A Question on Parallel Circuits

[High Efficiency]

I'm making a one off array of 6x LEDs in parallel and have a quick question.

I understand the logic behind not running LEDs in parallel but this application requires it.

Apologies for the ugly drawings but from an electronic perspective, is there any difference in these parallel circuits?

In essence, I'm wondering if drawing A) is two separate parallel circuits or whether the entire thing is still one parallel circuit (like B).

In both examples, the LEDs will be powered via circuit so there are no inline resistors.

[/URL][/IMG]

Thanks in advance.

 

[magellan]

Not that I can see. But what would be the reason to do #1 since it involves more wire?

 

[High Efficiency]

The risk, as I understand it, in running LEDs in parallel is that the LED with the lowest forward voltage could begin to hog the current, warm up, and begin down the path of thermal run away until it eventually self destructs.

The thinking in the drawing above is that the fewer LEDs running in parallel the lower the chance of thermal run away and eventually failure.

So, is A above 2 sets of 3 LEDs in parallel or one set of 6 LEDs in parallel?

I used A and B as a simplification but A would preferably be further divided into 3 sets of 2 LEDs.

I'm just wondering if there's a "safer" way of wiring together 6 LEDs than in B.

 

[Timothybil]

If you flip the first set of three LEDs in A over to the left, you will have B with a little extra wire. No electrical difference. The save voltage appears across the leads of each LED, and removing any LEDs from the circuit will not have any effect on the remaining LEDs.

 

[mattheww50]

Circuits A and B are effectively identical as long as the resistance of the wire is very small compared to the LED. Circuit A feeds the power to two groups of parallel LEDs in Parallel, so the current in each wire run to each LED string is half of what it is in Circuit B.
Losses in the wiring are I^2*R, and the current in each segment of the wire in figure A is half of what it is in Figure B. As long as I is reasonable, and R is very small, I^2*R will be very small compared to the energy the LED devices are receiving. As LED's become more powerful, the I^2*R losses can become significant when the strings are low voltage (6 or 12 volt) and the current is many amps.

 

[RetroTechie]

The thinking in the drawing above is that the fewer LEDs running in parallel the lower the chance of thermal run away and eventually failure.

That thinking is wrong. The difference in current results from small differences between the LEDs, and the (potentially large) differences in current that may cause. Failure then follows if LEDs overheat. As long as each LED stays within its rated power, no problem.

With 6 LEDs in parallel, the worst case is 1 LED eating all the current that was meant to go through 6 LEDs. With 2 LEDs in parallel, the worst case is 1 LED eating 2x the current that was meant to go through one LED.

But those worse cases rarely occur, unless LEDs with very different characteristics are placed in parallel. Or a LED totally dies, wiring breaks, etc. In practice, how well the characteristics of the LEDs are matched, will determine how even current is divided between them. LEDs with near-perfect match of characteristics -> no problem. Of course working conditions can be slightly different for each LED, the LEDs will age (and age differently), and then troubles begin.

If LEDs are not run @ full power but adequately cooled, and say 2x the normal current would be okay, you could place 2 LEDs in parallel. And then those pairs of LEDs in series. If 3x the normal current is okay, you could place 3 LEDs in parallel, and a # of those 'triplets' in series. And so forth.

I understand the logic behind not running LEDs in parallel but this application requires it.

Care to explain why? No room for series resistors? (tip: in a pinch, even wiring itself could be used as a small series resistor to even out differences between LEDs - just use thinner wiring and/or wires of a worse-conducting material than copper. Same for pcb tracks). No driver with high enough output voltage available? No single LED with the required output available?

 

[Mr Floppy]

many of those 3xAAA budget shower head lights are a number of LED's (usually) in parallel, but that is normally a 3xAAA alkaline battery power source.

Like mentioned above, dead-bugging a small series resistor doesn't take up that much space. You could use just a single dropping resistor if you had to.

 

[High Efficiency]

Thanks everyone for the replies. This has been helpful.

Electrically there is no difference between A and B so I'll likely move forward with B out of simplicity.

