Why closed-loop sensing provides the best white color-tuning performance in LED lights at the lowest cost

July 26, 2018 Tom Griffiths, ams AG

The lighting industry has recognized the value of color tuning to support added-value applications such as human-centric lighting. There is a stark difference between the three models of color tuning control in use today. This article describes them and shows why one model – sensor-enabled closed-loop control – is superior in terms of performance and cost. It also describes the optical design considerations which need to be taken into account when adding closed-loop control components to a lighting fixture.

With the rapid decline in the dollar-per-kilolumen price of LEDs, the economics of LED lighting have turned decisively in favor of white color tuning in new commercial and high-end residential lighting installations.

For instance, a typical office-lighting troffer can have a selling price between $175 and $350; the linear strip of LEDs in it will generally contribute less than $5 to the bill-of-materials (BoM) cost. This huge differential between the cost of the main component and the price the user pays for the fixture gives the manufacturer scope to add features and value, and thereby resist pricing pressure from low-cost, and potentially lower quality, competitors.

Color tuning is a particularly attractive way to differentiate an otherwise standard lighting fixture. The term color tuning generally denotes the ability to adjust the white correlated color temperature (CCT) of a fixture’s light output. This would typically be bounded by very cool-white at a CCT of up to 6500K and very warm-white with a CCT of 2200K. A typical tunable white luminaire would employ less extreme tuning in a range from an incandescent-like 2700K to the 5000K favored for office lighting. In the example of the commercial troffer, this requires dual linear strings of LEDs at CCT values of ≤2700K and ≥5000K, and a sophisticated control circuit.

Even if this raises the troffer’s manufactured cost for LEDs and electronics to $30, it is still a small fraction of the selling price. And the color tuning capability allows the fixture manufacturer to market the light under the banner of “human-centric lighting” (HCL). HCL is aimed at enabling the lights to cover CCT ranges which can help support the human body’s natural circadian rhythm.

For example, a tunable white system could be configured to produce a cooler (blue-white) light in the middle daylight hours, gradually warming the color temperature in the approach to dusk to give warm (yellow-white) light, potentially mitigating some of the blue light influence at night (see Figure 1). Studies show that such adjustment of the color of artificial light produces hormonal responses in humans which promote productivity and alertness in daylight hours, and which assist relaxation and sleep in the night.

[Figure 1 | HCL systems adjust lighting from cooler to warmer color temperatures towards dusk.]

In other words, HCL offers the promise of increased user comfort and, potentially, productivity as well as health-supporting benefits when compared with fluorescent or other non-tunable lights. These attractive benefits can help sustain the premium pricing of many of today’s commercial and residential lighting products.

Now, the question for the lighting industry is shifting from whether to implement color tuning, to how to implement it. There is a stark difference between the three models of color tuning control in use today. This article describes them and shows why one model – sensor-enabled closed-loop control – is superior in terms of performance and cost. It also describes the optical design considerations when adding closed-loop control components to a lighting fixture.

The three methods of controlling tunable white lighting

The most basic, and least accurate, approach to tunable white lighting control is to manage the luminaire’s warm and cool LED strings with a small microcontroller which implements a simple look-up table, to calculate the ratio of drive current through the warm and cool LED strings needed to achieve an approximate target CCT value. With careful selection and characterization of the LEDs, and high-precision power systems (LED drivers), this approach can provide an adequate result when new, and in laboratory conditions. But in the real world, the variation in LED behavior over time and under different operating conditions exposes the limitations in the look-up table.

For instance, an LED's color point shifts and its flux varies from the nominal values when dimmed, or when the ambient temperature differs from the nominal value specified by the manufacturer. This variability can seriously compromise the accuracy of the system's color control. The human eye is sensitive to light-to-light variations as small as a few tenths of one percent, so changes are often immediately perceptible. These variances only become more pronounced over time as aging adds to the divergence between the LEDs' initial characterisation and their actual performance.

The second approach to color-tuning systems is a more sophisticated version of the first: a look-up table-driven compensation system. This approach models the predicted variations in LED behavior caused by changes in operating temperature and by aging and incorporates them into a more complex look-up table. When combined with current-feedback circuitry to better manage the LED driver’s current variation, this approach produces more accurate color control than the first approach, and more sustainable performance. When this approach is properly implemented, light-by-light calibration is performed on the assembly line to cancel out the effect of the small variations in components from one production unit to another.

While the performance of this method is superior than the first, it has serious drawbacks:

  • Added component cost for temperature and current compensation
  • High-precision, multi-channel LED drivers
  • High dependency on the initial LED characterisation and aging predictions. Over a 10-year luminaire lifetime, almost any prediction model will show some variation from real-world performance. And because the LED market is dynamic, the model is likely to be derived from an earlier device than that integrated into the luminaire, potentially causing further variance from actual performance.
  • The luminaire manufacturer is locked into the initial component choice, since any change requires a complete re-characterization of the revised system design to achieve repeatable results.

Closed-loop control method

The third approach takes continual measurements of the blended light using a calibrated color sensor within the luminaire to regulate a closed-loop feedback control system. In this type of implementation, the microprocessor and algorithms make continuous, active adjustments to the drive current supplied to each string of LEDs to keep the mixed light output at its target CCT value. The control loop continues sensing the light output and adjusting the power inputs constantly to maintain the target color temperature: color shifts caused by temperature swings, aging or dimming operations are continuously compensated for.

This feedback system needs to know nothing about the LEDs it controls - not even their nominal color temperature. It needs no predictive models of the LEDs’ behaviour over time and temperature. It does not need carefully selected batches of LEDs from tightly specified bins at an exact color point. And with a pre-calibrated sensing element, the luminaires require no calibration on the production line.

