Leds in parallel?

[misterT]

yes I did write the report. There is no other report about this on the www, and led foundrys wont write about it in the detail required.

I found plenty:
https://ieeexplore.ieee.org/search/searchresult.jsp?newsearch=true&queryText=led parallel&x=-1154&y=-177

 

[Nigel Goodwin]

we all know that series parallel led arrays without limiting resistors must use vf matched leds....

No we don't, that's just YOUR made up theory - the Philips datasheet only referred to the same model of LED, no matching required - as long as you had enough in each series chain (I 'think' it was a minimum of 4).

 

[Flyback]

I do have parallel LEDs in production and they are sorted to with in 100mV

..please don't answer if its your IP, but does your company sort them to 100mV, or does the LED foundry do that for you.?
-also, do you operate each led at near the knee, or at rated current for that led.?
-Given that led vf varies after being turned on as the led warms up, how long after switch on do you measure the vf?
-At what led current do you measure the vf (near the knee current, or rated current)?
-Whats the junction to ambient thermal resistance of the leds that you parallel?
-How close are the leds mounted to each other in the application's pcb
-how many leds comprise each of the series strings that gets paralleled?
-do you do extensive production testing to assure good current sharing between the paralleled leds?
-do you have any series limiting resistance in your series paralell array (ie in each series led string)?

MisterT: your links actually support NOT using series parallel led arrays without limiting resistors/circuitry...most of your links concern current mirror led current equalisation schemes.

Nigel Goodwin..I did read the Philips report that you previously linked..i didn't find any clear indication that it said long led strings in parallel were fine to do without limiting resistors.
Even the most hardened parallel led enthusiast would call for voltage matched leds for parallel operation without limiting reistors.

**broken link removed**
...this is from Philips and they don't anywhere recommend series parallel array operation without current limiting circuitry.

 

[misterT]

MisterT: your links actually support NOT using series parallel led arrays without limiting resistors/circuitry...most of your links concern current mirror led current equalisation schemes.

All papers that provide some sort of solution to the problem, also include a study of the problem itself.. that is how a good scientific paper is written. And many times you find great studies in the references also. I thought you were interested about the nature of the problem, not solutions.

 

[Flyback]

OK thanks MisterT, I should have said I am (and many others are too) specifically interested in series parallel led array operation whereby no limiting resistances or current equalisation circuitry is used in any of the series led strings.

This would seem to be the cheapest way of driving loads of leds from a single current source, and we all know that with 2 cent Far Eastern LED lighting products hitting the market every day, everyone needs to find the cheapest way of doing it.

 

[misterT]

OK thanks MisterT, I should have said I am (and many others are too) specifically interested in series parallel led array operation whereby no limiting resistances or current equalisation circuitry is used in any of the series led strings.

This would seem to be the cheapest way of driving loads of leds from a single current source, and we all know that with 2 cent Far Eastern LED lighting products hitting the market every day, everyone needs to find the cheapest way of doing it.

There are many kinds of "cheap". I would say that the cheapest is the most reliable solution that does not require testing and measuring of all the products at manufacturing line (finding matching led-strings).
And I don't think "everyone" should find the cheapest way. Market like this gets dominated quickly by couple of large companies that sell the cheapest products.. then there are smaller companies that make quality products for those who need them and for those who are willing to pay for quality.

Also, I think there is a need for some kind of standard, so that the LED drivers could be integrated into the sockets instead of the "bulbs".

 

[Flyback]

There are many kinds of "cheap". I would say that the cheapest is the most reliable solution that does not require testing and measuring of all the products at manufacturing line (finding matching led-strings).

Of course, In the Far East, its very cheap to employ hundreds of lowly payed people to do just that at the production line.

 

[misterT]

Maybe this is interesting to you:

 

[alec_t]

I wish to evaluate it before doing it..

Why don't you just do a test to convince yourself, by connecting ~50 nominally identical LEDs in parallel, with no individual current-limiting resistors, and see what happens? If any survive that test , repeat the experiment but this time with paralleled strings, each string having, say, 10 LEDs in series.

 

[killivolt]

Wash, Rinse, Repeat.

Edit: I'm waiting for my simulation to load, I just keep reading this thread over and over.

