Resources | LpR Article | Technologies | Applications | Jan 15, 2018

Multi-Pixel LED Technology Opens New Horizons for Smart Lighting Applications

The evolution of Multi-Pixel LED technology has initiated a giant leap in the development of intelligent lighting systems which are most visible in the automotive industry. Now the first hybrid LED provides smart headlights with more than 1000 individually controllable pixels. Ralph Bertram, who is working on advanced LED device concepts, and Norbert Harendt, who is developing optics solutions for general lighting at Osram Opto Semiconductors, show that automotive lighting is just one of the potential areas in which intelligent selective pixel control can be applied. Options for the use in general lighting, such as information display for outdoor, indoor, retail or industrial applications, are very versatile.

Smart lighting has become an increasingly visible trend topic, not only drawing the attention of the industry as an attractive growth market, but also of the general public. Trends such as smart homes, IoT penetration, and advancements in LED technology are some of the key factors driving the market growth. Energy savings made possible by highly energy efficient LED technology, intelligent control mechanisms and sensors to regulate light settings for user comfort - as well as new concepts like Human Centric Lighting - are at the forefront of consumer’s minds.

Up to now, adaptive lighting includes changing the light intensity and color in response to usage parameters like occupancy or time of day. But there are many additional application fields in which intelligent spatially adaptive lighting can provide numerous benefits to end users, and open new potential markets for solution providers.

The automotive industry traditionally leads the way in many technological developments. Today’s cars often represent the most sophisticated technology owned by many consumers. Virtually every aspect of the modern automobile is high-tech, and uses state-of-the-art materials and solutions. As a result, cars have become a platform showcasing the evolution of technology as well as the potential and progress among engineers and innovators. Adaptive Frontlighting Systems (AFS), for example, help to increase driving and road safety. These systems adjust the direction of the light to offer drivers the best possible visibility by illuminating curve progressions, the side of the road, or help to protect oncoming traffic from glare through the so-called adaptive driving beam (ADB). This concept of adaptable light beams and emission characteristics of light sources is now also finding its way into General Lighting in application fields including shop lighting, or lighting solutions for hotels, gastronomy or business environments.

So, why didn’t the transfer to spatially adaptive lighting happen earlier? Most probably because the technological challenges are even higher than with other lighting systems. Changing the path of light requires moving parts such as tilting mirrors or shifting lenses. In the past it simply wasn’t possible to provide an affordable, efficient and adequately robust solution which would also fulfill the long operation lifetime required in professional lighting systems. With the continuing miniaturization of LED technology, new possibilities are arising that promise to add another dimension to adaptive lights: changing the beam pattern of a light source without moving parts.

Applications Fields and Requirements

Smart spatial lighting installations are already commonplace in some applications: corridors light up when you walk by, parking lots are only illuminated where people are present and the lights above an office desk or in the pantry area automatically dim down when not in use.

All this can be done by adding sensors and intelligence to both existing and new luminaires. For example, in corridors it is sufficient to light up or dim down ranges of 3-5 meters. In the office, it is already a different story: some people might want to light up only their desk and not the area around it. Others want to have a light beam on the paper they are working on. Thinking about airplane, trains or car interiors, it might be required to light up only the crossword that a passenger is solving, and not disturb the passenger in the next seat who would like to sleep. This is also a useful feature for home lighting – for example if you like reading at night while your partner goes to sleep early.

Museums or restaurants have different requirements with regards to adaptable lighting. Today, they usually have to pick one lighting arrangement with more or less flexibility through track systems where luminaires can be manually adjusted. If the arrangement of their tables or art on display is changed, the originally chosen light arrangement often doesn’t fit very well. To change the light fixtures, however, is time consuming and thus costly. To avoid this, novel lighting systems require digital control and a much finer granularity of the light spots, which makes it hard or impossible to realize with common means.

In shop displays, the benefits of spatially adaptive lighting are even more apparent: setting light accents by a tap on a tablet computer, even from a remote location, can greatly save effort and money associated with setting up the merchandise display. It also enables the user to change the look and feel of the display at an ad-hoc notice without even touching it.

Technical Challenges

Each of these application fields has different requirements towards the necessary light levels as well as the “granularity” – meaning the size of each “pixel” that is to be illuminated. In table 1, we tried to estimate typical scenes including feature sizes, the light source distances and the resulting beam angles.

