WO2008139369A1

WO2008139369A1 - Lighting device with a plurality of light emitters

Info

Publication number: WO2008139369A1
Application number: PCT/IB2008/051753
Authority: WO
Inventors: Bernd Ackermann; Peter H. F. Deurenberg; Gerardus H. F. W. Steenbruggen; Wido Van Duijneveldt
Original assignee: Philips Intellectual Property & Standards Gmbh; Koninklijke Philips Electronics N. V.
Priority date: 2007-05-10
Filing date: 2008-05-06
Publication date: 2008-11-20
Also published as: TW200904251A

Abstract

The invention relates to a lighting device (100, 200, 300) with N light emitters (11) (e.g. LEDs) of different primary colors (e.g. R, A, G, B, C) that shall be controlled in such a way that a number k of target values (e.g. color point and/or brightness) are optimally matched, wherein k is typically smaller than N. The control problem is solved by applying at least two different control schemes simultaneously to different light emitters or during different operating conditions to all light emitters. In a first particular embodiment (100), some of the light emitters (G1) may be controlled by a feed forward controller (Cff) while the rest (G2) is controlled by a feedback controller (Cfb). In another embodiment (200), the driving commands of some of the light emitters (G1, G2) may be coupled, thus lumping these light emitters to a virtual light emitter for the purpose of control. In still another embodiment (300), different optimization criteria (C1, C2) are pursued under different operating conditions (Op1, Op2), for example the optimization of color rendering index on the black-body line and the optimization of lumen output elsewhere.

Description

LIGHTING DEVICE WITH A PLURALITY OF LIGHT EMITTERS

FIELD OF THE INVENTION

The invention relates to a lighting device with a plurality of light emitters, particularly LEDs, and means for controlling the light emitters individually according to given target values.

BACKGROUND OF THE INVENTION

From the US 6 507 159 B2, a lighting device comprising red, green and blue light emitting diodes (LEDs) is known, wherein the LEDs are controlled in a feedback loop such that given target tristimulus values are optimally matched. The effectiveness of such a lighting device with respect to color gamut and color rendering is however limited as only three primary colors are available.

SUMMARY OF THE INVENTION

Based on this situation it was an object of the present invention to provide a lighting device with improved characteristics, particularly with a large color gamut and good color rendering properties.

This objective is achieved by a lighting device according to claim 1. Preferred embodiments are disclosed in the dependent claims.

The lighting device according to the present invention comprises the following components: a) A number N > 2 of light emitters with different primary colors, i.e. with different emission spectra under comparable operating conditions (temperature, driving currents, etc). Each light emitter may be a single lamp or a combination of several, identical or distinct lamps. Moreover, it is understood that the light output of the whole lighting device is the superposition of the light output of all its N light emitters. b) A controller for selectively driving the aforementioned light emitters such that a number k > 1 of given target values (e.g. coordinates of a desired color point) are optimally matched by the common light output of the light emitters, wherein at least two different control schemes are applied. In this respect, an "optimal match" means that the light output of the lighting device (i) exactly meets the target values, or (ii) approaches said values as close as possible (e.g. approaches a given color point in a predetermined color space with a predetermined metric of color-distances as close as it is possible with the used light emitters).

Moreover, a "control scheme" shall denote any unique algorithm or mathematical equation by which the driving commands for a light emitter can be calculated based on actual input signals (e.g. the target values and measurement signals) and given parameters (e.g. characteristics of the light emitter). Two control schemes are then considered to be different if the associated equations are structurally different (thus a mere difference in parameters would not make the equations different; moreover, it should be noted that the equations shall not comprise a case sensitive branching as this would allow to formally combine completely different control approaches into one single equation). Examples of control schemes will be described in more detail with respect to different embodiments of the invention.

With respect to the number k of target values and the number N of light emitters, three cases can be distinguished: If k > N, there are generally not enough independent control variables (i.e. primary colors) to match all target values; an optimal approximation of the target values can then be tried instead. If k = N, there is generally a unique set of driving commands of the light emitters by which the target values can be reproduced. Finally, if k < N, the available number of primary colors provides excess degrees of freedom. The target values can therefore generally be reproduced, but the control problem becomes non-trivial. The proposed lighting device with its application of at least two different control schemes is particularly suited to manage the latter case.

It should be noted that the number k of target values defined above usually refers to targets values that can be measured using a sensor or can be derived from sensor measurements in real-time using e.g. an observer algorithm. If e.g. color point and brightness of a lamp are measured using a color sensor, then it will not be possible to optimally match e.g. a target of good color rendering since this can not be derived in real-time from measured color coordinates and brightness. k=3 will apply in this situation. For a lamp with N>3 primary colors it will then be possible to optimize the color rendering properties in an approximate way by taking into account the results of offline calculations in the design of one of the two control schemes as outlined in more detail below.

In a first basic variant of the present invention, the lighting device is characterized in that there are at least two groups of light emitters, wherein each group is associated to another control scheme that is applied to all light emitters of said group. With other words, the driving commands of the light emitters of one group are calculated by one equation, while the driving commands of the light emitters of the other group are calculated by a different equation. If there are more primary colors N than target values k, the division of the light emitters into several groups can particularly be done such that the excess degrees of freedom can be handled.

