"10 watt" LED

"And furthermore (Jacques Clouseau :)"

May I suggest bench testing samples using a precision power supply?

The 10 watt chips are The Wild West. Take 2 different samples DUT

Apply threshold voltage to illuminate: Note value.

Apply each to a given heat sink...

Note Delta T at recommended drive current...
...then back off the high temp DUT to = temp of the lower chip

Does the illumination seem to be the same?

This exposes (strengths - weakness) in manufacturing techniques

MEXICOWANDERER wrote:
What is missing is compensation for a two volt spread in supply voltage. This is the primary justification for using PWM rather than pure resistance

x2 ... active circuitry, either linear or switching is much better suited to an automotive environment that is an RV. I designed my personal fixtures to handle at least 18 volts input, with 35 volt ceramic caps on the input stage.

Lets not even get started on load dumps and other transients that SAE specs worry about :)

GordonThree wrote:
3s 3p led array

Vf(typ) 3.3vdc per chip, x 3 in series = 9.9 volts. Let's round to 10

Ohms law P=V x I ... So 10=10 x I, so I = 1amp

So let's say 13v supply voltage means a resistor needs to drop 3 volts.

Ohms law V=I x R ... So 3=1 x R, thus R=3.3 ohms (nearest easy value)

Ohms law P=I²R ... So P=1 x 3.3, thus P=3.3 watts

I'm a bit rusty... What did I miss?

Of course you can under drive the module to reduce loss but also output.

Technically, your math is right but your ASSUMPTIONS ARE WRONG bringing the entire equation into error.

Your assuming 3.3V per diode, that is not correct.

White LEDs can be had with voltages ranges from 3.1 up to 4.1V depending on the different materials used to make the diode junctions.

The 10W COB that the OP has, is most likely similar to the ones I have which are specd at 11.1V of the series string, that puts the diode junction at 3.7V.

Now, since I have messed with 10W, 20W, 30W and 50W versions I can't say for sure at this time the exact max current for each, I would have to dig around to find the manufacturers spec sheet and my notes.

I can tell you this, the 10W versions ARE 11.1V for full brightness and current draw. The 20W, 30W, 50W require 34V as a min voltage which to operate from 12V would need a booster power supply.

I did retrofit a couple of my TTs outdoor porch fixtures with the spare 10W versions since I didn't need a boost regulator.

Unfortunately, it IS a very dreary and wet evening tonight and the rain is forecasted through Sat so I am not about to bother getting a ladder out to pop the lens off to verify the resistor I used..

My point to you is a couple of watts heat wasted by a resistor is not in any way shape or form "substantial" compared to the wattage consumed by the COB.

Statistically speaking it can be said that the difference IS NOT STATISTICALLY SIGNIFICANT, since the increase in loss is very small compared to the overall power used by the entire device.

People tend to forget that a switching power supply regulator does indeed WASTE energy in the form of heat. It just does it a a SLIGHTLY LOWER AMOUNT.

Typical switching supplies often have efficiency levels of 80%-85%, some may reach 90%-92% but not very many do. That means a switching regulator will waste 15%-20% of the power that goes into it as heat..

If the power wasted by the resistor was EQUAL OR MORE than the COB then I would consider that to be STATISTICALLY SIGNIFICANT.

And YES, I ERROR on the safer side by UNDER DRIVING THE LED. There is a point where the current increases rapidly and yet the light gain is extremely small, at that point pushing it to max yields very little light gain and severely reduces the operating life..

That is intentional, and vastly increases the life of the LED (or any device for that matter).

This also allows for the fudge factor when things don’t go as planned.

Yup and with flaky chips that laugh at their spec sheet the bell-curve voltage/ma/lumen output relationship can assume the absurdity of Monte Hall, and let's-make-a-deal.

Lin, is whom set me up with SHARP chips. But they cannot compete with scumbag low-cost-at-a-high-price specials in eBay. Many 10-watt chips exhibit multiple colors and intensities throughout the plate.

CREE chips are sensitive and utterly unforgiving about drive current.

A sort-of-clue
A company that offers 365 nanometer grade ultra-violet chips usually is a lot stricter with chip quality. Costs more but the chips are uniformly top notch.

You mentioned the 20-100 watt family of chips. This is where I spend time with my lab power supply between 34-36 volts. The Sharp units seem to prefer 34.7 volts as the top of the bell compromise.

To suggest a Chinese company is going to use top drawer chips then custom power them to an optimum compromise would be really stretching it.

The Karma Sutra of Chinese LED philosophy seems to be "Buy an X wattage chip then drive it at 60% so maximum savings are realized on the cost of heat sinks. Frugality to the point of running the chip above the thermal upper limit.

Down here, one of the pastimes is noting exactly how the new LED streetlights have failed. flashing, mass area blowout, intermittent operation wotta hoot.

Then there are the new cars with interesting faults in LED tail lights.

I'll stick with my severely controlled engineering environment - thank you...