Screen Shot 2018-01-15 at 13.44.18.pngTable 1: Estimated “granularity” (=smallest spot on target) and required optical parameters for different applications

Looking at the set of parameters, the biggest challenge lies in the very fine granularity of each single beam. Osram has started to demonstrate this kind of application scenario with the “Omnipoint” concept in 2015 [1]. In this scenario an assembly of almost 100 Oslon Square High- Power LEDs is placed on a hollow sphere in a downlight configuration. Each LED, equipped with its own narrow-beam optics, points to a different direction in the room. By switching or dimming each LED individually, the light distribution in the room can be smoothly adapted.

Screen Shot 2018-01-15 at 13.44.23.pngFigure 1: Spot size on the target equals “granularity“ in a multi-spot source

With this enormous effort in terms of mechanics and optics, the fixture is able to fulfil the requirements for a demanding office application as explained above and gets closer to fulfilling shop lighting specification. However, applications aiming at small object illumination, like a reading light or sophisticated merchandize illumination, demand even more pixels and smaller beams. In order to develop a system fulfilling these requirements, a different, more integrated technical approach is necessary.

More but smaller pixels call for a dense arrangement of LEDs that can be addressed individually. Miniaturization of the system and its components is therefore a key requirement, which would allow the use of common optics for hundreds of light sources. In turn, this is essential to keep the system simple and affordable.

Looking at the requirements in table 1, the necessary light levels per pixel for many applications can be fulfilled by using an arrangement of 1 mm. LEDs, each able to deliver a typical 100 lm, going up to 300 lm in overdrive mode.

However, in order to illuminate the whole room with these beams, hundreds or even thousands of pixels are necessary. Thus, it is essential to have high-luminance emitters that can be packed very closely together. Figure 3 shows an array of a new chip sized “Package” under development, which is not larger than the chip itself. It has, in fact, been designed as a surface-emitting chip without any package or frame around it. Its compactness is perfectly suited for dense arrays while still having a size manageable by standard SMD equipment.

Screen Shot 2018-01-15 at 13.59.50.pngFigure 2: Detail of the “Omnipoint” demonstrator: Each LED with its own lens shines into a different angle and therefore illuminates a different area in the room

An array of this new clusterable LED provides the best possible compact size realizable by LED components and will help to fulfil the requirements of numerous application scenarios. But even with this accomplishment, in some cases the overall LED array will amount to a size difficult to manage by optics. Additionally, driving the LEDs would still need to be realized in a passive matrix arrangement with outside electronics. The next step in the integration process will therefore be the multi-pixel LED.

Screen Shot 2018-01-15 at 13.39.00.pngFigure 3: Array of surface-emitting Chip Sized Packages

Heading Towards the Multi-Pixel Light Source

To date, adaptive LED lighting systems, including automotive headlamps, have operated with individually controlled chips for each illuminated area. Now the evolution of multi-pixel LED technology is initiating a giant leap in the development of intelligent lighting systems, which are observable in the automotive industry.

Under the “μAFS” research project (pronounced “micro AFS”), funded by the German Ministry of Research and Education a group of German companies worked for three and a half years up to September 2016 on the groundwork for a new class of energy-efficient LED headlamps for adaptive front-lighting systems. Osram Opto Semiconductors took on the leading role as project coordinator and contributed its extensive expertise in the area of LED light solutions for the automotive sector as well as in the area of chip and conversion technologies.

The project partners, have developed a pixel-light source with 1024 individually controllable light points. These provide about 3 lumens (lm) at only 11 milliamperes (mA) for an individual pixel surface of 0.115x0.115 mm from a closed emission surface of 4.00x4.00 mm with a grid size of 0.125 mm. They are arranged in an array of 32x32 on an active matrix IC, so each pixel can be individually addressed. Developed originally for the automotive headlamp application, it can possibly also enable the fine granularity needed for shoplighting and reading light applications.

Another major advantage are the control options, e.g. through the interaction between a camera and a controller. The camera acts as the “eyes” of the system, capturing the information about the surroundings and forwarding it to the controller. This “brain” processes the information and forwards a suitably adapted light distribution pattern to the pixels in digital format. Each of the pixels can be switched on and off with different currents more than one hundred thousand times a second, and can therefore be dimmed. Depending on the situation, the system decides which pixels will be affected. In automotive applications, traffic signs (for example) will be illuminated so drivers can see them clearly without being dazzled by the reflected glare from their own headlights.