Instead, the luminaire manufacturer can choose LEDs from a wider range of color bins at each end of the color temperature scale, and the mix of bins can change over time with no effect on the CCT performance of the color tuning control system. Provided the color sensing element which drives the control loop can maintain its calibration over life, no further calibration of production units is required.

The performance of the closed-loop system will be superior to that of the first two approaches when the luminaire is new. That’s because the control loop measures actual light output in practice, regardless of operating conditions, rather than relying on a model for what the light output should be in theory under assumed conditions. It is an adaptive rather than a predictive method.

This superiority over other methods will grow over time, since the closed-loop method will perform as well after a normal LED luminaire lifetime as it does when new, while the performance of the first two approaches necessarily degrades over time.

The superior performance is also provided at lower cost, because of the looser specification of the LEDs and power system. By widening the LED binning range, the luminaire manufacturer has the freedom to choose LEDs from lower-cost or more readily available bins to the left and right of the tunable CCT range, reducing overall exposure to supply-chain risk. In addition, inventory management is greatly simplified, and its cost is reduced. And by eliminating the need for production line calibration, end-product assembly time and cost are reduced.

Finally, the luminaire manufacturer can continually update the board design to replace older LEDs with newly introduced, improved products, without any need to go through lengthy and costly processes to characterize the new LEDs and develop new look-up tables.

An integrated approach to closed-loop sensing

It is quite possible to implement a closed-loop control system for tunable white lighting with a few discrete components: a high-quality color sensor, a microcontroller and in-house developed algorithms for adjusting the drive current to the LEDs in response to the sensor measurements.

The development of this software is a complex and difficult task, however. It requires expertise in a variety of disciplines including dimming/tuning methodologies (for a smooth tuning effect), embedded programming, optical design, optical system characterization, and closed loop methodologies to prevent oscillation or synchronization defects. As is often the case in electronics system design, an integrated system-on-chip (SoC) solution for closed-loop control of color tuning saves space and cost and is quicker and easier for the system designer to implement.

Such an integrated solution is available from ams, a supplier of high-performance color sensors to the smartphone and instrumentation markets. It provides two types of color tuning SoC: The Smart Lighting Manager and the Smart Lighting Director. The manager type, such as the AS7221, includes a small processor core integrated with an XYZ color sensor in a 4.5mm x 4.7mm x 2.5mm package (see Figure 2). An XYZ sensor has a color response curve that mimics that of the human eye. Its highly robust interference filters, fabricated on-wafer as part of the CMOS processing, are extremely stable over both temperature and time.

[Figure 2 | The compact AS7221 is easily accommodated in downlights and other space-constrained designs.]

In its most cost-effective implementation, the Smart Lighting Manager has a multi-channel output, providing either a standard 0-10V or PWM-type dimming signal to the LED driver to determine overall brightness values. In addition, two other PWM channels would control switching transistors at the bottom of the warm and cool LED strings, steering the current using a complementary duty cycle to achieve the on-target CCT mix of the two LED strings.

A smart lighting director, such as the AS7225, also includes the high-performance XYZ color sensor and digital logic to continually calculate the needed ratio between the LED strings. The results of these continuous calculations are supplied to an in-luminaire application processor in the form of directives to tune each string to a specified percentage value. It would then be the job of the application processor to manage the transition from one directed string mix to another, as well as managing DALI or other network signals.

The use of a Smart Lighting Manager such as the AS7221 offers the easiest and quickest way to implement a closed-loop control system and requires no other control components. The ams Smart Lighting Manager also includes native support for several add-on sensors, including an ambient light sensor to add daylight harvesting capabilities, as well as environmental sensors such as temperature, humidity and VOC-based air quality sensors. 

The use of a Smart Lighting Director such as the AS7225 is ideal for smart luminaire designs that already have a discrete microcontroller and gives the luminaire manufacturer more flexibility to add other sensors controlled by the MCU.

Optical design considerations

In a closed-loop control system, the sensor must have a representative view of the mixed light output emitted from the dual strings of LEDs.

To obtain a view of fully mixed light, it is also possible to mount the sensor on the LED board in view of a light guide mounted at the edge of the luminaire’s reflector or housing (see Figure 3).

[Figure 3 | A light guide directs light from the edge of a downlight’s reflector to the color sensor on the LED board.]

The ams sensors are supplied with a 45° field of view and include configuration controls to allow adaptation to different field of view or mixing-chamber characteristics.


The addition of closed-loop sensing to an LED luminaire enables direct BoM and manufactured-cost reductions, both for directly tunable lighting systems and for those which use the sensor loop to maintain a single CCT or lumen output target over the lifetime of the luminaire. For these tunable systems, a closed-loop system enables the use of less precisely specified components while producing more precise color tuning.

Lighting manufacturers should expect to gain additional competitive advantages and premium pricing opportunities as the result of implementing added-value applications enabled by color tuning. These include circadian-optimized and other HCL applications which could improve workforce productivity and user satisfaction.

Tom Griffiths is Senior Marketing Manager for Sensor Driven Lighting and Spectral Sensing at ams AG. He has spent the past 18 years as a strategic consultant, communicator and evangelist immersed in LEDs, LED lighting, and the associated markets, including his roles as the founder and publisher of LIGHTimes, Solid State Lighting Design and CompoundSemi Online. Tom holds a bachelor’s and a master’s degree in economics from the University of California.

Previous Article
How fog computing disrupts the cloud services market

For business, one unresolved challenge remains important: providing scalability for cloud technology.

Next Article
Monetizing blockchain-based M2M service
Monetizing blockchain-based M2M service

Today, several publications and other media outlets writing about how to apply blockchain in fog computing ...