and reject; it cannot create light frequencies that did not already exist. So if we want blue or violet from an incandescent light, we have a problem, because there is just not much blue there in the first place.

In a colored LED, selected semiconductor materials produce only a narrow range of pure color frequencies. Filtering colored LED light is not particularly helpful, since there are no unwanted frequencies to be filtered out. In general, we cannot change the color of LED light with filtering as we can with incandescent light. Manufacturers do tend to use colored diffusers with colored LED strings, but those have little effect on the operating color. Presumably the colored diffusers are what the public expects, and also are useful in manufacturing when heaps of bulbs get tipped over and mixed. Diffusers are helpful because the high-output LEDs tend to produce a beam or narrow cone of light which we might want to deflect in many directions.

The white LED bulb is a special case: The white light produced by a white LED is fluorescent light, at least in part. Typically, bright yellow fluorescent phosphors are activated by a high-efficiency blue LED, and the result includes yellow (with some orange and red) as well as the original blue. Even though the result is white-ish, it does not cover the full incandescent spectrum, so filtering may be disappointing.

LED Failure

LEDs are semiconductor diodes, and diodes almost always fail as a short (a "short circuit," a low resistance path). Shorting upon failure conveniently takes the place of the "shunts" added to series string incandescent mini bulbs, because those shunts often fail to work. One of the hardest problems in mini light repair is to find which one of all the bulbs in a series circuit has burnt out with a shunt that did not short out like it should. The failing open problem should be rare with LEDs, but sometimes the tiny bond wires do break, especially if excess heat softens the plastic and allows the external leads to shift.

LED strings can fail when LED leads corrode or wires break or fuses blow. Then a whole string or perhaps a half-string circuit would go out. Fixing any broken circuit involves locating the break, then repairing it. Unfortunately, locating a break in an LED string can be more challenging than in an incandescent string, because each LED reduces the overall voltage and modifies any test signal.

The rated life for these LED bulbs is given as 25,000 hrs, average, but that may sound like more than it really is. For example, we are running 7 strings of 60 bulbs each, 24/7. That is 7 strings * 60 LEDs-per-string = 420 LEDs, for 24 hours-per-day * 7 days-per-week = 168 hours-per-week. So we burn 420 LEDs * 168 hours-per-week = 70,560 LED hours-per-week, which could imply a failure rate of almost 3 bulbs per week. And, indeed, 3 bulbs per week is the failure rate actually observed.

Normally, however, when manufacturers talk about LED "failure" over time, they are not talking about LEDs actually going dark, as many of the Philips LEDs do. Instead, LEDs normally do not die, but simply get dimmer to a point where the result is no longer acceptable. In these Philips strings, one might hope that would mean 25,000 hrs for each LED, or 25,000 hrs/168 hours-per-week = 149 weeks or about 3 years of continuous full-power operation. Unfortunately, the high rate of LED total failure in these Philips strings has obscured any LED fading issues.

String Construction

[Philips LED String Boxes]

The Philips 60-LED multi-color strings are arranged in two circuits of 30 LEDs each with a total forward voltage of about 75 volts. The LEDs are off until at least 75 volts is applied. Abstractly, one could have any number of circuits, so we could see 30, 60 and 90 bulb strings which are virtually the same except for bulb count and length. Other manufacturers apparently use circuits with 35 LEDs, thus having a forward voltage of about 87.5 volts, which should be a little more efficient. Since red LEDs have less forward voltage, they can occur in larger numbers, such as a single circuit of 50 with a total forward voltage of about 100 volts.

In the Philips strings, two wires go all the way through the string from the fuses in the power plug to the power socket at the end; these power wires allow another light string to be plugged in where the first ends. A third wire goes from near the power plug to near the center of the string, and this connects a circuit of 30 LEDs in series with a molded resistor. A similar wire goes from near the center to near the end, so most of the string has three wires. The two circuits are independent: an open in one will not affect the other. Each resistor is enclosed in plastic and will get warm in operation, but should never be too hot to hold in the hand.