Automotive lighting however is just one of the potential areas in which intelligent selective pixel control can be applied. Options for the use in general lighting, such as information display for outdoor, indoor, retail or industrial applications, are very versatile.

Challenges and Opportunities for Optics Designers

Moving from discrete lenses to an integrated array, we also face significant challenges in optics design. Illumination optics in the traditional sense always is about blurring the light source and not creating an image of it on the wall. It is also about removing any focal points so a spotlight emits a single, homogeneous, slightly diverging beam.

If we want to create an optical system that can send light from different pixels into different directions, we need to get back to an imaging system again. In optics terms, this is the simple task to convert a spatial pattern into an angular pattern.

The task is not much different from what a wide angle (“fisheye”) lens does when used for camera imaging in the opposite way: it projects rays coming from different angles to different camera pixels on the detector. It is also not so much different from an image projector optics that brings light from different points on the image to different angles in the room.

However, using LEDs as a source, the major difference is the acceptance angle as illustrated in Figure 4: For a camera lens, it does not matter at what angle the rays hit the sensor. In projectors, the light is usually pre-collimated so the image is formed from light that is more or less parallel and the optics only needs to accept a limited range of incoming angles. In addition, efficiency is not the primary target for these systems.

Screen Shot 2018-01-15 at 13.39.29.pngFigure 4: Difference between an imaging camera system (left) and an illumination system (right) - note the large angular aperture needed at the light source of the illumination system

For our system of pixels formed by single LEDs, each pixel emits light into the full hemisphere. Thus, the optics needs to not only transform a spatial into an angular pattern but also catch as much of the light emitted as possible, - also the light emitted to the sides - and bring it to the right direction. This is definitely not possible by standard optics and needs a complex optics design with several, relatively large lenses. Since it is still an illumination optics, ultimate requirements towards image quality are not necessary. However, the demands for color correction are high in order to transfer a 3 MacAdams steps distribution of the light source to the imaging area. Either way, a slight “blurring” of the image is required to hide the structure of the LED chips in the room.

This is definitely a new, interesting task for optics designers: merging the “two worlds” of imaging optics and illumination optics to create really novel solutions for spatially adaptive illumination.

Competing Technologies

Switching pixels on and off does not require new technology. Digital projectors, both with LCD as well as micromirror technology, are available and offer millions of pixels. However, these are devices designed for the display of information, not for illumination purposes. Thus, they operate with RGB colors - which is perfect to display images but will create a horrible color impression when used as a source to light up a room. In addition to that, they operate by permanently creating a high light level and absorbing the light again at pixels that should be dark. This is not only highly inefficient, it also ends up with limited black and white contrast.

Using an LED light source where pixels only light up when their light is actually needed is the enabler for an adaptive system energy-efficient enough to be used for General Lighting purposes. It also allows for the design of systems with smaller heat sinks, and passive cooling without the offending noise of a fan is possible.

Since the technology is extremely robust, it can also be utilized in more harsh environments like outdoor lighting. It could therefore be introduced into architecture lighting or mounted on moving fixtures in stage and movie lights.

Screen Shot 2018-01-15 at 13.40.30.pngFigure 5: Artist's view of an adaptive shoplight system


When realizing “adaptive lighting”, light levels and color temperatures as well as the distribution of light emitted by each single luminaire can be changed to really create different lighting scenes during the course of the day or be adapted to the situation.

First demonstration systems by O researchers based on single LEDs have been showcased with the “Omnipoint” system, which impressively demonstrated the concept. Since then, miniaturization of light sources enables shrinking of form factors and adding more and more features. Realization of designs with chip sized packages (CSP), especially with the most compact new LED generations, allows adding more and more pixels and a flat design. In the future we will see even more pixels and integrated (active matrix) designs when LED sizes really enter the micrometer scale.

Multi-pixel LEDs are in the early stage of coming to the market with additional options for applications in general lighting. This technology will take lighting to the next level since it adds another dimension to adaptive lighting: spatial steering of the light, without any moving parts or compromise to the high level of energy efficiency we are used to with LEDs.

[1] Video reference:

(c) Luger Research e.U. - 2018