The use of LEDs as light sources has become more popular. In many cases, LEDs are now mandatory to reach energy efficiency and color rendering standards. Unlike incandescent bulbs, LED light engines feature light-emitting devices as well as a heatsink, a constant current driver, and sometimes secondary optics to fulfill the light pattern requirements. All of these components contribute to the lumen-per-watt efficacy of the system and need to be optimized.
Table of Contents
- Managing Junction Temperature
- Optic Selection
Long-term cost savings is a key motivation for introducing LED light engines. Savings result from low power consumption and low maintenance due to longer life with stable light output. These benefits have to be evaluated against the cost of LED light engines. Ultimately, good design is the key for success.
Managing Junction Temperature
LED system design can be approached from many angles. One way is to start with the maximum expected LED junction temperature, or TJ, which can be estimated by assuming a TJ max of 85°C. As higher junction temperatures reduce the LED light output and LED life, TJ becomes one of the central design parameters.
Along with potential light losses in the light path, junction temperature helps define the required number of LEDs to reach a specified lumen goal. In combination with the maximum ambient temperature, the TJ limit also determines the performance of the heat management system, which includes an LED, a printed-circuit board (PCB), a heatsink, and ambient air.
LED production lots differ in their lumen per watt (lm/W) efficacy. Therefore, the system has to pass the lumen requirements with the lowest LED efficacy values from the considered lumen bins. Further, as LEDs age, their light output can drop, which makes it important to know whether the lumen requirements are stated for a new light engine or for one that has been operating for a period of time, as well as whether the requirements are at room or maximum ambient temperature.
LED efficacy can be improved by applying a smaller current per square millimeter of LED chip. For example, the efficacy of an Osram Oslon SSL LED can be increased approximately 25% by reducing the current from 700 mA to 350 mA.1 Several factors affect this increase.
First, light output is less than linear with the LED current, so doubling the current produces less than twice the amount of light. Higher currents require a slightly larger LED voltage, and a larger current also leads to a higher junction temperature, causing thermal degradation of the light production. The application of less current per LED requires a larger combined LED chip area, which means using more LEDs or LEDs with larger chip sizes. This measure can be the difference in passing the light fixture requirements.
Secondary optics for LEDs can be used as powerful tools to manipulate the light pattern. Today’s lenses typically use a reflector for efficient light collection and a lens/diffuser combination on top to further shape the beam. Efficiencies of greater than 90% are common for acrylic lenses. Other optics employ only a reflector, often with a segmented mirror surface to prevent a picture projection of the LED. Diffusers by themselves can be customized to generate more complex light patterns like an elliptical beam field. Made out of plastic, these diffusers add very little cost per LED.
Also, secondary optics are designed to manipulate light and are optimized for low light loss. Light fixtures, however, contain other elements that can dramatically reduce the overall system efficiency. For example, cover glasses are added to protect the lamp’s light engine from outside elements. At each glass surface, a Fresnel reflection of about 4% or higher occurs. The exact reflection value depends on the refractive index of the glass and the angle between light and glass. This light gets thrown back into the lamp, and a white lamp interior, including a white PCB surface, can recover some of the light. Cover glasses made from highly transmissive glass types and anti-reflective layers can improve the performance.
Finally, the lamp’s mechanical structure, including mounting poles or decorative artifacts, can absorb part of the light as well. Lamps with a mounting pole directly beneath the light engine are particularly susceptible to losses, up to 20% in some cases. These factors make it increasingly important to seek innovative lamp designs to increase the amount of light hitting the ground.
LEDs typically are soldered onto a PCB. The soldering process is important for ceramic LED packages in particular as thermal expansion differences can stress the solder joints between the LED and a metal core board. Optimizing the soldering parameters, the solder paste or the dielectric material can strengthen solder joints.