Just like the incandescent strings, the Philips strings can be cut in the middle (with the power off), where there are only 2 wires. The first half can be used after insulating the cut wires with shrink tubing or electrical tape. The second half can be used by finding another fused plug, perhaps from an old incandescent string, and splicing that into the cut wires. Or maybe the second half could be used for light replacements.

The AC power line is alternating current; that means it flows in one direction, then reverses and flows in the other direction, repeatedly. The AC voltage varies in the "hill and valley" pattern known as a "sine wave." In the US, the standard AC line is about 120 volts AC at 60Hz; the line goes from 0V to +168V back to 0V, to -168V and back to 0V, 60 times a second. But LEDs are diodes and want to conduct in one direction only. When the LEDs conduct, we have about 30 LEDs eating up about 2.5V each, on average, so little current flows until the AC gets above about 75V. The strings do not seem to include any other power diodes: When the AC wave goes negative, presumably the sum of all the LED reverse voltages exceed 168V so that significant reverse current does not occur.

Not every circuit which uses LEDs is necessarily efficient. The LED light strings described here are efficient largely because they use many LEDs in a series circuit. That reduces the line voltage by about 75 volts before it gets to the current-limiting resistor. That means only a modest fraction of the total line voltage is wasted heating the resistor. With this efficient topology, we can use the line voltage directly, without voltage-changing transformers or power supplies. And we can use just a single resistor for the whole circuit, instead of needing a separate resistor for each bulb.

LEDs on AC

[Reading Corner]

For an imaginary cycle of 360 degrees of the AC line, about 26.5deg (1.2msec) pass before the line gets to 75V and the LEDs start to emit light. Then the LEDs are on for 127deg (5.9msec), and off for the remaining 206.5deg (9.6msec). That means we get a 5.9msec hump or pulse of light every 16.7msec: The LEDs are actually ON only about 35 percent of the time (the current and light ramp up then down during the pulse). The pulsing effect is similar to what we see in fluorescent lights, but worse because it is slower: only one side of the AC wave is used. In particular, it is possible to move a hand in the LED light and see a stroboscopic effect. Flowing water is particularly interesting. But I have no problem reading; the light does not seem blinky unless it is the sole illumination for something moving. In 5 strings I have tested, both circuits in each string each use the same line polarity, which is wise.

I would like to see the manufacturer add a single line-voltage power diode to avoid depending on summed LED reverse voltage limits. One can imagine various other design alternatives:

It is easy to criticize any particular design if one does not have to provide a viable alternative. At our house, LED blink might have been slightly disconcerting at first, but now we rarely notice it, or perhaps even find it amusing. I have read several books in white LED light alone without noticing any blink. LED clocks blink all the time, but are still used because we do not notice any blink. Of course LED light can make flowing water interesting, in the same way that a strobe light can make dancing interesting. Anyone with an idea for a better LED topology can take a couple of these strings with a few other parts and see just how well their design would work. Actually building one and calculating power loss and so on may expose problems not apparent in the imagination.

Plug Fuses

Like the incandescent strings before them, the LED strings have a fuse in series with each side of the AC line, located in the plug. Usually there is a trap-door in the plug to open and access the fuses, or there may be a section that pulls out. Quite often these are hard to open. The fuses consist of perhaps 1/4 inch of tiny wire inside a glass tube. The intent is for the tiny wire to get hot, melt and separate when too much current flows.

Fuses are used to protect the wires. If there were no fuses in the plug, and for some unknown reason too much current was flowing, that could heat the wires, melt the insulation, and cause a direct copper-to-copper short circuit across the AC line. Without fuses to protect the wires, we would have to hope the outlet circuit breaker would flip before a fire started. Since the copper wires in these LED sets seem to be similar in size to those used in incandescent sets, a similar sized fuse offers similar levels of protection.

The replacement fuses which come with the Philips LED lights seem to be rated at 3 amps. (One end of each fuse is labeled "(UL)3A".) These seem to be exactly the same as the fuses for incandescent mini lights. Looking at my replacement fuse collection (from a decade of buying lights), over half are marked, and every one of those is a 3 amp fuse.