In a particular realization of the aforementioned first basic variant, the controller of the lighting device comprises a feed forward controller and a feedback controller, wherein one group of light emitters is controlled with a (pure) feed forward control scheme by the feed forward controller, while the other group of light emitters is controlled with a feedback control scheme by the feedback controller. By applying a feed forward control to some of the light emitters, the feedback control objective for the residual light emitters can be significantly simplified. It should be noted that the feedback controller may internally comprise a feed forward component, too, e.g. for providing basic control signals which are fine-tuned in a feedback loop. Said feed forward component may optionally apply the same feed forward algorithm as the feed forward controller does for the other group of light emitters. Preferably, there are just as many light emitters in the aforementioned feedback controlled group as there are target values, i.e. a number of k. Thus the control objective of matching k target values can uniquely be achieved with the available primary colors, allowing a well-defined design of the feedback loop. The lighting device with the feedback controller preferably comprises a sensor that is coupled to said feedback controller and that can determine the brightness and/or the color point of the common light output of the (active) light emitters. The sensor thus measures important characteristic values of the common light output which are often also (directly or indirectly) specified as target values. The mentioned feed forward controller may be designed in many different ways. Preferably, it is designed such that it optimizes the color rendering properties, particularly the color rendering index (CRI), and/or the power efficiency of the lighting device. These target values can be optimized offline using for example a model of the lamp or experiments. As the related optimization problem is however rather complicated (multidimensional, nonlinear, etc.), it is practically impossible to solve it exactly in realtime. The theoretical optimum for target values like CRI and efficiency is therefore preferably used in the feed forward controller to reduce the control problem to a number and kind of control parameters and target values that can be easily handled. These remaining control parameters and target values, e.g. brightness and/or color point, are then dealt with in the feedback controller.

The mentioned feedback controller is preferably coupled with the feed forward controller in such a way that the target signal and/or the feedback signal which is provided to the feedback controller are freed from components that are already handled by the feed forward controller. In particular, the target signals that are provided to the feedback controller should be the difference between the basic target values and that part of the basic target values that is already achieved by the feed forward control. Similarly, the feedback signal that reports the actual state of the control system to the feedback controller should be the overall measurement signals minus the components of these signals that are caused by the feed forward control loop. The feedback controller can thus be focused on that part of the control task that still remains to be solved after the feed forward control has been done. In a second realization of the first basic variant of the present invention with differently controlled groups of light emitters, there is at least one group of light emitters for which the driving commands are at least temporarily coupled by a static relation (and accordingly another group of light emitters for which the driving commands are uncoupled or coupled by a different static relation). The static relation allows to calculate the driving commands of all light emitters of said group from the driving command of only one light emitter of the group. The static relation therefore freezes the degrees of freedom of the considered light emitters, thus simplifying the control problem significantly. While the aforementioned static relation between the driving commands is in principle arbitrary, the driving commands of all light emitters of the group preferably have a fixed ratio with respect to each other. In particular, all driving commands may have the same value or the ratios are chosen such that combined with a subsequent higher level control it results in optimum color rendering properties and/or power efficiency of the lighting device.

In a further development of the aforementioned embodiments, all light emitters of the considered group (i.e. the driving commands of which are coupled by a static relation) are treated like one virtual light emitter in a higher level control scheme by a higher level controller. The higher level control scheme may then simply treat real light emitters and the virtual light emitter (which consists of several real light emitters) structurally in the same way. The higher level control scheme may particularly be designed such that it optimizes brightness and/or color point of the lighting device.

The aforementioned higher level controller may optionally provide a feed forward control of all light emitters based on at least some of the target values, e.g. brightness and color point. Moreover, the feed forward control may be based on operating parameters of the lighting device, for example on its temperature. A straightforward feed forward control (e.g. by matrix multiplication) is often available if the control problem is well-defined, i.e. if there are no excess degrees of freedom. This condition can be achieved by the proposed "freezing" of several degrees of freedom with a static relation. The aforementioned higher level feed forward control scheme may further be complemented by a feedback controller acting on at least some of the light emitters. Said feedback control may operate independently of the higher level control scheme, i.e. control different light emitters, or it may operate in cooperation with the higher level control scheme and for example fine-tune its feed forward commands.

In the following, a second basic variant to realize a lighting device with different control schemes will be described, wherein it should be noted that all described embodiments of the invention can be combined. According to this variant, there are (at least) a first region of operating conditions and a second region of operating conditions of the lighting device in which different control schemes are applied. The individual light emitters are therefore treated similarly at a given point in time, but this treatment may structurally change if other operating conditions are reached. The "operating conditions" may for example comprise the target values, the driving commands of the light emitters, the actual light output of the lighting device, environmental conditions like temperature and/or any other parameter that has an influence on the operation of the lighting device. In general, any kind of control scheme may be applied in the first and the second region of operating conditions, respectively, for example feed forward vs. feedback control or an individual vs. a lumped control of several light emitters (cf. discussion above). In a preferred embodiment, the control schemes applied in the first and the second region of operating conditions, respectively, differ in the optimization criteria for the common light output of the light emitters that are pursued. This allows to further improve the control of the lighting device as a situation-dependent optimization can be realized.

