LEDs in parallel...well thermally coupled due to being in "tube"

Hello,
This concerns a LED lamp with Vin = 20V to 140V, and output needs to be 20W.
We had a LED lamp done for us. They did a Buck converter with output of about 12V...it actually is a buck converter with an output current regulated to 1.66A….its made up of paralleled strings of three_LEDs_in_series. Actually ten strings of three_LEDs_in_series…all paralleled. There are no dropping resistors to equalise the currents.
It is said that equal string current is obtained by using matched LEDs.
Also, since the LEDs are totally enclosed in a plastic tube, (it’s a tube light) and sit on a metal core PCB, it is said that the LEDs are all obviously going to be well thermally coupled.
I suppose its true that the LEDs are likely be well thermally coupled, since they are all encased in this totally un-ventilated enclosure……the LEDs are about 2.5cm apart, but still, the total enclosure in this internally hot enclosure should ensure good thermal coupling.
However, I must admit I would have done a buckboost converter and put all leds in a single series string.
But am i just being old-fashioned?
Is this the age of leds_in_parallel with no dropping resistors, and just rely on the encasement in a non ventilated enclosure to give good thermal coupling between paralleled leds

Also, you make sure the LEDs are operated at the lower end of the current range.....where voltage changes more with changes in current.....so eg the luxeon rebel plus led woudl be used with just 160mA in it...(its max is 700mA)

LUXEON REBEL LED
https://www.lumileds.com/uploads/380/DS107-pdf

...this inceases chances of sharing of current..because the led voltage changes more as the led carries more current, and this voltage gets onto the other leds and they start cinducting more, so you dont get so much higging if led current.

I ma worried that they put just 3 in the paralleled series strings, because the more leds the better for good current sharing. (due to vf spread probability)

Do you at least agree in pronciple that thermal runaway is not possible for any of the paralleled leds in the tube?, ...because they are all 'frying' in each others heat?...despite being 2.5cm apart

Hi,

LED forward voltage varies with temperature (and other parameters)
The relevant temperature is the die temperature. Not ambient temperature, not heatsink temperature.
Maybe (and I doubt this) you can ensure equal heatsink temperature...but this doesn't mean equal die temperature, because there is a thermal resistance between die and heatsink.

Assuming same heatsink temperature..then the LED with the higher power dissipation (or the higher current) will have the higher die temperature. This will lead to lower forward voltage and thus to higher LED current...and thus to even higher power dissipation...and so on....
It's like an avalanche.

The system is more critical with:
* higher thermal drift of forward voltage (I see no way to influence this)
* higher thermal resistance of die to ambient/heatsink (improve thermal coupling)
* lower series resistance of LEDs. (Adding a series resistor is a simple and often used method)

Klaus

Thanks, though Rth(jc) is about 4degC per watt for the leds.
So the die temperature will never be able to get that far above the die temperature of the other leds for it to be a problem......especially since we are running the leds at only 16% of nominal current, and Vf increases more with current at these low levels...which will mean less current and avoiding thermal runaway.

Luxeon rebel plus LEDs
https://www.lumileds.com/uploads/380/DS107-pdf

I answered on the other forum. Why are there so many thread duplicates on more than one forum?

I also have noticed that.