The First Commercial 200 Lumen/Watt HB-LED

Coming off its record-breaking 303 lm/W lab device, Cree Inc. has announced the 200 lm/W XLamp XP-L discrete LED, almost doubling the efficacy stakes in the high-power LED market.

Cree XLamp XP-L (not to scale with following illustration).

The single-die device, available in Cree's 3.45mm x 3.45mm XP family package, offers an Optical Control Factor among the highest (for a more compact and controllable optical system design), and delivers upwards of 460 hot (85C) lumens in a 65 CRI, cool white CCT (5,000K-8,300K), with 1,050 mA of drive current. The device is specified at a maximum drive current of 3 amps, meaning north of 1,200 lumens is available from a single 3-volt device. The devices are also binned in warmer CCTs, including an 80 CRI white (2,700K-4,300K) CCT range that delivers 380 hot lumens (1,050mA) and a 90 CRI white (2,600K-3,200K) that puts out 340 lumens at 1,050mA.

Wow.

This means that what was typically done with the company's single-die XLamp XM-L in a 5mm x 5mm package or the quad-die XM-L EasyWhite and Philips LumiLEDs (sponsor of this site) most recent high-power LED, the Luxeon MZ, can now deliver the same class of lumens with a much smaller emitter surface, resulting in a higher Optical Control Factor.

Philips Lumileds Luxeon MZ (not to scale with preceding illustration).

Wow.

Perspective

But, this is all from a LED-centric supplier's perspective, not that of a system designer. While it's fine to tout super efficacy in the LED device, what customers see is the system wall plug efficacy. Yes, if you're doing a 1,200-lumen design, you can do it now with one device that has its light emanating from a much smaller surface area than a quad-die solution. But, you need to deliver about 2.5 amps in a single-die 3V device, instead of the 800mA or so you'd need with a quad-die 12V device.

Apart from the parasitic loss of the driver control electronics and all the other parts you need to support triac dimming, the largest losses in an LED driver are typically in the Rdson of the switching device and in the ohmic and core losses of the magnetics. Taking the highest-efficacy topology, the buck topology, cranking the current up by 3x increases the power losses in the MOSFET and inductor by... are you ready?... 9x.

So, if you were tickled that you managed to find yourself a 1.5-ohm MOSFET and a 1.5-ohm inductor and had about 2W of losses with those "crappy, old," 100 lm/W 12V devices to deliver 960 lumens and 12W at the wall plug, you now have to provide about 2.5 amps at 3V to deliver the same lumens to the optics, about 7.6W into the LEDs. Marketing is going to recommend you for a bonus for that kind of efficacy.

But, wait a minute

Hang on. 2.5 amps into that 1.5-ohm, 400V MOSFET alone is 9W of additional loss. That's over 18W if you can use the same inductor, which you can't, because the current will saturate the core. So, now, you need a MOSFET rated for much higher current, 3x the size. Since there's no free lunch, you just increased the gate capacitance, increasing the MOSFET switching losses and drive requirements.

Oh, by the way, that MOSFET's area just increased by about 3x, so count on paying for that one. That higher-current inductor you found has much lower inductance to deliver that current in the same volume, which means much higher LED current ripple. You could go with a much higher volumetric device, for which you pay money per cubic centimeter of copper and ferrite.

You were further ahead with the likes of the 12V Luxeon MZ. Cree keeps playing the game of ever-increasing device current for bragging rights and to support some mythical $/k-lumen price point, but who can afford to use those devices? Granted, the XM-L's 200 lm/W at 350mA is very, very usable. The 3 amps and 1,200 lumens, or even 480 lumens at 1 amp -- either case is nothing short of marketing bull to prop up average selling price and stock price, from this system designer's perspective.