In a particular embodiment, the optimization of the color rendering properties is given priority in the first region of operating conditions. The color rendering properties can for example be quantified by the color rendering index and describe the capability of a light source to reproduce the colors of various objects. As the term "priority" indicates, the optimization of the color rendering properties may not be the sole optimization criterion, however the one that receives the largest weight in a weighted combination of optimization criteria. Similarly, the optimization of the lumen output can be given priority in the second region of operating conditions.

According to a further development, the control scheme changes continuously in an intermediate region of operating conditions between the first and the second region of operating conditions. Thus a continuous transition between the different control schemes can be achieved.

The first region of operating conditions may particularly comprise the black-body line of the common (target or actual) light output of the light emitters. The black-body line represents the color points of a radiating black-body at different temperatures and is particularly important for the generation of white light.

The second region of operating conditions may accordingly comprise all color points of the common (target or actual) light output of the light emitters that have more than a predetermined distance from the black-body line.

In the lighting device with different control schemes in different regions of operating conditions, at least some of the light emitters may be controlled with a feedback control scheme by a feedback controller. Such a feedback control allows to match the target values in spite of temperature changes, aging of the light emitters and the like.

The controller of the lighting device optionally comprises a memory in which a look-up-table is stored that contains control parameters, especially for a feed forward control. The use of a look-up-table allows a real-time implementation of complex control schemes, for example an optimization of the color rendering index, wherein intermediate values may be determined by a simple interpolation.

While the light emitters may in principle be any kind of lamp (or group of lamps), it is preferred that they comprise LEDs, phosphor converted LEDs, organic LEDs (OLEDs), LASERs, phosphor converted LASERs, colored fluorescent lamps, filtered (colored) halogen lamps, filtered (colored) high intensity discharge (HID) lamps, and/or filtered (colored) Ultra High Performance (UHP) lamps.

The target values may particularly comprise the color point, the brightness, the color rendering index and/or the power efficiency of the lighting device, wherein this list is far from complete. These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter. These embodiments will be described by way of example with the help of the accompanying drawings in which:

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 illustrates the chromaticity diagram relating to a lighting device with five primary colors; Figure 2 shows equations relating to a state of the art general feedback control loop;

Figure 3 shows a block diagram of a state of the art general feedback control loop;

Figure 4 shows equations relating to a first control approach according to the present invention which comprises both a feed forward and a feedback control;

Figure 5 shows a block diagram of the first control approach; Figure 6 shows equations relating to a second control approach according to the present invention which comprises the lumping together of several light emitters;

Figures 7 and 8 show two variants of the feed forward control according to the second control approach combined with a feedback control;

Figure 9 shows diagrams comparing the flux and the CRI of a lighting device when being controlled according to different optimization criteria (left/right) and at different temperatures (upper/lower);

Figure 10 shows an example of a look-up-table of duty cycles for color points on the black-body line for a third control approach according to the present invention which comprises situation-dependent optimization criteria;

Figure 11 is a flow chart of the generation of a look-up-table for a CRI optimization; Figure 12 summarizes in principal drawings the three described control approaches according to the present invention.

DETAILED DESCRIPTION OF EMBODIMENTS

Like reference numbers in the Figures refer to identical or similar components.

LED lighting devices based on additive color mixing have high efficiencies, high color rendering indices (CRI), adjustable color temperature, and allow for controlling the color of the light. In order to obtain a large color gamut, known lighting devices are equipped with three primary colors (typically red, green and blue: RGB). With these colors, it is possible to generate warm and cool white light (e.g. from 2500 Kelvin to 6500 Kelvin on the black-body locus). A drawback of such LED light sources with three colors is however that they do not enable sufficient color rendering (i.e. CRI > 80). This can partly be solved by replacing red LEDs by amber LEDs, but this only increases the R_a whereas the Rg is decreased (Rg shows the lamp's ability to render red objects, cf. definition of CRI by the International Commission on Illumination (CIE)), so there is no overall color rendering improvement. Fortunately, this can be obtained by adding an additional color, e.g. an amber LED, thus equipping the lamp with four LED colors (red, amber, green and blue: RGBA), or a white LED, thus equipping the lamp with four LED colors (red, green, blue and white: RGBW), or by using even more additional colors, e.g. an amber and a cyan LED, thus equipping the lamp with five LED colors (red, amber, green, cyan and blue: RGBAC). Moreover, using more than three primary colors increases the color gamut of the lamp. A complication associated with additional colors is however the increased number of degrees of freedom (one for each LED color). The known three-color LED source (RGB) has three degrees of freedom, which are all restricted by the chosen color point (e.g. x, y coordinates) and brightness, whereas the fourth degree of freedom, introduced by adding e.g. an amber LED, is still open for restriction. Furthermore, it has to be observed that the optical characteristics of LEDs vary with manufacturing spread, aging, temperature and forward current. Therefore, the target color point and brightness can hardly be obtained without a suitable feedback system.