When fuses are rated at 3 amps, that theoretically means they should take a 3 amp load indefinitely, and blow at some unknown higher current after some unknown delay. In practice, fuses are rarely manufactured to tight standards, and the closer one gets to the rated value, the more likely it is that a fuse will fail from time to time. The 3 amps these fuses allow theoretically represents about 360 watts of resistive load. That would be about 9 incandescent strings of 40 watts each, although 3 strings would be a good limit to keep the fuses cool.

If we neglect pulse effects, an LED string which takes 4.8 watts at 120VAC should take about 120V / 4.8W = 0.04A (for 2 circuits) which is about 1/25 of an ampere. If the 3 amp fuses are the weakest part of the circuit, as they should be, they should handle about 75 strings. To keep the fuses cool, one might want to have a 25 string limit.

The documentation with the Philips products indicates that no more than 3 strings should be daisy-chained. It may take some time for rules to catch up to the reality of just how little power the LED lights use. However, the stated limit of 3 strings may be more related to the idea that high-power incandescent strings could be plugged into LED string AC sockets. A chain of 3 incandescent strings at 40 watts would take about 1A total, and would be a reasonable value for 3A fuses.

When strings are connected in a "daisy-chain" powered from a single plug, all the power used by the chain goes through the first plug fuses. The second plug takes less current, because it does not power the first string, although fuse manufacturing variations may hide the difference. When plug fuses blow repeatedly, we need to see if we are close to some limit. If so, we can assume the plug is just handling too much power, and we need to shorten the daisy-chain. If we are not close to a limit, we seriously need to track down and fix or replace the source of the problem.

In practice, things other than the fuses can be problems. Any poor connection, such as might occur in the AC plug, can heat up under current load, and there more current there is, the more heat will be created. If the heat is sufficient to distort the plastic plug or even tilt the connections to create a line-to-line short, that could be a much more serious situation than a mere blown fuse. If any molded plastic part ever gets too hot to hold, something is seriously wrong, and the issue should be investigated and corrected immediately.

In these strings, fuses are required by design. Since LEDs can and do fail short, the possibility exists that all of the LEDs in one or both circuits in a string might short out. That would leave the resistor directly across the full might of the AC line. Never short out the fuses.

GFCI Trips

Many newer homes are equipped with Ground Fault Circuit Interrupter (GFCI) breakers, especially for outside power. These devices continually sense whether the same current is flowing back from a socket as is flowing out to it. When those currents are not the same, some of the current must be getting loose into ground. Since the lost current may be going through a person, that is ample reason to trip the GFCI, and possibly flip the breaker.

When ordinary inside strings are used outside, rain can cause some electrical leakage from one or more bulb sockets into foliage and then into ground. Even a fairly tiny current leak could flip a GFCI breaker. The obvious solution is to somehow find the leak and insulate it, but finding a leak that gives no outward indications can be a daunting task. The usual approach is to do all we can and hope for the best. Perhaps a clear bag or cup could be placed around each lamp socket.

Another possibility for avoiding GFCI trips from outside strings may be an "isolation transformer." The isolation transformer lifts the 120VAC line from ground, and so avoids ground faults. That should fool the GFCI, and probably is safe enough for outside lighting. But isolation transformers are expensive, and local electrical code issues may be involved, so licensed experts should be consulted.

Perhaps an overall better outside installation would be to use LED "rope lights." Since the lights are contained inside a vinyl tube, they are much better protected from rain than an open light string. While LED rope lights are expensive, the ability to avoid GFCI trips may make them more worthwhile than at first they might appear. Rope lights also minimize the rusting of LED leads.

