LEDs in parallel offer better value for money?

[Nigel Goodwin]

...and you are saying that its only the LEDs that are available to manufacturers in large quantities that can be used in this way? (3 in series chains, paralleled up)

No, that was for single selected LED's in parallel - not series chains.

Also, do you have one of these app notes we could look at?

I've not looked at one since a number of computers ago

But I seem to remember one of them was from Philips Semiconductors?.

I wasnt sure if you meant its ok to parallel chains of 3-in-series of any led, and whether or not you meant that paralleling the "3-chains" can be done without using a series resistor in each "3-chain"?

Presumably the LED's need to be of the same type, so use identical LED's from the same manufacturer - no separate resistors are needed for each separate chain, just current limiting for the entire assembly.

What are you trying to do anyway?.

 

[Flyback]

What are you trying to do anyway?.

...thinking about whether or not parallel led lamps need redesigning...they have no series resistors. They just use LEDs which have the same part number, goodness knows if the pcb assembly house bother to make sure all the parallel leds come from the same reel or wafer

 

[kinarfi]

I have 2 lights on my off road vehicle that have 24 3 watt LEDs each and at present, I am running them as 6 strings of 4 in parallel and no balance resistors, they are rated at 800ma max and even when the engine is running and the alternating maintaining voltage, they are running at about 750ma. Each LED is rated at ~3.6V and so far, my only fear is I don't have an adequate heat sink, they do dim out some when the engine is off or just idling.
So what I'm saying is that I think it is possible to run LEDs from a voltage source if you know it won't exceed what the LED can handle and I do believe that strings of series'd LEDs can be paralleled and that the more in the string, the more the voltage can vary, as long as you keep the max V some what below the accumulated voltages of the LEDs in the string.
Kinarfi

 

[Flyback]

ok thanks

my only fear is I don't have an adequate heat sink

....i see your point, and thats a pretty big fear, because overheating is about the worst thing that can happen to LEDs.

 

[Nigel Goodwin]

...thinking about whether or not parallel led lamps need redesigning...

What do you mean by 'LED lamps'? - do you simply mean torches/flashlights?.

 

[Flyback]

yes, generally....many companies, eager to jump on the LED bandwagon, but clueless about how to do a LED driver, have used parallel LEDs in combo with hysteretic mode led drivers, very cheap, but very dodgy , and not robust.

LEDs in parallel is bad news...wholesale...why have to worry about super thermal coupling when you can just put them in series. As you know, superb thermal coupling is a must with leds in parallel.

 

[audioguru]

I do not believe that LED forward voltages are random. An LED or a string of LEDs need a series resistor to limit and determine the current.

What if all the LEDs at your distributor are new and are from the same batch with most low voltage or most high voltage?

What if a huge product manufacturer took all the low voltage or high voltage ones?

Then the LEDs you get will be extremely bright and burn out soon or they will be very dim.

 

[Flyback]

audioguru i respsect what you say.


..but this product (SD24) uses three paralleled strings of 8-LEDs-in-series..................

http://www.strobe.com/component/opt...duct_id,154/category_id,57/manufacturer_id,0/

also here...

https://www.electro-tech-online.com/custompdfs/2013/02/install-_sd24.pdf

..LED string currents (if balanced) are 150mA.......each of the three series strings has a 10R, 1206 resistor to equalise the currents...but this seems rather small.
The 150mA is the LED current that flows when the LEDs are ON, and its worth remembering that this product is generally used as a flasher, so it would only be on for some 50% of the time.....also, even when on steady burn, it may be being pwm dimmed at 50%, say....so maybe its somewhat less than 150mA in each LED.

...but SD24 is a MASSIVELY popular product......and apparently failure rates are close to none.

....the paralleling of the led strings means a lowered output voltage, and so a single stage boost converter can drive the leds, which is nice and cheap, as the boost can just use a cheap off the shelf inductor.

...try putting those 24 LEDs in series and driving them with anything anywhere near as cheap and small as what drives them in the SD24......the boost is indeed very small , which is ideal for this products use.....................as i said, LEDs in parallel is maybe not such a bad idea.....just look at sales volumes on SD24.

 

[Mr RB]

It's common in "pro" designed products to use LEDs in parallel, especially white LEDs. This is done in all sorts of torches and auto lights, camping lights etc.

Modern LEDs are made in batches where the Vf is reasonably reliable, and with white LEDs in particular the Vf turn-on "knee" is quite gentle and the Vf is high, so they are suitable for paralleling.

 

[Flyback]

and the Vf is high, so they are suitable for paralleling.

...i'm sure you realise that high Vf does not mean "suitability" for paralleling

...though i appreciate that paralleling high Vf LEDs is "especially" a good way to make the manufacture of the required switch-mode lamp driver cheaper and simpler.
-since high Vf LED strings can get rather high in voltage, and this voltage can be brought down by paralleling...(remember that cheap boost converters cant boost to very high voltages, due to topology limitations)

In other words, i know what you mean, but you have abbreviated your intended meaning a little too much, and it comes out a little ambiguous.