The US 6 441 558 Bl describes a system for controlling the color point and brightness of a LED lamp with three primary colors which uses a lumen- feedback and temperature-feed forward control. The system requires switching-off the LEDs briefly for light measurements. Therefore, the LED drivers must have fast response times. In addition, a PWM driving method is required to overcome the LED variations with forward current. With the PWM control, the implementation becomes complex and, in addition, the LEDs are not utilized to their full capacity. In the following, different control approaches according to the present invention are described that are based on the application of at least two different control schemes and that are particularly suited for the control of lighting devices with more than three primary colors. 1. Example: Combined feed forward and feedback control In the first example of a LED lamp that mixes the light of N > 3 primary colors, feedback control is applied to 3 colors and feed forward control is applied to the residual N-3 colors. For the feedback control, the tristimulus values of the light of the LED lamp are measured and the control signals for the 3 colors are determined from the measured tristimulus values and the target tristimulus values in a feedback controller. For the feed forward control, the control signals for the N-3 colors are set such as to achieve as good as possible good color rendering and high efficiency of the lighting device. This concept will be explained in more detail in the following with reference to Figures 1 to 5.

Figure 1 illustrates in a CIE chromaticity diagram the situation that the light of five LEDs with color points Pi, P₂, P3, P4, and P5 shall be mixed to obtain light with color point P_x. The color points have been drawn such that Pi corresponds to red, P₂ corresponds to amber, P₃ corresponds to green, P₄ corresponds to cyan, and P₅ corresponds to blue. There are several possibilities to subdivide these five primary colors into one group of N-3 = 2 colors (N=5) to which feed forward control shall be applied and one group of 3 colors to which feedback control shall be applied. Applying feedback control to red, amber, and blue and feed forward control to cyan and green will be considered in the following.

As a preparation, the case of a lighting device with three primary colors RGB will be considered first with reference to Figures 2 and 3. The light to be emitted by the LED lamp may be specified by its chromaticity coordinates x and y and its luminous flux Φium- From these quantities, the tristimulus values X, Y, and Z can be calculated according to equations (2.1) of Figure 2. The tristimulus values are grouped into a vector TV ("Tristimulus Values"), wherein it has to be distinguished between the tristimulus values TV₀ of the light perceived by the observer, i.e. a person looking at the LED lamp or an object illuminated by it, and the tristimulus values TV_S determined in the feedback path of the color control system.

Ideally, the color sensor used in the lighting device should sense the tristimulus values directly. However, this will not be achieved in practice. The values R, G, and B sensed actually by the color sensor are grouped into a vector SR ("Sensor Readings").

In a similar way, the control signals for the drivers for the red, green, and blue LEDs will also be grouped into a vector CS^ ("Control Signals"). These may be duty cycles for a pulse width modulation control or current amplitudes for an amplitude modulation control. The mentioned vectors TV, SR, and CS^ are listed in equations (2.2). Figure 3 shows the general setup of a LED color control system for a lamp mixing three primary colors that uses a color sensor. It is indicated where the signals discussed above occur in the system. As to the tristimulus values, the input signal TVset and the error signal TV_err are indicated in addition to the output signals TV₀ and TVs. The transfer functions depicted in the block diagram represent the following parts of the system:

Gc - controller

GD - driver

Goso - optical system, from LEDs to observer Goss - optical system, from LEDs to sensor

Gs - color sensor GicAL - calibration matrix.

As indicated by dashed lines, the parts of the control system can be grouped into modules for which the transfer function can be easily determined in a calibration procedure. The first module corresponds to the transfer function G:c2τ from the control signals CS^ to the tristimulus values TV₀ perceived by the human eye, cf. equation (2.3). The second module corresponds to the transfer function Gic2s from the control signals CS^ to the sensor readings SR, cf. equation (2.4).

The calibration matrix Q_CAL can be determined from the requirement that the tristimulus values TV_S in the feedback path have to be equal to the tristimulus values TVo perceived by the observer, cf. equation (2.5). The block diagram of Figure 3 further indicates that the tristimulus values determined in the feedback path of the system are linked to the sensor readings by the calibration matrix, which results in equation (2.6). If the light of more than three primary colors is mixed, then there are still three tristimulus values grouped into TV and three sensor readings grouped into SR. However, there are now more control signals. For the example considered with 5 primary colors there are 5 control signals, one for each of the primary colors red, amber, green, cyan, and blue. These are grouped into a vector CS^ that has now 5 elements (cf. equation (4.1) of Figure 4.

Additionally are defined the vector CSfF of the control signals for the primary colors that are under feed forward control and the vector CSfh of the control signals for the primary colors that are under feedback control. In general, CSfF has N-3 elements, for the example considered it has 2 elements. CSfh has 3 elements, cf. equation (4.1)

Gic2τ and Gic2s are now matrices that have three rows and N = 5 columns, i.e. one column for each primary color. The transfer functions Gic2τ and Gic2s from the control signals to the tristimulus values and sensor readings, respectively, are determined in a calibration procedure. N measurements are taken, i.e. 5 for the example considered.