Dimming

In general, all lights last longer if they are dimmed, and there are several possibilities:

Sadly, most normal triac lamp dimmers will not turn the LEDs quite to OFF: The LEDs do get unusably dim, just not OFF. In multi-color strings, the blue, white and red may be most apparent, because they typically have the highest efficiency and thus produce the most light from a tiny trickle of current. The dimming is far beyond what is needed to extend lamp life, but having lights be partially lit can be an issue in other ways. And the same issue occurs with most devices using triac power control, such as:

Most triac based control circuits are two-wire devices: They have one wire to connect to the power line and another wire to connect to the load. That is an easy series connection and all the power going to the load flows through the controller. That works fine, especially since the worst that could happen, even if the triac shorted out, is that the controlled light would just be full ON. But a little power is needed to run the control. If actually no power at all was flowing, the controller could not read the knob setting, and so would not "know" to keep the power OFF. Fortunately, even a small non-LED load may be enough to recover full triac control.

A 4 watt night light typically is sufficient to keep the triac controller active while turning LED lights all the way OFF. I have tried a night light on the end of our patio strings, which solves that problem while creating another: The incandescent light comes on with the LEDs. If that destroys the "effect," the night light might be:

On the other hand, some of the motive for automatic light control goes away when the full-ON lights take 1/10 the power previously used, and almost-OFF lights may take 1/20 of that. Until better dimmers become widely used, there may be a tradeoff between what we want and what we can get.

New Opportunities

[Rope Light around Mirror in Bathroom]

Most of my interest in LED strings is the light itself, and not the presentation. The history of lighting has seen a profusion of lighting fixtures which can be beautiful in themselves, but fixtures made for a few large bulbs may not be appropriate for large numbers of small bulbs. Ever more powerful LEDs are being developed for central lighting, but those LEDs need substantial heat-sinking simply to survive. I would like to consider the possibility that light from many small and cool individual sources lends itself to practical use in new ways.

These LED strings cannot compete with the concentrated light of a 60 watt incandescent bulb, or a 40 watt fluorescent fixture. Accordingly, their best use may be in the distant corners that are not well lit, or perhaps where hot lights are irritating or dangerous. For example, an area of the wall or ceiling covered with small bulbs can be considered as much of a light source as a standard ceiling fixture. If we start to like the new form of lighting, we may find that visible wires are not so bad after all.

One big advantage of LED strings is usable light with much less electrical power. Previously, we would expect 100-bulb 40 watt incandescent strings to take about 1/3 amp of current. Now we have a 60-bulb 4.8 watt LED string (which apparently gives more actual light), that takes about 1/25 amp. Using less power allows longer daisy chains and is a cost savings around the holidays, but is even more important for lights which are on all the time.

With less power used, we can afford to be less interested in turning lights on and off, and more interested in lighting areas that central bulbs do not reach well. Perhaps half-strings would be appropriate for some places. Various applications that use "Christmas lights" can be more practical than decorative.

One option is to collect 10 or 15 lights in a "flower" where all LED bulbs point the same way. There would be 6 or 4 such flowers in a normal string. Twist-tie can be used to hold together all the sockets and wires for each flower. Since most light comes straight out the top of super-bright LEDs, each flower is like a mini spotlight. The amount of light in each flower is comparable to large LED flashlights and is quite effective for reading. These light flowers also might be used for lighting pantry shelves or displays.

LED lights can be blinked without the decrease in life we see in incandescent strings.

An unusual design advantage of these particular strings is that they use only one side of the AC power cycle. When we connect several strings in a daisy-chain connection, we can rotate the plugs and cause any string to use either half of the cycle. There is no apparent visual difference, until we start fooling with the power. With appropriate power diodes, it is easy to build a controller to switch each side of the AC cycle on and off, and so independently control two different groups of LED strings from the plug end alone without separate wiring. Depending on the orientation of the plug for each string, any LED string in a long chain can belong to either group. More clever circuits could allow each group of strings to be dimmed completely independently.

Modifications

Remove socket or plug. It is possible to remove the trailing AC socket, perhaps to better fit an enclosed space. With the power off, cut the 2 wires to the socket. On the light string side of the cut, bend each of the wires over and separately insulate with heat-shrink tubing. It is just as easy to remove the AC plug, but this is a fused plug, and fuses are absolutely required for safe operation.

