White LEDs in parallel

[leonelabs]

I've been working on a project that employs a large array of white LEDs in parallel and I wanted to get some feedback from the community.

I have 10x white LEDs in parallel and only a single resistor to restrict the current supplied from a battery. If I say the battery is 3V and the LEDs have a Vf = 2V then with a 10ohm resistor I should be getting about (3V-2V)/10 = 100mA coming from the battery. Now this 100mA will get distributed between the 10x LEDs roughly equally, but not exactly (each LED has a slightly different IV curve), so that each LED gets approx. 100mA/10 = 10mA/LED.

With each white LED rated at up to 100mA this seems all well and good yet I have a gnawing feeling that I'm missing something about this arrangement. I know the general rule of thumb is to include a resistor with each LED but this arrangement seems to work just find and could scale to 100's or 1000's of LEDs in parallel. So what am I forgetting?

I imagine failure modes could be problematic i.e. if one LED were to short-out then the other 9x would dim as the majority of the current flowed through the short.

Any other thoughts?

[mikeselectricstuff]

Shouldn't be much pf a problem as long as the LEDs are the same type. Vf will be around 3v, not 2v.

[sleemanj]

I imagine failure modes could be problematic i.e. if one LED were to short-out then the other 9x would dim as the majority of the current flowed through the short.

Any other thoughts?

Consider the other failure mode, led going open circuit.

1 goes, current through others increases, next weakest goes, current increases further, next weakest...

As long as there are enouh leds in parallel or thy are running suitably lower than their rating that the current from a few failures divided over the rest is not significantly an increase you can get away with it of course, but if for example you have 3 in parallel running at their rated current and one goes out now the 2 left behind are at 150%.

[fantasy2]

I would never do something like this. There is still a variation between LED's and even from the same batch(that's why usually they sort and rate high power LED's) you see differences. It's just a matter of time before this schematic fails.

At much lower than rated current you probably get away with it though...

[pmbrunelle]

Just use a boost converter and put the LEDs in series.

[Paul Price]

The idea that a LED will short-out is nonsense. LED's fail by giving off less light, it is usually hard to get them to short-circuit except with massive over-current abuse.

This should work fine and it is simple and so it is a good idea, with the more LED's in parallel, the more reliable this approach will work. This is especially true when running the LED's at less than max. current rating currents of the LED's, while   not having very high operating temperatures. The Von of most LEDs in a batch are usually matched fairly close.

Newer LED's are much more rugged than older devices, but I've seen many things, like flashlights and bicycle lights use this approach without any problem over a wide temperature range.

Newer LED's are used in the today easy to find direct replacements for incandescent bulbs and are very reliable even at high temperature. And although the power supply in the shell can fail after thousands of hours, the 20 or so LED's being lit don't.

For example, if one LED in the paralleled group drops in operating voltage(gets a little warmer), the increase in current causes the voltage to again rise across that same LED (to match the V of all the others at the same time) and this tends to equalize current sharing. Even if a LED has a V/I curve like an ordinary diode, it still has a fairly high enough internal resistance that ballasts each LED.

[Zero999]

Yes, this can be done but it's not ideal. Resistors are cheap so why not use a separate resistor for each LED?

[Srbel]

Every LED should have its own resistor. Otherwise, some LED will pull more current than others.

[Fungus]

Every LED should have its own resistor. Otherwise, some LED will pull more current than others.

Only important if you need carefully matched brightness (eg. scrolling LED displays).

I'd only do this if I was planning on running them at less than maximum power. Aiming for exactly 20mA (or whatever) will be difficult. In general though, the more LEDs, the better. It could save a lot of snipping/soldering if you're building something with thousands of LEDs.

[leonelabs]

Thanks for the feedback guys!

I can definitely see subtle brightness variations but I'm running the LEDs at <10x their max operating current so visually it appears OK.

It could save a lot of snipping/soldering if you're building something with thousands of LEDs.

This is the main reason I want to avoid the resistors; although the cost/resistor is low adding them in will double the part count per PCB and will add a non-trivial amount of cost/time to the assembly process.

[mikeselectricstuff]

A resistor costs way more to place than the part cost. Parallel LEDs of the same type are unlikely to be a problem.