For each measurement the control signal for one LED color is set equal to one and the control signal for the other LED colors are set equal to zero. Sensor readings are taken and the tristimulus values of the light observed are determined using a spectrometer. The results obtained for the first calibration measurement of the red LED R are shown in equations (4.2). Similar results are obtained in the further calibration measurements for the other four colors, which can then be used to construct the matrices Gc2τ and Gc2s according to equations (4.3) and (4.5) as well as the associated feed forward and feedback components of these matrices, cf. equations (4.4) and (4.6). The relations used in Figures 2 and 3 to determine the calibration matrix for controlling the light of a lamp mixing three colors have to be adapted according to equations (4.7) to (4.11) to the present case where more than three primary colors are mixed.

Figure 5 shows the simplified block diagram of the resulting LED color control system for a lamp mixing more than three primary colors that uses a color sensor. The transfer function Gτ2c describes how to determine the control signals for the primary colors that are under feed forward control from the tristimulus values of the target color and brightness. Any suitable approach can be used for this feed forward part. The node X of the system has the effect that (i) the theoretically achieved feed forward component Gc2τ,ff -CSfF of the resulting tristimulus vector is subtracted from the target

and

(ii) the measured feed forward component GCAL • Qc₂s,ff • CSfF of the resulting tristimulus vector is subtracted from the tristimulus vector TV_S that is used in the feedback path. The same sensor can be used for measuring both the complete light output SR and the feed forward component if the feed forward component is not measured continuously.

The feedback controller Gc for the control system for more than three primary colors can be designed in the same way as that for a control system for three primary colors, for example in the way known from the US 6 507 159 Bl. This document (which is incorporated into the present text by reference) describes a method for controlling the color of a LED lamp that mixes the light of three primary colors (usually RGB), wherein a color sensor is used to measure the tristimulus values of the mixed light. Said color sensor comprises three light sensors with peak sensitivity in different parts of the visible spectrum (usually also RGB). In summary, the described example discloses a methodology for controlling the target color point and brightness of the mixed light of a LED lamp using N > 3 primary colors. Firstly, brightness values optimizing the balance between high luminous efficiency and good color rendering are determined for all primary colors. Then feed forward control is applied to N-3 colors and feedback control is applied to 3 colors. For the feed forward control, the control signals for the N-3 colors are set such as to achieve as good as possible the predetermined color rendering, efficiency, and tristimulus values. For the feedback control, the 3 tristimulus values of the light of the LED lamp are measured and the control signals for the 3 colors are determined in a controller from the measured tristimulus values and the tristimulus values of the target color. 2. Example: Lumped control for several light emitters According to a second approach for controlling a lighting device with more than three primary colors, the number of degrees of freedom is reduced. A typical RGBA LED system would then have for example three degrees of freedom instead of the usual four. This reduction can be achieved by temporarily merging two (or more) degrees of freedom, consequently allowing a simple algorithm to determine the power ratios at which to drive the LEDs. In the most simple case, the merged colors are driven at identical ratios. This approach has two important advantages: a microcontroller can determine the required power ratios for every target color point online; the algorithm also yields an optimum in lumen output of the lighting device when it is supplied with the available lumens of each color at the present operating temperature.

The following description of this example will focus on the combination of red and amber, which is an advantageous combination with respect to the color rendering. The proposed approach is however also suited for the combination of other colors, like blue and cyan, blue and green, but also for phosphor converted LEDs with a substantial, intended blue leakage.

Several ways are known in the state of the art to convert a color point and lumen level (x, y, L or u', v', L) to the LED colors' power ratios or duty cycles (cf. P. Deurenberg et. al, "Achieving color point stability in RGB multi-chip modules using various color correction methods", Proc. SPIE 5914, 2005; WO 2002/47438;

WO 2002/52901). The main difference between these methods is what optical data (e.g. X, Y, Z tristimulus values, x, y, L color coordinates and lumen output) is stored. However, in general the procedure can be explained as follows (for the tristimulus example).

A target color point and light level with tristimulus values X_T, Y_T and Z_τ can be expressed as in equation (6.1) of Figure 6, where the matrix C describes the CIE set points of the LEDs as a function of the duty cycle D₁ for LED color i (with i = R, G, B). The C-matrix contains the CIE 1931 tristimulus values for each LED color (X₁ , Y₁ , Z₁) on a column basis. The inverse of the C-matrix, also called "Calibration matrix", can be used to determine the required duty cycles for a certain target color point. For any three color (LED) system, this can be applied straightforward.

However, when more than three colors are applied, matrix C is no longer invertible because of its dimensions (3x4 in the considered example). Merging two LED color channels (during the computation only) solves this problem. The resulting duty cycles for these two channels may for example be chosen to be equal. In this case, red and amber are preferably merged onto a single duty cycle, because these LED colors are closest together in the color space. The resulting equations (6.2) of Figure 6 show that the C-matrix contains the (lumped) CIE 1931 tristimulus values for each LED color (X₁ , Y₁ , Z₁) on a column basis, wherein the index RA corresponds to a "virtual light emitter" comprising the lumped colors red and amber. The duty cycles for a given target color point can be found during a subsequent color control by multiplying said target color point with the inverse of the C-matrix.