Make a half string. In most cases, in the middle of a string there is a place where only 2 wires are twisted together. By cutting there, the first circuit can be used alone. Again, bend over each wire and insulate with heat-shrink tubing. The second circuit could be used independently by adding another fused plug perhaps taken from a string of replacement lights.

Get rid of the AC wires. In these strings, two wires take AC power from the fused plug to the AC socket at the end of the string. If we do not need AC power there, we can think about removing both of those wires, leaving only 1 in most places. We end up with 2 circuits of 30 LEDs and a resistor each. We do need to be particularly careful to know which wire to cut as 2 wires become 3 inside a lamp socket or resistor enclosure.

If we splice 1 or both of the separate circuits across the wires from the fuse plug, we get 1 or 2 single-wire circles of LEDs using only a third of the wire used originally. This amount of effort could be worthwhile in artistic projects such as lighted frames where the unneeded wire would be too large or distracting.

Specific Light Strings

[Blue Dome Display]

We have these specific light strings. They are all Philips brand, bought in Austin, Texas in late 2006 at the full retail price of $11.99 per box, plus tax, at Target:

Measured Voltages

LED forward voltages exhibit substantial variation, even for the exact same color in the exact same string. Here are few actual values for 3 bulbs of each color from these strings, measured at 20.1mA DC:

I have been unable to find a datasheet for the LEDs themselves; perhaps Philips uses outside suppliers.

Maintenance

Reversed Socket Fault

One of our white strings arrived with multiple bulbs dark. By replacement with known-working bulbs, all but one of the sockets were filled with working lights. But in one socket, even a known working lamp would not light, even though the rest of that circuit would light as soon as a lamp was inserted. Clearly, current was flowing, yet the known working lamp did not light.

As I looked at the dark bulb, the rest of that circuit suddenly brightened, and it was clear the dark LED had just died. Trying the bulb in a known working socket showed the formerly-working bulb was now bad. So I had actually watched an LED fail due to reverse current flow. The failure took 5 or 10 seconds, and for that time the rest of the circuit was somewhat dimmer than it should have been. That should be a tip-off for the future.

Inspection of the lamp wire showed that the polarity or direction of the dark socket was the opposite or reverse of other sockets. One possible fix might have been to simply reverse the one LED in that socket. However, since that would surely set up a sequence of similar failures in the future, it is fortunate that is not really possible. The socket connections are at different heights and the LED leads have been cut to different lengths. The correct solution is to reverse the socket.

To reverse the socket, I cut both wires from the socket, one on each side, about in the middle of their lengths. Then I stripped the insulation from all 4 ends back about half an inch. Then I reversed the socket, and twisted together the copper wires for each pair of close ends. Both metal connections were then insulated with electrical tape. When tested, the reversed socket worked properly. The whole repair took perhaps 5 minutes.

Since this string arrived with 4 bad lamps, it is tempting to speculate that somebody, somewhere, tried to fix the dark lamp, and in the process killed 3 others. The disturbing part of this is that the result ended up on the shelf, even though somebody, somewhere, might have known the string was faulty. Of course, for me the result was educational.

Another Reversed Socket

After working fine for almost 2 months, a reversed socket was found in our wonderful blue dome string. Various bulbs had gone out in that string and were replaced with bulbs from a spares string. This time, however, spare bulbs would not light in that socket. Several spares were tried, and, probably, fried.

Close examination showed that the plastic latch on that socket faced away from the AC plug, whereas all others faced toward the plug. Thus, the socket was installed backwards. Moreover, the original LED in that socket was also reversed! This exposes a deliberate manufacturing response to a known manufacturing error. It is hard for me to express just how insane that really is: The sockets are there for the obvious reason that these LED lamps do frequently fail and must be replaced. But when a socket is reversed, no normal replacement lamp will work, and all those tried will die.

