Modular design techniques overcome tunable SSL development obstacles (MAGAZINE)

Sept. 19, 2017
Emerging tunable lighting applications are creating demand to drive down development cost and complexity. Mark Shepherd illustrates how a closed-loop system designed with smart lighting modules simplifies LED specification.

Emerging tunable lighting applications are creating demand to drive down development cost and complexity. MARK SHEPHERD illustrates how a closed-loop system designed with smart lighting modules simplifies LED specification.

Human bodies have always been regulated by the 24-hour cycle of night and day. Our internal circadian clock, which takes its principal timing cue from the types and patterns of light and darkness, plays an important role in controlling the body's cycles of activity and rest. Mounting evidence from scientific studies appears to confirm that interfering with the circadian clock poses a risk to people's wellbeing. One example of this: Exposure to the wrong type of artificial light in the evening can impair quality of sleep and even health. Increasingly, the lighting community is pursuing new LED-based products that can have a positive impact on human health and wellbeing. Developing such products is a complex task, but here we will discuss cost-effective avenues for realization of tunable-white solid-state lighting (SSL).

Interested in articles & announcements on tunable lighting and other smart lighting technologies?

The potential health effects of lighting are widely applicable: In the US, the Institute of Medicine has estimated that between 50 million and 70 million Americans suffer ill health and risks to safety from sleep disorders and sleep deficiency. These effects include higher risk of heart disease, stroke, depression, obesity, and diabetes.

FIG. 1. A module with an intelligent lighting manager and integrated color sensor is mounted behind a diffuser for this tunable lighting design.

It's also important to note, of course, the aforementioned positive effects on human behavior that spectrally-controlled artificial lighting can have on people. For instance, cooler or bluer hues of light are believed to promote more sustained focus on work tasks than warmer, yellower hues, so the design of lighting schemes has an important effect on productivity in the workplace. Terms including human-centric lighting (HCL) or lighting for health and wellbeing are being used to describe this emerging lighting application.

Incandescent sources hid the problem and opportunity

In the past, the research into the circadian effects of artificial lighting would have been of mostly academic interest, since most indoor lighting was provided by incandescent sources of light with a fixed warm correlated color temperature (CCT) of 2700K, or by fluorescent light sources typically with a cooler, fixed CCT of more than 4000K. The CCT of a light source is a single number value used by the lighting industry as a shorthand way of denoting its color appearance. It provides a useful guide to the relative warmth or coolness of white light, but lacks the depth of information about chromaticity offered by x,y color coordinates mapped against the CIE 1931 chromaticity diagram.

The introduction of modern LED light sources, however, has given the lighting industry the opportunity to provide control over the color temperature of their products' light output. LEDs are readily available in a wide range of CCT options stretching from 2700K to 6500K. The intensity of an LED's output is also easily controlled by familiar analog or digital dimming techniques.

As a result, it is now technically possible to design a luminaire that provides a variable or tunable white-light output — a tunable white point. This evolution in design allows manufacturers to market products which, for instance, can change the CCT of their light output over the course of the day, in tune with the user's circadian rhythm, or to enable workplace lighting to be adjusted dynamically to balance workers' comfort and productivity.

FIG. 2. The CIE 1931 color space, showing the black-body locus and the chromaticity range of certain CCT values. The calibrated-for-life color sensor discussed here measures x,y coordinates typically to within four MacAdam steps. (A MacAdam step marks the point at which two colors are barely able to be distinguished by the average human eye).

Implementing the control element of a tunable white-light design poses unfamiliar problems, however, for system designers who have only previously developed products with a fixed CCT. This article describes the methods available today for realizing tunable white lighting, and introduces one potential implementation scheme that provides both precision and control while freeing the system designer of the need to implement complex algorithms.

Potential paths to tunable white

There is no limit to the number of ways a user might want the CCT of a white lighting system to be made adjustable. The most common requirement is likely to be for lighting that is harmonized with the natural circadian rhythm. In simple terms, this means providing bluer, cooler white light with a relatively high CCT of around 4000K through most daylight hours, and gradually lowering the color temperature in the late afternoon and evening toward a warm-white CCT of 2700K by dusk.

A basic hardware architecture for this variable-CCT output includes a string of warm-white LEDs and a string of cool-white LEDs, both with a fixed CCT, which are mounted alongside each other behind a diffuser that mixes their light output before it reaches the illuminated plane.

On first consideration, the color control of such an architecture might seem simple. A low-cost 8-bit microcontroller (MCU) could easily adjust the drive current through each LED string with reference to a real-time clock, producing a greater proportion of the total light output from the cool-white string during daylight hours and a smaller proportion from the warm-white string, and vice versa in the evening. Since LED chips have experienced substantial price reductions over the past few years, such a system is cheap and simple to build.

Handling LED binning, ageing, and color shift

Such a system is, however, imprecise, unstable, and difficult to manufacture consistently in volume. The CCT specifications of LEDs can be highly variable across production batches, and LED manufacturers therefore supply LEDs in bins characterized by chromaticity. The MCU-based control system requires two LED strings each with a fixed, known CCT in order to achieve a given CCT at the illuminated plane. If the design assumes that the cool-white LED string has a CCT of 4000K, the tuned color output will be wrong if a unit's cool-white LED string is actually at 4100K because of variation in the bins used in the luminaire's production run.

This means that the luminaire manufacturer has to tightly specify the bin or bins from which the LEDs are supplied. Unfortunately, LED manufacturers charge a high premium for orders confined to a tightly specified bin. Limiting bin selection also exposes the luminaire manufacturer to supply-chain risk.