The described approach results in a maximum lumen output of the unit at the chosen color point for the temperature the above C-matrix is valid at. Updating the matrix for other temperatures, results then in maximum lumen output at all temperatures. The matrix updating can be accomplished by multiple calibrations at different temperatures or, if some parameters of the LED's output spectrum are available, by a series of calculations to compensate for temperature drifts.

Alternatively, one can also store the x, y, L values of each LED color. Equation (6.3) can then be used to describe the LED's optical output for all duty cycles D₁ at 100% and at reference temperature T_ref, wherein the index RA indicates the coordinates of either a red, amber or lumped red and amber primary color and wherein L is the maximum lumen output of the lighting device at T_ref. The lumen fractions of each of the (combined) color primaries can be determined by using the inverse matrix according to equation (6.4) in which the index T is used to indicate the target color coordinates and lumen output. To find the duty cycles for each of these primaries, each lumen fraction must be divided by their respective maximal lumen output at D₁ = 100% and calibration temperature according to equation (6.5).

If any of the calculated duty cycles D₁ is larger than one, the desired lumen output cannot be reached. By normalizing the duty cycles with respect to the largest duty cycle, at least the desired color point can be reached, though the flux output is reduced. In case any of the duty cycles D₁ is negative, the desired color coordinates cannot be reached with the used color primaries.

If red and amber LEDs are lumped together, these will get the identical duty cycles, thus optimizing the duty cycles for maximum flux output. The X_RA and V_RA coordinates can be calculated from the tristimulus values as denoted in equation (6.6). Apart from the maximum lumen output, another clear advantage of the described approach is that every color point can be uniquely translated to duty cycles, for any number of LED colors. Obviously, more than two LED colors can also be lumped together.

Figure 7 shows schematically a complete controller arrangement for a lighting unit 10 with four LEDs 11 of the primary colors RAGB. The temperature T of the heat sink of the device is measured with a temperature sensor 13 and communicated to a feed forward controller 20. The feed forward controller 20 determines four control signals according to the described approach (Figure 6), i.e. by combining two colors (R and A) to one virtual light emitter (RA) and by multiplying the target color point (Xχ,Yτ,Zχ) with the inverse of the associated calibration matrix C(T) that is valid for the measured temperature T. Moreover, the color point (R,G,B) of the lighting unit 10 is measured by a color sensor 12 and communicated to a feedback controller 30, e.g. a PID controller. The feedback controller 30 then determines correction factors for the feed forward control signals (D_R, D_A, D_G, D_B)ff. The LEDs 11 are finally driven with the resulting corrected control signals. Figure 8 shows a variant of the system of Figure 7, wherein the lighting unit 10 comprises a flux sensor 12 and wherein the LEDs 11 are driven by pulse-width modulation (PWM). PWM results in the LEDs having a constant forward current, but varying duty cycle or on-time, which essentially means they are switched on periodically and for a certain amount of time only. This drive method has the advantage that the forward current no longer changes the emitted wavelength (as the current remains constant over time). And this means that the mixed color of the unit has less dependencies. In order to determine the output (power) of each LED color with the single sensor 12, the LEDs must be turned on sequentially in time to be able to discern each separate LED color (or alternatively frequency separation might be used). This means that the sensor 12 measures the instantaneous output. Through a number of measurements and some simple calculations, the output of each LED color can then be determined from the measured (single LED) Φ by a color signal extractor 31. However, measuring the instantaneous output and using PWM driven LEDs means that the measured signal does not change when the duty cycle is changed. As the PID setpoint represents the desired output of each LED at the (constant) forward current applied during a duty cycle, the setpoint does not depend on the chosen color point.

To improve accuracy of the system, the temperature T of the heat sink is measured by the temperature sensor 13 and communicated to the PID controller 30. Based on the measured temperature T, the PID controller 30 can then change (in a feed forward way) its setpoint to match the peak wavelength shift of the LEDs due to changes in their temperature. Further, the system functions as described in Figure 7.

In general, the approach to lump two (or more) LEDs is primarily used to determine the initial driving commands that can be used to obtain a desired color. However, these initial commands are only valid at a certain temperature T (the temperature at which the calibration matrix C is determined). Therefore, if the LEDs are also operated at different temperatures (or when they have aged considerably), these feed forward commands alone do not yield the desired color accurately enough, and the feedback controller 30 is needed to achieve the needed accuracy. The feedback algorithm can have a lot of different forms and may for example use the shown color sensor 12 or a simple, unfiltered sensor (cf. US 6 441 558 Bl). In all cases, the feedback algorithm can be constructed such that it works independently of the chosen color point (and thus of the initial driving commands). In such a situation, the feedback algorithm corrects the initial driving commands so using them also obtains the desired color point. 3. Example: Changing optimization criteria According to a third approach for controlling a lighting device with more than three primary colors, color rendering properties are optimized for color points on the black-body line (BBL) and lumen output is optimized for color points off the black- body line. Additionally, it is proposed to perform the complex CRI optimization procedure offline and to supply a microprocessor with a look-up-table of the results. Interpolation may be used to determine the driving ratios in between color points present in the look-up-table. The driving ratios for maximum lumen output can be calculated online by a simple algorithm.