 

[Nigel Goodwin]

Maybe this is interesting to you:

Pretty cool

A far more sensible 'report' of LED lighting than some might present

 

[Flyback]

Alec_T..
I don't do a test because its a batch thing, doing the test with a different batch with rogue non-same-batched leds having crept in could reveal different results.
MisterT.., I enjoyed vid
There is of course no performance data for that product. The products brand, "muller light" is a contrived name, probably set up by someone who would simply close down that brand if it wasn't successful in the field.
The leds were each operated well below rated current, so whether from batch to batch you get the sufficient amount of lumens is still in question.
The auther himself suspects the leds are Chinese imitations of a Philips led......maybe the Chinese added series resistance to each led , we don't know.
Also, the LEDs are in a string of 24 in series. -then 4 of these are paralleled without added limiting resistance, though of course, the stray resistance of such a long string will be relatively high, so that helps.
The input voltage, is 230VAC, and this gives a rectified DC of some 270VDC.............The question remains as to why they didn't do two strings of 48 leds in series, then parallel these, since that would have been possible...though maybe the switching frequency would have been too low with thetype of controller used.
So in short, we have no long term success data for the product, we also don't know whether it was actually made at the Chinese led foundry itself, which if it was, would mean there was certainty about all the leds being from the same batch..we also don't have lifetime data, because leds (even from the batch) degenerate there specs at different rates, and so the sharing of current may become unequal as time goes on.
So in short, there's no well-kown brand (eg Thorn, GE, Tridonic, Sylvania, Harvard) attached to this thing, and maybe one of those set up a separate company and called it "muller light" just in case somewhere down the line, a whole huge production batch of these things fail..if that happened, the brand "muller light" would simply disappear and you know what would happen if you wanted your money back.
Also, the question now goes out, is 24 leds in series the minimum acceptable number?
Anyway, are they also “doctored” leds comprising enhanmced resistance…..and what really is the lumen per watt output of those leds, which as confessed, are running at well below their spec’d current.

By the way, muller light doesn't come up on google search....accept as yoghurt.
Its probably a "made in china"name, or rather made in china by a western entrepreneur who went to china and asked them to build it there, so that if it went wrong then he can jack out of it pronto.

 

[misterT]

By the way, muller light doesn't come up on google search....accept as yoghurt.
Its probably a "made in china"name, or rather made in china by a western entrepreneur who went to china and asked them to build it there, so that if it went wrong then he can jack out of it pronto.

http://www.mueller-licht.de/

 

[Flyback]

ok, it doesn't of course , mean that its not secretly owned by one of the big lighting conglomerates who are using it as a vehicle to bring in these cheap Chinese concepts of series/parallel LED banks....because the big western guys like Thorn and GE are not doing parallel/series led banks without limiting resistors.
I just wish I could be a fly on the wall in the place where they are made to see if they are made in the actual Chinese led foundry itself, and whether they are doctoring the leds with resistance.

 

[ronsimpson]

..please don't answer if its your IP, but does your company sort them to 100mV, or does the LED foundry do that for you.?

both, Mosly get the foundry to sort.

-also, do you operate each led at near the knee, or at rated current for that led.?

at rated current

-Given that led vf varies after being turned on as the led warms up, how long after switch on do you measure the vf?

If the girls sort, about 1 second, if the machine sorts 0.1 second. I don't care how they are testes just that they are sorted the same way

-At what led current do you measure the vf (near the knee current, or rated current)?

rated current

-Whats the junction to ambient thermal resistance of the leds that you parallel?

? Some products not good and some have large heat sinks.

-How close are the leds mounted to each other in the application's pcb

very close

-how many leds comprise each of the series strings that gets paralleled?

parallel LEDs are 3 in parallel. in a different product there are 8 series four times with some resistance.

-do you do extensive production testing to assure good current sharing between the paralleled leds?

in the first year of production YES but not now that the problems went away.

-do you have any series limiting resistance in your series parallel array (ie in each series led string)?

in the parallel series, yes, the resistors are of a low value.

All LEDs are driven with a constant current switching power supply.

 

[Flyback]

Thanks Ron Simpson, when you just do "three series LEDs" in parallel, that's not many leds...why don't you just series them all?
Also,aren't matched leds more expensive thannon matched leds,?., so wouldn't you be better off just using one series string, or say three series strings each with its own switch mode current regulator?

 

[alec_t]

I don't do a test because its a batch thing, doing the test with a different batch with rogue non-same-batched leds having crept in could reveal different results.

But you've already said that you can't guarantee the same Vf, even from the same batch. You don't trust suppliers apparently. So in the real world if you want to parallel LEDs you are going to have to do it with non-matched LEDs.