Long Life for a Cheap LED Torch Light

While being busy with another Instructable (https://www.instructables.com/Cheap-Quick-Dirty-Emergency-Light/), I soon realised that of the nine LEDs in the very cheap torch light I used for this project, one did no more illuminate.

This was reason enough to completely disassemble the front section of the torch light and checking how it's made. To my horror I saw that the nine LEDs simply were connected in parallel, and then in series with a 5 Ohm resistor. The whole assembly is supplied by three 1.5 Volt batteries connected in series (resulting in a supply voltage of 4.5 Volt).

Connecting incandescent bulbs in parallel is a perfectly normal procedure that you know from your first electricity experiments in school physics. The different lamps in your living room - all connected to the same AC supply voltage, i.e., in parallel as well - are a perfect second example.

The trick is: Incandescent bulbs have a positive temperature coefficient - which means in clear, that their internal resistance rises with rising temperature (and as you perhaps learned the hard way, they get rather hot in operation). In other words: Should the supply voltage of an incandescend bulb rise by a bit (courtesy of your electric power company), its temperature rises as well, which, together with its increasing resistance, limits its own current - up to a certain degree. And this happens, of course, to every single bulb in your group of bulbs that are connected in parallel.

As opposed to that, connecting LEDs in parallel without limiting the individual currents is a cardinal sin nobody ever must commit.

The voltage/current characteristics of an LED is similar to the one of a standard diode, which means that it is highly non-linear. When you increase the (forward) voltage, the (forward) current increases too. BUT: The current increases disproportionately faster than the voltage due to the non-linearity, refer to the diagram above (which is borrowed from Application Note 3256 by Maxim Integrated, published in 2004).

I measured one of the white LEDs from my torch light. For an increase of forward voltage from 2.76 V to 2.85 V (i.e., by only about 0.1 V), the current went from 1 mA up to 10 mA (which is a factor of 10! which still is on the safe side for an LED specified for 20 mA - but imagine what happens when you increase the voltage a tiny bit more...). So, limiting the current through the LED is mandatory, which can be done in several different ways. The cheapest of these methods is using a series resistor provided that the supply voltage is more or less constant.

So you have several LEDs, a series resistor, and a battery (the voltage of which is more or less constant over a rather extended period of time). Why, then, do I tell you that you cannot connect the LEDs simply in parallel, and then to the resistor and the battery (as shown in diagram no. 1 above)?

The real LEDs on the desk in front of you are somewhat different from the ideal models you normally use when designing or trying to understand a circuit. Real components always have certain tolerances, and apart from that, LEDs from different manufacturers can have significantly different forward voltages. This means that the voltage/current characteristics varies from LED to LED, even if they are from the same manufacturer and the same production batch.

When connecting a group of LEDs in parallel, they all are supplied by exactly the same voltage. Due to their slightly different characteristics, not all the LEDs of the group have the same current flowing through them. This can, under worst-case conditions, overload the one LED with the smallest forward voltage (and therefor the largest forward current), quasi the 'weakest link' - and by that, strongly reduce its life span. Once this particular LED has gone to the happy LED hunting ground, the remaining LEDs are still connected to the same voltage. And since the same current is distributed now over less LEDs, it is increased by a bit... and the next-weakest LED will fail faster... and so on, a vicious circle!

The only method for keeping all the LEDs that are connected in parallel alive for a long time is using one individual current limiting resistor per LED, as shown in the diagram no. 3 above.

I fail to understand why a torch light manufacturer ignores this very basic rule. Can it just be economical reasons, aka stinginess? But the individual resistors cost next to nothing, their placement on the PCB is done by a robot, and the soldering happens in one go anyway...

Perhaps the manufacturer believes that a customer having a prematurely deceased torch light will soon shop for a next one? Hm. I'm sure that any reasonable customer will buy a better one somewhere else, so his plan will not work out for sure.

BTW, there is, of course, a second method to make sure that all the LEDs in a group have the same forward current: By connecting them in series. This requires, however, a higher supply voltage that is not available from the battery pack in our application. On the other hand, much less forward current is required then, and a single current limiting series resistor is sufficient in this case.

I had a second, identical torch light that was, of course, opened and checked as well. The guy who had assembled it obviously had a bad day - the red (positive) wire from the battery was connected to the wrong side of the resistor (see diagram no. 2 above).

Huh?

This means that there was no current limiting at all for any of the LEDs! Or, to be completely honest, the current was somewhat limited, but only by the battery's rather low internal resistance - which, I'm sure, was far from being adequate. I am amazed that the LEDs even survived their first power-on.

This is, in fact - after a simple calculation - an easy task, and performed in a few minutes.

The forward voltage of the LEDs is, as I measured before, approximately 3 V. Subtracting this from the nominal battery voltage (3 x 1.5 V = 4.5 V) results in a difference (U) of about 1.5 V. This is the voltage that will drop across the series resistor. In order to have a current (I) of about 15 mA through this resistor (R) and its LED, it must be about 100 Ohm, because

R = U / I

according to Ohm's law.

In the first photograph above you see the tiny, round PCB from my torch light. It had accommodated nine white LEDs - eight in a circular pattern, and one in the centre. I removed the one in the centre (that was dead anyway) and cut the outer, circular track that connected all the cathodes of the LEDs (cuts marked in black in the picture). The next picture shows a star-shaped array of eight 110 Ohm resistors soldered onto this PCB. One terminal of each resistor is connected to the PCB and the cathode of one of the LEDs, the other terminals are all connected together in the centre; this point is now the new negative terminal for connecting to the battery with the black wire.

After stuffing all this successfully back into the torch light's body, you can test it. It might produce a tiny little bit less light than before, but all (now eight) LEDs in your torch light will live happily ever after - or will, at least, survive you.