How to alleviate the negative effects of using parallel LEDs with a common resistor?

We all know the rookie mistake of using LEDs in parallel and not giving them separate series resistors. They are often doing it for ill-founded incentives like "I want to save money on resistors", "I don't want to solder that much", and "They all have the same color anyway so why bother?".

The correct answer in such a situation is to persuade the OP to use separate resistors. Forums and QA sites (including this one) are full with such examples.

However, consider a case when it's completely impossible to avoid using a single resistor for separate parallel LEDs. I know it's bad. I know it's evil. But in a case it cannot be avoided, what methods are there to at least try to slightly alleviate the problems such a circuit can will cause?

Consider a case with 2 to at most 5 LEDs of same color from the same production batch. The LEDs don't have to shine with their maximum brightness, so the current can be smaller than the maximum allowed current (although 1/x for the case of the device with x LEDs would probably be too low). The LEDs are inside a proverbial box, in parallel, and the box can not be changed. With the two wires hanging out from it, I can do whatever I want. Still, I would prefer simple solutions with few and only passive elements. (of course, complex microcontroller-driven solutions will also be appreciated, especially if no simpler solutions can be found)

Clarification: By "alleviate" I mean to reduce the brightness differences as much as possible, and, of course, to avoid them being damaged.

• The methods to alleviate the problem are more onerous than just fixing the problem and not paralleling LEDs. – Andy aka Apr 18 at 11:44
• What do you actually want to accomplish? What does "alleviate" mean to you? – Elliot Alderson Apr 18 at 11:45
• @Andyaka : With respect, the reason I had such an introduction instead of just posting a one-liner is to avoid exactly these kinds of replies. I wanted to avoid writing a lengthy novel why exactly it is completely physically impossible to not parallel these LEDs, and I hoped that the introduction helps you believe that I know what I'm asking. – vsz Apr 18 at 11:46
• I don't know what you are asking and I don't understand the point of the question. – Andy aka Apr 18 at 12:07
• @Andyaka : You mentioned "methods to alleviate the problem". Could you list some of them? No matter how onerous, they could become a nice and informative answer. – vsz Apr 18 at 12:10

Have R so big that the current wouldn't be too high even in case where only one led actually conducts. This prevents smoke-effects, but doesn't fix the possible non-uniform current distribution.

The leds can all seem to illuminate acceptably, but do not expect it stay true when the parts get older or the temperature changes. If there's already a substantial brightness difference, you can try paint the brightest one slightly to make it dimmer.

My suggestion: Get a can opener and fix the circuit.

You accept the fact that not all of your LED's will be the same brightness, and some of them might not light at all. There is no other way, other than using wire cutters along the green line, and replacing the part that can't be changed with a properly built circuit.

"I want to save money on resistors"

Do you really think you can find an alternative that allows you to replace resistor that cost 0.05$or € each? "I don't want to solder that much" Do you really think you can find an alternative that does not involve soldering? "They all have the same color anyway so why bother?" We bother because we have to, improper current limiting WILL lead to the destruction of your LEDs, color does not change anything. But to try to answer the question nonetheless, you could • Select your LED and group those that have the closest Vf; • Make sur they are evenly cooled, to avoid thermal runaway of an already unstable system; • Route the PCB in a way that creates small resistor in the tracks (using similarly long and thin tracks for each LED And last option (and this is really pushing it) would be to magnetically couple your LEDs, and drive them with an AC signal. simulate this circuit – Schematic created using CircuitLab • Sorry to be this blunt, but have you actually read the question? The quoted reasons you try to debunk one by one are exactly the reasons I listed as bad examples to show that it's not why I'm asking the question. But OK, I'll edit that part to make it more obvious. – vsz Apr 18 at 13:06 • I did, but debunked for completeness and to mirror the flow of your question – Sclrx Apr 18 at 13:08 • What kind of fancy-ass resistor has the exorbitant price of$0.05? ;) – pericynthion Apr 18 at 16:52

Better solution would be to have the LEDs in series, then they would all get the same current. Of course, the source voltage needs to be higher for them to all turn on. If one fails open, they all turn off. If one fails shorted, the current for the rest goes up until they fail too.

Example, 12V source, Vf = 2.7V, 15mA: (12V - 2.7V - 2.7V - 2.7V)/.015A = 120 resistor for current control.

One LED fails shorted, then (12V - 2.7V - 2.7V )/120 ohm = 32.5mA, above the normal 20mA continuous rating of a typical LED.

In parallel, if one fails shorted, the current jumps up and possibly takes out the resistor while the other LEDs simply don't turn on. If one fails open, the others see what? Their Vf is still the same, so one could argue the current draw from the source actually drops.

I made a timer box with MAX7219 driving 7 segment displays, each segment of a digit was made of 3 individual LEDs in parallel. They were very bright LEDs, rated for thousands of mcd at 20mA, something like this http://www.dipmicro.com/store/LED5W , so I ran the MAX7219 at a reasonable current level (I think the current limit resistor selected only allowed 15mA) and all three LEDs drew similar current so they shared that current, and were still very bright (not like the old indicator LEDs that needed 20mA and output just tens of mcd). After 10 years of use several days a week for several hours at a time, all the LEDs were still going with no failures.

So the answer to "Can I run xx LEDs in parallel?" could be "It depends", and not a hard & fast "No, Never". If the LEDs were run in pulsed mode (like from a MAX7219 which multiplexes across digits at 800 HZ rate), the LEDs are not on continuously and thus don't develop the heat that high current continuous on creates that typically kills an LED.

If to be used as on continuously than the smart design would have a resistor per LED, or ensure that the max current allowed by the current limit resistor did not allow a single LED to go into over current situation.

There is no "magic" solution. But you may not need one.

Most of the posters here will tell you that LEDs in parallel will never work.

But every Halloween and Christmas, where I live, the shops are full of battery-operated strings of LED "fairy lights". Each consists of a small battery box, one resistor, and a whole string of LEDs wired in parallel. Sometimes, for blue or white LEDs, the manufacturer won't even bother with the resistor, and will rely on the internal resistance of two AA cells.

Despite the crude design, they do actually work. The resistors are all the same, and are likely to all be at much the same temperature. The curves of voltage against current will be very similar for each LED. Because of this, the current naturally tends to balance out between them.

The only thing you may see is that if you run the LEDs at a very low current, some may be obviously dimmer than others. Increase the current, and they will all be reasonably bright.

Mount the LEDs on an array of individually-controllable thermoelectric Peltier devices such that you can independently bias each LED die warm or cold. Position a calibrated camera in view of the LEDs such that you can estimate how much light each is putting out. Wrap a feedback control loop around it, adjusting the heating/cooling to achieve even light distribution.