Falling LED prices are yielding an unexpected bonanza for system developers: a “virtual” gain in device efficiency, which can be as much as 50 percent, or even more in some cases.

LED systems have traditionally run below the LED datasheet efficiencies. In addition to power conversion losses, LED efficiency drops as they are driven harder, and also as they run hotter. The former phenomenon, called “droop,” follows a cube law (and is currently ascribed to a phenomenon known as indirect Auger recombination). A cutting-edge LED, when driven at its nameplate power rating, might output only 60 percent of its nameplate efficiency rating, which is often specified at reduced drive level. Similarly, other physics effects cause output to fall dramatically with increasing temperature. In a case recently brought to the attention of Greentech Media, a special-application LED lamp that used only passive fin-cooling showed a 30 percent increase in output when customers started pointing fans at the units.

High-power LED models from players such as Cree or Osram have cost as much as $4.00 in recent years. At such prices, the LED cost dominated the bill of materials for a product. Designers were under intense financial pressure to reduce the LED count by driving the devices harder. A part such as a Cree XPG might end up being driven at 1.0 or even 1.5 amps, when it could get rather higher efficiency at, say, 0.35 amps. Similarly, the cost of the LEDs left very little budget for cooling measures. Heatsinks were often undersized, and advanced measures such as heat pipes, nanomaterials, and smart fans were usually out of the question.

At Strategies in Light 2012, Philips speaker Dirk Vanderhaeghen, presenting on automotive headlights in the High-Brightness Technology session track, remarked, “The end result of practical realities is that designs have come out at little more than half the theoretical efficiency of the emitters.”

Think 60 lumens per watt, instead of 120.

Now, however, with 3 to 5 watt high-brightness LED prices dropping, in some cases as low as $0.90 (particularly for new, cost-optimized devices such as the Cree XBD), the game is rapidly changing. Designers can convince management to use more emitters, drive them less, and allocate budget for improved cooling. CEO Kevin Wells of Lumigrow Inc., a maker of high-power LED greenhouse fixtures, states, “We can now afford to crank the LED count in a way we never could before.”

In another recent example, a pending off-grid lighting design by Simplistic/Alglo, based on a Cree XPG, was initially specified to drive the LEDs at only 0.2 amps for extreme efficiency in a battery-dependent environment. While this was later increased somewhat, drive levels this low for an XPG would rarely have been dreamed of in previous years.

Power conversion may instead eventually come to dominate the budget of the LEDs (as with solar PV), causing a whole new wave of innovation pressure.

The hidden bonanza of “virtual efficiency” will be an important game changer for LED applications, particularly as Haitz’s law (the Moore’s law of LED) reaches physics limits to efficiency in the coming years.

With the Department of Energy setting an LED manufacturing cost reduction goal of 20 to 1 for the coming decade, drive levels low enough to yield over 200 lumens/watt are likely to become feasible.

Tags: bridgelux, efficiency, leds, light emitting diodes, lighting, solid state lighting