As was already mentioned, it is very difficult to obtain sufficient color rendering for general lighting by using just three LED colors. Therefore, at least one further color (e.g. amber) can be added, which implies however that the driving ratios for the LED colors can no longer be uniquely calculated by just specifying the desired color point and lumen output.

The additional degree of freedom in a (e.g. RGBA) lighting device can be used to optimize light technical properties like CRI, lumen output, or power efficiency (lumen per used Watt of electrical power). The graphs of Figure 9 show in this respect the significant difference in lumen output (left side) and color rendering index (right side) for driving ratios optimized on either maximum flux output (black lines) or maximum color rendering (gray lines). The results change when the heat sink temperature rises (top diagrams corresponding to room temperature 25°C, bottom diagrams to an elevated temperature of about 50⁰C).

To improve the aforementioned situation, it is proposed to optimize the duty cycles for a four (or more) color LED system depending on the chosen target color point. If the user desires a white color (on the black-body line BBL), where the color rendering is important, the power at which the lamp drives its LED colors should be optimized on the color rendering properties of the lamp output (CRI, Rg etc). In areas clearly off the BBL, where color rendering is not very important, the power applied to each LED color or the total lumen output of the lamp can be optimized. For color points very close to the BBL (e.g. with ΔuV < 0.005), the driving ratios can be determined by interpolating between both algorithms.

Unfortunately, for a common microprocessor, it is impossible to perform color-rendering calculations online, because these computations are too complex. A solution to this dilemma is optimizing the driving ratios for color points on the BBL offline. A PC is easily capable of performing these complex calculations for a number of points. The results can be stored in the memory of the microprocessor in the form of a look-up-table (LUT). The driving ratios for target color points in between the stored color points can then be interpolated. An example of such a LUT is shown in Figure 10.

Figure 11 presents an exemplary flow diagram of how to determine the content of such a LUT. In block 1, the required range of Correlated Color Temperatures (CCT) over the BBL is specified, while block 2 specifies the allowable visible difference between the discrete steps on the BBL. Block 3 specifies the available amount of memory for storing the LUT. In block 4, discrete CCT values are determined based on the inputs of blocks 1 and 2, while in block 5 discrete CCT values are removed from this list to fit the table into the available memory. Between the remaining points, an interpolation has to be done. Finally, the power levels D_1;Opt are generated in block 6 for each LED color for the remaining CCT values while optimizing on the chosen criterion (e.g. maximal CRI).

It should be noted that the parameters of the offline optimization algorithm can be tuned specifically to the application or to the lamp in question. For instance, it may not make much sense to increase the color rendering properties from excellent to perfect when the lumen output decreases significantly. A better solution may therefore be found by optimizing on a weighted combination of both color rendering properties and lumen output. Additional optimization parameters (e.g. lumen/Watt) may also be introduced. Depending on the number of available LED colors, even more restraints may be required.

Assuming the microprocessor is already equipped with some kind of color feedback algorithm to make sure the output color of the lamp stays constant (although the system ages and changes in temperature), the system is typically aware of the possible lumen output of each LED color at the current temperature and/or age.

Utilizing this information to determine the driving ratios for each color directly (e.g. with an algorithm that follows the equations of Figure 6) yields the driving ratios at which maximum lumen output of the unit is reached. Thus the best driving ratios for each color point clearly off the BBL can be found. Target color points a little off, but close to, the

BBL can also be found by interpolating between the closest BBL point and the lumen optimized calculation.

The applied algorithm finds the driving ratios for a certain color at reference temperature while also achieving maximum lumen output. By continuously substituting the available lumen output for each (lumped) LED color at the present junction temperature, the algorithm can be used to determine the driving ratios for maximum lumen output.

The technologies described above can particularly be applied to LED fixtures using more than three LED colors. They are especially suitable for general lighting or LCD backlighting, but may also be applied in other application areas where lumen output and color rendering is very important.

Figure 12 summarizes in principal drawings a) to c) the three control examples described above. The drawings show lighting devices 100, 200, and 300 with

N = 5 light emitters 11 (e.g. LEDs) of different primary colors R, A, G, B, C that shall be controlled in such a way that a number k of target values (e.g. color point and/or brightness) are optimally matched, wherein k is typically smaller than N. The control problem is solved by applying at least two different control schemes simultaneously to different light emitters or during different operating conditions to all light emitters.

In the lighting device 100 of drawing a), one group Gi of light emitters is controlled by a feed forward controller Cff, while the group G2 of residual light emitters is controlled by a feedback controller Ca that receives measuring inputs from a color sensor 12.