Replacing a socket is usually not too bad, but this one was at the start of the second light circuit, and had 3 wires going into the socket. Two of the wires are there just as a convenient join, but now we have to extract them from the socket and join them up. Probably that connection should be soldered, since this is the path that provides current to subsequent strings. Soldering should also add some mechanical strength since the shortest wire will tend to bear the load of wire and lamps. Then we take a socket from the spares string, and attach it between the join and the third wire. Be sure the bulb latch points the correct way before soldering the new socket wires.

There is a sad lesson here, which is that buyers need to examine Philips LED strings before they are installed in places where we would prefer not to work. If a reversed socket is found soon after purchase, normally the string can be returned to the store so someone else can enjoy the problem. But if we find a reversed socket in a hung string, we are typically forced into correcting the manufacturing fault. After we reverse or replace the socket, the reversed LED from the original socket should be clearly marked as backwards or discarded.

LED Leads Rust

My picture of LED lights on our patio shows 4 strings on a dimmer, and they stay on (dim) most of the time. Recently, one of the circuits (a half-string) went out.

For a circuit to light, there must be a continuous metallic connection through all LEDs (and the resistor) to AC line voltage. Every wire must be unbroken, every socket making a good connection, every LED conductive (not necessarily producing light) and the resistor must have near the appropriate value. Only a limited number of things can go wrong.

When wires are dragged about, sometimes they break, most often at an end where they cannot bend, or perhaps at a place of bending or strain. But these lights are just hanging. Resistors rarely change value, and any change should be fairly minor. So that leaves the socket to lamp connections.

Closer examination showed that the circuit was not completely out, instead showing a tiny glow at the highest dimmer setting. That meant the connection was complete, in a sense, just no longer good. This was a big advantage for removing and replacing lamps, because the lamp to socket connection is problematic. Even a tiny glow allows one to see that a lamp has been properly seated. Without that, removing and re-seating lamps can create more circuit problems than were originally there.

I dawdled around for several weeks trying to think of an electronic way to avoid a brute-force search of 30 lamps and sockets. The hum tracers used for incandescent strings really do not get to the problem. One might try measuring the voltage between various lamps, but with tiny currents, the AC source and the forward voltage drop in each LED, the situation may seem complex until the problem is found. (Then everything will be obvious.) In the end, brute force seemed the best alternative. This involved examining at each wire, pulling each lamp and looking for abnormality.

Pulling lamps from sockets quickly revealed one with orange-brown rust on the LED leads and socket contacts. Although one might expect all connections to be copper (wires), brass (socket contacts), or tin-lead (solder), clearly some steel was present. Since rust does not conduct, having rust on contacts is not a good thing. But we can scrape off the contacts (using an insulated screwdriver), and scrape the LED leads so they will make good connection. We also can stick the screwdriver tip into the socket to short across the contacts, and if this socket was the only bad connection, the circuit should light. Fixing the first lamp did not restore operation.

Eventually, a down-pointing lamp was found that showed a small amount of orange-brown color under the diffuser at the base. That would have been a good hint, but finding it required very close examination. Removing that lamp from the socket exposed very rusted leads. This was ordinary iron rust. During subsequent examination, one of the leads actually fell off, having rusted completely through. The iron part of this is the LED lead. Tinned, they look silver, and one might expect them to be copper or bronze underneath, but in fact they are steel, which is a big problem for use out in the weather. A down-pointing lamp has an up-pointing socket opening which can collect water on the LED leads which will rust.

In the process of pulling lamps, several of the diffuser lenses came off. Normally these can just be pushed back on again, although they never seat quite as tightly as they did originally.

It also became apparent that not all LEDs are positioned similarly. One might expect that the small LEDs would be held up by the plastic holder so that the LED will be visible from the side of the diffuser. But some LEDs are positioned down into the holder and cannot be seen from the side. The light, of course, goes mostly out the top and is then diffused, so little light is lost, but there is a difference.

Examination of a retreated LED showed it to be firmly in position, held tightly down and back by the bent-over LED leads. In contrast, the better-positioned "exposed" LED was loose, flopping out and in. Those leads were bent to allow the LED to be positioned into the diffuser, but only being plugged into a socket actually enforced that. When that holder is not in the socket, the leads are free to move and drop the LED back inside the holder.