1. A densely packaged two-color light engine, like the Osram Sylvania PrevaLED, works best with a metal-clad board.
Metal core boards have better heat conductivity than FR4 material equipped with thermal vias and are necessary for light engines with dense LED placement (Fig. 1). For larger PCB areas per LED—for example, one square inch or more per watt of heat—metal-clad boards reduce the junction temperature by only a few degrees over FR4. Here, the lower-cost FR4 board might be sufficient.
As discussed previously, a low LED junction temperature is beneficial for LED light output and is influenced by the performance of the heatsink. Thermal resistance characterizes heatsinks. In general, a heatsink is a compromise between good heat sinking, size, and costs.
Heatsinks transfer their heat to an outside medium like air. Since the thermal conductivity of air is low, convection dominates the mechanism. Heat transport increases based on the heatsink-to-air temperature difference, the exposed heatsink area, and potential airflow, which can be fan-driven.
Also, heatsinks produce their own micro airflow as lighter, warmer air moves up. This process makes convection and the heatsink performance orientation-dependent. A PCB temperature ends up lower for a heatsink with low thermal resistance, and systems allowing good vertical airflow usually perform best.
Because LED PCBs are often screwed onto the heatsink, enough screws need to be in place so the PCB doesn’t warp at high or low temperatures. A good interface material such as thermal conductive tape or grease between the PCB and heatsink is beneficial because it increases the heat-conductive cross section between the two components.
Due to the complexity of the heatsink’s function, it should be tested or simulated to optimize performance, size, and shape. Many heatsink suppliers can help with application support. It should also be noted that dust will settle on the surface of the heatsink and that the heatsink may require a protective paint, both potentially reducing performance.
To verify the heat management performance, thermocouples are popular for checking temperature. Though they provide straightforward measurements, they aren’t always practical to use. To be accurate, thermocouples need to have good contact with the object, read one PCB spot only, and add wires to the test setup.
Thermal imaging works much faster, as it immediately reveals all the hotspots on the PCB or a heatsink in operation. For proper function, the imager should be calibrated and operated with the right emissivity factor as it detects radiation, which depends on the surface’s temperature and its capability to radiate. For example, unpainted shiny metal surfaces radiate significantly less than painted metal or PCB material. Respective factors are available from thermal imager suppliers.
As part of the lighting system, LED drivers are another factor in the overall efficiency of the solution and need be optimized. LEDs are best driven by constant current sources as the LED light output is fairly linear with the current.
Switching current sources turn their output on and off while an inductor and capacitor limit and store the LED current, creating a constant current with reduced ripple that is largely independent of the supply voltage. Driver efficiencies of 90% or more can currently be reached. For a light fixture, the constant current source needs to be integrated in a module that might, for example, need to satisfy the following requirements:
• High efficiency, i.e., greater than 85% when dimmable, greater than 90% otherwise
• High power factor correction, greater than 90%
• Total harmonic distortion of less than 20% at full load
• Thermal shutdown protection
• Open output protection (no LEDs attached)
• UL listed
• Meets applicable Energy Star requirements
• Lifetime comparable to LED life requirements
• Compact size to fit in various light fixtures
• Rated for operation in a specific environment
• Easy to replace
• Low bill of material
• Five-year warranty
Though Energy Star compliance might not be required in all cases, it should be considered a goal since LED light engines are designed to save power and provide high-quality lighting.
In the past, it was common in light engines with multiple LEDs to place the LEDs in several parallel serial strings and to adjust the current per string by serial resistors. In other cases, the differences in forward voltage and the potential difference in current per string were simply ignored. The result was either a lower efficiency due to serial resistors or transistors, or uneven light production within the light engine.
2. In a linear light strip supporting Dragon LEDs, individual constant current sources control serial strings of LEDs.
Today, many switching current sources are offered with several, individually controlled constant current outputs. With this concept the LEDs are still organized in serial strings, but each string runs with a constant current at a high efficiency level (Fig. 2).
There have been many advances in LEDs, lenses, and LED driver technology. In each area, high-efficiency components are available. The light engine designer should strive to optimize each system component as the product of the efficiencies determines the final lumen-per-watt level of the light fixture.
Awesome post. Like it very much.