In the lighting device 200 of drawing b), the driving commands issued by a controller CO to light emitters of the groups Gi, G₂ are coupled, thus lumping these light emitters to a virtual light emitter for the purpose of control. The only light emitter of group G3 is controlled individually. As was described with reference to Figure 7, the controller CO is preferably combined with a feedback controller for adjusting the control signals in view of temperature drifts, aging etc.

In the lighting device 300 of drawing c), different optimization criteria are pursued by control schemes Cl and C2, respectively, under different operating conditions OpI and Op2. This may for example comprise the optimization of color rendering index on the black-body line and the optimization of lumen output elsewhere.

Finally it is pointed out that in the present application the term

"comprising" does not exclude other elements or steps, that "a" or "an" does not exclude a plurality, and that a single processor or other unit may fulfill the functions of several means. The invention resides in each and every novel characteristic feature and each and every combination of characteristic features. Moreover, reference signs in the claims shall not be construed as limiting their scope.

CLAIMS:

1. A lighting device (100, 200, 300), comprising a) a number N of light emitters (11) with different primary colors; b) a controller (20, 30, C_ff, Ca, CO, Cl , C2) for selectively driving the light emitters such that k target values are optimally matched by the common light output of the light emitters, wherein at least two different control schemes are applied by the controller.

2. The lighting device (100, 200, 300) according to claim 1, characterized in that the number k of target values is smaller than the number N of light emitters (11).

3. The lighting device (100, 200) according to claim 1, characterized in that there are at least two groups (Gi, G2, G3) of light emitters (11) which are controlled according to different control schemes.

4. The lighting device (100) according to claim 3, characterized in that one group (Gi) of light emitters is controlled with a feed forward control scheme by a feed forward controller (Cff) and another group (G2) of light emitters is controlled with a feedback control scheme by a feedback controller (Ca).

5. The lighting device (100) according to claim 4, characterized in that there are k light emitters in the group (G2) of feedback controlled light emitters.

Claims

6. The lighting device (100) according to claim 4, characterized in that it comprises a sensor (12) coupled to the feedback controller (Ca) for determining the brightness and/or the color point of the common light output of the light emitters.

7. The lighting device (100) according to claim 4, characterized in that the feed forward controller (Cs) is designed to optimize brightness, color point, color rendering properties and/or power efficiency.

8. The lighting device (100) according to claim 4, characterized in that the target signal and/or the feedback signal provided to the feedback controller (Ca) is freed from components that are associated with the feed forward controller (Cs).

9. The lighting device (200) according to claim 3, characterized in that there is at least one group (Gi, G2) of light emitters (11) for which the driving commands are at least temporarily coupled via a static relation.

10. The lighting device (200) according to claim 9, characterized in that the driving commands of all light emitters of said group (Gi, G2) have a fixed ratio with respect to each other.

11. The lighting device (200) according to claim 9, characterized in that all light emitters of said group (Gi, G2) are treated like one virtual light emitter by a higher level controller (20, CO).

12. The lighting device (200) according to claim 11, characterized in that the higher level controller (20, CO) is designed to optimize brightness and/or color point.

13. The lighting device (200) according to claim 11, characterized in that the higher level controller (20, CO) provides a feed forward control of all light emitters (11) based on at least some of the target values and optionally on operating parameters of the lighting device, particularly on its temperature (T).

14. The lighting device (200) according to claim 13, characterized in that the higher level controller (20, CO) is complemented by a feedback controller (30) acting on at least some of the light emitters (11).

15. The lighting device (300) according to claim 1, characterized in that there are a first and a second region of operating conditions (OpI, Op2) of the lighting device in which different control schemes (Cl, C2) are applied.

16. The lighting device (300) according to claim 15, characterized in that different optimization criteria for the common light output of the light emitters (11) are pursued in said first and second region of operating conditions, respectively.

17. The lighting device (300) according to claim 16, characterized in that the optimization of color rendering properties is given priority in the first region of operating conditions.

18. The lighting device (300) according to claim 16, characterized in that the optimization of the lumen output is given priority in the second region of operating conditions.

19. The lighting device (300) according to claim 15, characterized in that the control scheme changes continuously in an intermediate region between the first and the second region of operating conditions.

20. The lighting device (300) according to claim 15, characterized in that the first region of operating conditions comprises the black-body line of the common light output of the light emitters.

21. The lighting device (300) according to claim 15, characterized in that the second region of operating conditions comprises all color points of the common light output of the light emitters that have more than a predetermined distance from the black-body line.

22. The lighting device (300) according to claim 15, characterized in that at least some of the light emitters are controlled with a feedback control scheme by a feedback controller.

23. The lighting device (100, 200, 300) according to claim 1, characterized in that the controller comprises a memory with a look-up- table containing control parameters.

24. The lighting device (100, 200, 300) according to claim 1, characterized in that the light emitters comprise a LED, phosphor converted LED, organic LED (OLED), LASER, phosphor converted LASER, colored fluorescent lamp, filtered (colored) halogen lamp, filtered (colored) high intensity discharge (HID) lamp, and/or filtered (colored) UHP lamp.

25. The lighting device (100, 200, 300) according to claim 1, characterized in that the target values comprise the color point, the brightness, a color rendering property, and/or the power efficiency of the light emitters.