LED Replacements

Unlike incandescent strings, which greatly benefit from having matched lamps, matched LEDs are not needed. The distinction is between a series string of variable resistors, and a series string of diodes. Having equal voltages across individual resistors in a series string depends not upon an absolute resistance value, but just that all are similar. That allows wide variation in production, as long as the lamps in a string are similar. But incandescent filaments dramatically change resistance as they heat, meaning that it is an advantage to have all lamps made in the same production line. In contrast, having appropriate voltages across individual LEDs is virtually automatic, provided sufficient current flows.

Since LED strings do not benefit from matching, there is less need to buy identical replacement strings at the same time as the originals. Almost any similar LED that can be made to physically fit should work. The tradition in the Christmas lights business seems to be to use different socket designs each year, thus encouraging the customer to purchase all new strings every year. However, in most cases, bulbs can be extracted from a socket base by straightening the leads. Then a new bulb can be placed into an old base, but polarity counts. Provided the leads do not break, the LEDs from similar string designs can be reused as spares.

Modern super-bright LEDs occur in two general voltage groups: about 3.2V (white, blue or green) and about 2.0V (red, orange or yellow). Ideally, we would replace a higher-voltage LED with one of the higher-voltage colors. Similarly, we should replace a lower-voltage LED with any of the lower-voltage colors.

If it becomes necessary to use a replacement from a different voltage group, that will change the string current slightly. If 3.2V LED is replaced with a 2.0V LED, an additional 1.2V will added across the resistor. A single LED change could increase current by a couple of percent and the power dissipated in the resistor by twice that. With enough changes the resistor may get significantly hotter.

Product Quality

My hard-learned policy for incandescent mini strings was to always get spare strings simply for replacement bulbs. With highly-reliable LED lights I expected spares would not be needed, but alas I was wrong. LED lamp failures still occur every week. For these Philips strings, it is important to buy extra LED strings simply to use as replacement bulbs.

Initial Failures

Operational Failures

By operating week:

  1. 2 Faceted Green, 2 Faceted Blue, 1 Faceted White (finding/fixing reversed socket)
  2. 3 Faceted Green
  3. 1 Faceted Green
  4. 2 Faceted Green, 1 Blue Dome
  5. 1 Blue Dome
  6. 1 Faceted Blue, 1 Blue Dome
  7. 3 Faceted Green, 2 Blue Dome
  8. 4 Blue Dome (1 finding/fixing reversed socket)
  9. 4 Faceted Green, 1 Blue Dome

LED Failures in 10 Weeks of Use

We have seen 34 bad LEDs (over half of a complete string) in under 10 weeks with 7 always-on Philips strings. There were 29 operational failures in 9 weeks, so on average there were about 3 LED failures per week. Over this period, a little under 1 percent of the operating LEDs failed each week, a figure much higher than we might assume in the absence of actual experience.

We got one multi-color string as replacements for the 4 multi-color strings we run continuously, and in just over 2 months we have run out of replacement faceted greens. Now the greens get replaced with yellow or orange, which is very sad. This failure rate with LED strings seems much worse than the incandescent strings they replaced.

After 6 Months

We have a blue string running all the time inside which has been a lot of trouble. A total of 30 blue LED bulbs (half a string) have been used in about 6 months. That usage rate is about 5 bulbs per month or a full string per year. Although the failure rate does seem less now than at the start, I just replaced 2 blues this last week.

The 4 multicolor strings running all the time inside continue to lose greens, and our replacement string has none left.

In contrast, the 3 white strings running all the time inside have had few failures. We also have a white rope and 2 blue ropes on all the time, with no LED failures.

The 4 white strings on the patio (dimmed most of the time) have needed only a couple of replacements. Those strings are not directly in the weather, and were thought to be protected by the patio cover. But this last week 2 of those bulbs were found to have developed lead rust, with fewer than a dozen bulbs examined. That seems ominous.

Terry Ritter, his current address, and his top page.