That thinking is wrong. The difference in current results from small differences between the LEDs, and the (potentially large) differences in current that may cause. Failure then follows if LEDs overheat. As long as each LED stays within its rated power, no problem.

With 6 LEDs in parallel, the worst case is 1 LED eating all the current that was meant to go through 6 LEDs. With 2 LEDs in parallel, the worst case is 1 LED eating 2x the current that was meant to go through one LED.

But those worse cases rarely occur, unless LEDs with very different characteristics are placed in parallel. Or a LED totally dies, wiring breaks, etc. In practice, how well the characteristics of the LEDs are matched, will determine how even current is divided between them. LEDs with near-perfect match of characteristics -> no problem. Of course working conditions can be slightly different for each LED, the LEDs will age (and age differently), and then troubles begin.

If LEDs are not run @ full power but adequately cooled, and say 2x the normal current would be okay, you could place 2 LEDs in parallel. And then those pairs of LEDs in series. If 3x the normal current is okay, you could place 3 LEDs in parallel, and a # of those 'triplets' in series. And so forth.


Care to explain why? No room for series resistors? (tip: in a pinch, even wiring itself could be used as a small series resistor to even out differences between LEDs - just use thinner wiring and/or wires of a worse-conducting material than copper. Same for pcb tracks). No driver with high enough output voltage available? No single LED with the required output available?

Retro - I'm trying to squeeze every last bit of efficiency out of the flashlight. I thought adding series resistors, no matter how small would be working against me and add complication. Keeping in mind that the array is being powered/regulated by a circuit (no more than 300 mA shared by all LEDs) and if I were to run them in series, what is the minimum resistor recommended to even out differences between LEDs? I'd like to get a sense of how much additional energy would be burned up by adding these 6 resistors.

6x low power LEDs are needed for coverage and efficiency. That's why a single LED is not being used.

 

[SemiMan]

Thanks everyone for the replies. This has been helpful.

Electrically there is no difference between A and B so I'll likely move forward with B out of simplicity.



Retro - I'm trying to squeeze every last bit of efficiency out of the flashlight. I thought adding series resistors, no matter how small would be working against me and add complication. Keeping in mind that the array is being powered/regulated by a circuit (no more than 300 mA shared by all LEDs) and if I were to run them in series, what is the minimum resistor recommended to even out differences between LEDs? I'd like to get a sense of how much additional energy would be burned up by adding these 6 resistors.

6x low power LEDs are needed for coverage and efficiency. That's why a single LED is not being used.

That dreaded efficiency ...which if a concern you should be using a buck/boost regulator so you get as perfect of matching between the battery and LED as possible ... assuming the right choice of regulator. A bad choice can make it worse.

Realistically, your eyes can barely tell a 30% difference in intensity, so worrying about a few percent is a waste of time. If that is the worry, then use 12 LEDs at 1/2 the power and you will get 10-15% better efficiency in all likelihood. Other option is buying a better bin.

 

[RetroTechie]

I'm trying to squeeze every last bit of efficiency out of the flashlight.

In that case: put all LEDs in series. Start with the highest voltage you can use as a power source. Use a dedicated, high-efficiency LED driver that outputs the exact current you want to go through each LED (read: no series resistor(s), driver controls LED current). And which is designed to work most efficient around the in/output voltage ranges used. Buck if Vin​ > Vout​, boost if Vin​ < Vout​, buck-boost if input and output voltage ranges overlap. A linear driver can be very efficient if input voltage and (total) LED voltage are close together, wasteful if not.

Putting LEDs in parallel means you work with relatively high currents & low voltages, which is contra-productive when highest efficiency is desired. Wiring is also simpler with a series configuration.

Whether 6 separate LEDs are more efficient than 1 high-power one, depends entirely on the specific LEDs selected, the current they're run at, mounting/cooling options etc.

So I'm with SemiMan here. I suspect you're looking at this 'problem' from the wrong angle. It would help if you could share some more details concerning intended power source, the specific flashlight host, LEDs you have in mind, etc.