Worse, the LEDs' output is not in any case fixed: Both the CCT and the intensity of an LED's light output change perceptibly over time and temperature.

FIG. 3. External inputs to the proposed tunable lighting system architecture enable connection of external sensors, such as an ambient light sensor, as shown in this typical application circuit.

The simple MCU system described above is incapable of handling either variation in production units or ageing effects. If a luminaire manufacturer wants to market a product that offers the ability to precisely and accurately tune the color of white light for the lifetime of the luminaire, then a different solution is required.

A technique proposed by sensor manufacturer ams AG uses a closed-loop real-time feedback system based on measurements by a color sensor mounted with a view of the LED color mix, typically looking from behind toward the luminaire's diffuser, or into the reflective mixing chamber of an indirect or volumetric design (Fig. 1).

Considering a closed loop system

The architecture of this closed-loop system is as simple as that of the 8-bit MCU solution described previously, because all of the sensing and control functions are integrated into a single module, the AS7221 Smart Lighting Manager (SLM). But because the module includes a calibrated xyz chromatic white color sensor providing real-time measurements of the luminaire's light output, its operation is beyond the influence of LED current, temperature or ageing effects, or any other external influence on the LEDs' color temperature or intensity. It also virtually eliminates the effect of differences between one bin and another.

Developers have a lot of flexibility with a modular approach. Measuring just 4.5×4.7×2.5 mm and housed in a 20-pin land grid array (LGA) IC package, such a module can be mounted in any number of locations and orientations on the luminaire.

The module described here integrates a tri-stimulus color sensor that provides coordinates consistent with the CIE 1931 color space, which describes the human eye's perception of color ( With two strings of LEDs, warm and cool, the output of a luminaire may be made to track, in a fairly linear way, the black-body locus (Fig. 2), which marks the ideal target for white light over the range of CCT values from warm to cool white. The design is not limited to the control of two strings of LEDs, so if a third string of amber LEDs were added, for instance, it would be possible to track the black-body locus even more closely.

The means by which the SLM implements this tight color control is its "brain," an advanced Cognitive Lighting Engine (CLE). The CLE constantly processes the input from the integrated color sensor and compares it to the desired CCT value which, in a circadian-pattern light, changes during the course of the day. When it detects a variance between the actual CCT and the desired CCT, it adjusts the drive current to the two (or three) LED strings until the variance has been minimized to the extent possible with the chosen LEDs. This closed-loop feedback system operates continuously to maintain the light output at the correct color temperature and intensity.

LED intensity control

Control of the drive current to the LED strings is implemented via three channels of pulsewidth modulation (PWM) signals fed directly from the module to an LED driver. One of the PWM channels can optionally be operated as a 0–10V analog dimming output, allowing the other two channels to operate as a current steering mechanism, enabling the use of low-cost, constant-current power supplies.

This tunable lighting system has two important attributes that heighten its value to the developers of new LED lighting equipment. First, it is incredibly easy to implement. The developer does not have to implement an algorithm for calculating the temperature, ageing factors, or drive current at each LED string in response to a color sensor's measurements. The designer also does not have to implement a PWM controller. All the designer has to do is decide the target CCT values and program them into an external 4-MB serial flash (nonvolatile) memory array that connects via a serial peripheral (SPI) bus to the SLM.

FIG. 4. A development board allows product developers to jumpstart tunable white luminaire projects.

Second, the AS7221 provides a ready-made framework for implementing environment-aware lighting (Fig. 3). It offers inputs for external sensors such as occupancy (presence), internal air quality monitoring, and other types of sensors, enabling the implementation to operate as an Internet of Things (IoT) sensor hub. Users can also connect an ambient light sensor (ALS), such as the ams TSL4531, to implement daylight harvesting alongside the color tuning function. The energy and cost savings provided by daylight harvesting can often help the end user to recoup the additional cost of smart or tunable white lighting within a short period.

Control and commissioning

A serial interface is provided for configuration, control, and management of the tunable lighting system. This allows it to be easily networked, for instance, via a Bluetooth or ZigBee radio module or other wired or wireless communications interface. The SLM can in fact operate in standalone mode. Because it can be networked, however, it is also capable of pushing data to a central controller for data analytics purposes.

The proposed architecture relies on a smart lighting command set based on simple AT-style text commands to configure and control a wide variety of functions such as on/off, dimming and lumen maintenance (AT was the command set used by dial-up modems). A development kit from ams, the Smart Lighting Integration Kit or SLIK, lets the designer connect the AS7221 to a luminaire quickly and easily (Fig. 4). Traditional wall-mounted dimmers, occupancy sensors, and LED drive ballast units can all be wired to this board.

A library of configuration elements — smart lighting commands that define the transitions between different color and luminance settings — is included in the application programming interface (API) that ams provides with the SLM. These APIs enable the user to configure parameters such as the start time of transitions, the transition speed, and initial and final luminance levels. A PC-based graphical user interface (GUI) also gives the user the capability to set up system tests and to log test data.


The critical breakthrough in tunable white lighting provided by the architecture proposed in this article is to make continuous and accurate closed-loop control easy to implement, requiring only a single module and associated flash memory to be mounted close to the strings of emitters. The combination of an accurate, calibrated-for-life color sensor and a Cognitive Light Engine in a single module implements the entire color tuning function, and also provides the flexibility to connect external sensors for environment awareness. Such an architecture promises to enable the development of a new generation of smart, tunable lighting products providing more comfortable and more productive indoor spaces that promote the wellbeing of their users.

MARK SHEPHERD is a staff field applications engineer for ams (