Modern LEDs are made in batches where the Vf is reasonably reliable

..i am also sure that you appreciate that there is no way of knowing whether or not your PCB assembly house have bothered to use LEDs from a single batch or not.
I would say that Western/European PCB assembly houses will say they have done this, but will not bother, -their margins are tight, due to Chinese competition, and they will simply use whichever reel of "same part number" LEDs they can source cheapest.....they may well not be from the same batch.

...another point is that all the Vf matching in the world is useless if you do not ensure tight thermal coupling between the paralleled LEDs aswell.

Tight thermal coupling is a job thats done by your lowly payed assembly staff....its them that will be trusted to thermal-glue your LED PCBs to the heatsink, and its them that you have to trust to evenly spread the thermal-glue behind the mounted LEDs, so as to give close thermal coupling of the LEDs.........and once they are glued down, there is no way for your quality staff to check the evennness of the thermal gluing, so you really are laying your trust (and your future) at the hands of your lowly payed assembly staff.
-The only way to check the evenness of thermal gluing is to rip the product to pieces and inspect it, -even then , the glue will have dried and you can't be sure about it.

You could do production tests involving running the LEDs and then measuring the current in all of the paralleled strings over a long period of time....but thats expensive.

So, to summarise, paralleling LEDs (with no series resistors to equalise the parallel currents) is an absolute last resort.
You would be best off avoiding paralleled LEDs like the plague.

 

[Nigel Goodwin]

.
So, to summarise, paralleling LEDs (with no series resistors to equalise the parallel currents) is an absolute last resort.
You would be best off avoiding paralleled LEDs like the plague.

Paralleling single LED's is only done for things like torches, where you have no headroom to enable series resistors, and rely on the internal resistance of the batteries to limit the current - it's also pretty probable that such LED's are matched, and designed specifically for that use.

As I've previously mentioned, paralleling series chains of three or more identical LED's is perfectly fine, and a technique suggested by the manufacturers.

 

[audioguru]

take the SD24 product....this uses parallel LEDs, and is a superb product....it does have series resistors though, but only tiny 1206 ones.

Its LEDs are not in parallel. Instead they have three strings of 8 LEDs in series and in series with a tiny 10 ohms resistor (as you said in your post #28).
They have white and blue ones so each LED must be at least 3.2V. The current in each string is 150mA so the resistor uses 1.5V.
But the minimum supply for the product is 10V so it must have a voltage boosting circuit with an output of at least 27.1V and less for red amber and yellow ones.
We do not know the manufacturer nor the part number of the LEDs and we do not know how well they are matched.

 

[Flyback]

audioguru,

Please excuse my abbreviated writing.....What i meant was that they are *not* in series......the constant current driver has its current divided into three parallel "chains"...and by the word "chain", i mean 8 LEDs in series with one resistor of 10R, 1206 size.

we do not know how well they are matched

....interesting point, because really, *who* does know whether they are matched or not?

-is some highly payed quality inspector scrupulously peering over the production line in the assembly house to ensure that the LEDs were all from the same batch?.....i doubt it.

If they need 6 reels of matched LEDs from the same batch to complete a production run........then what happens if they can only source five matched reels within a week, say?......do they tell the customer that they can't complete the job?, or do they buy the 5 matched reels, and sneakily purchase a 6th reel of "same part number" LEDs and pretend that it was from the same batch as the first five?

Hmmmmmmmmmmmm.............................do we still like paralleled , matched LED "chains"?

 

[audioguru]

The 8 LEDs in series have a current-limiting resistor in series so the LEDs do not need to be extremely well matched.

Most Western LED manufacturers (like Philips Luxeon) already match LEDs and sell them in BINs. There is a BIN number for forward voltage, color (cool white to warm white) and brightness.
Of course they might not have enough LEDs in one BIN number that you need because the BINs are different for each production run (they are not matched when they are all in the same batch).

 

[Nigel Goodwin]

Hmmmmmmmmmmmm.............................do we still like paralleled , matched LED "chains"?

For the third time - which you seem to keep ignoring?, there's no problem with series chains - check with the manufacturers who produce datasheets suggesting their use in that way.

Matched LED's are only required for single LED's in parallel, and perhaps pairs in parallel?.

 

[ronv]

**broken link removed**

Here is an example of binning. It is not as hard on a supplier as one might guess.
Almost any bin might work for you.
If you notice a single LED Vf spec is .4 volts while a series string of 6 is .5 volts. And they can actually use some that would be rejects to make up the upper and lower bins.

https://www.electro-tech-online.com/custompdfs/2013/02/XLampMX_BL.pdf

 

[Flyback]

Thanks ronv

near the end of the article you pointed out...it says

You should include a current regulator in every parallel branch.

....this is ambiguous, do they mean a current regulator for each series string which make up the parallel branches, or do they mean a current regulator for a "bank" of four parallel strings?

Interesting that the advice to use multiple current regulators, comes from National semi, who make and sell....yes, you've guessed it, current regulators for LEDs.

 

[ronv]

In each branch like your 10 ohm resistors.

Take a look at the edit on bins in my post above for more info.

 

[ronv]

Here is an example of what might happen with 3 different methods.
Note: It does not include what might happen with temperature.

 

[Flyback]

thanks ronv, but that is low current, presumably below or at the knee.....higher up the curve and current changes drastically with tiny changes in Vf