EP2057503A1

EP2057503A1 - Led module

Info

Publication number: EP2057503A1
Application number: EP08715543A
Authority: EP
Inventors: Julius Muschaweck; Markus Zeiler; Josef Hüttner; Felix Michel
Original assignee: Osram Opto Semiconductors GmbH
Current assignee: Ams Osram International GmbH
Priority date: 2007-03-09
Filing date: 2008-03-05
Publication date: 2009-05-13
Also published as: DE112008001265A5; KR20090125267A; WO2008110142A1; US20100061081A1; TW200845360A; CN101627332A; CN101627332B; JP2010521066A; US8360601B2; TWI402963B

Abstract

The invention relates to an LED module (10) comprising a support and a plurality of light sources (1), every light source (1) being arranged on the support in such a manner that the LED module (10) has a radiation homogeneity H<SUB>BLU</SUB> that is smaller than the statistic mean H* of a radiation homogeneity distribution, the radiation homogeneity distribution being based on a plurality of LED modules (10) having a random arrangement of the light sources (2). The invention also relates to a method for producing said LED module.

Description

description

LED module

The invention relates to an LED module and a method for producing an LED module.

LED modules are used, for example, for backlighting LCDs. Many individual LEDs are arranged behind the display in one plane. One difficulty lies in this case is to arrive at a sufficiently ^homogeneous' backlighting. These homogeneity problems have two different causes: on the one hand, it would be difficult to build thin and homogeneous backlight units at the same time, if one had completely identical LEDs, since it already requires special efforts in the optical design to achieve this "intrinsic" homogeneity , on the other hand, the LEDs are not identical. One tries to obstruct possibly identical pre-sorted LEDs (so-called "binning").

For reasons of measuring accuracy, but above all for the sake of the otherwise "combinatorial exploding" logistics, it is not possible to set the limits for binning at the LED level so narrowly that the difference of LEDs of the same type makes no more difficulties. Particular problems arise when, due to sequential manufacturing processes or even due to chance, LEDs with similar deviations from the mean are installed close to each other.

It is an object of the present invention to provide an LED module with improved radiation homogeneity. These Task is solved by an LED module according to claim 1.

Another object of the invention is to provide a simple method of manufacturing an LED module with improved color homogeneity. This object is achieved by a method according to claim 11.

Advantageous embodiments and further developments of the LED module and of the method for producing such an LED module are the subject of the dependent claims.

An LED module according to the invention comprises a carrier and a plurality of light sources each having one light emitting diode or a plurality of light emitting diodes, each light emitting diode having a color locus Fi and a brightness Hi and the light sources depending on the color loci Fi and / or the brightnesses Hi in a predetermined position on the support are arranged such that the LED module has a radiation homogeneity H _BLU which is smaller than the statistical average H * one

Radiation homogeneity distribution, which results in a random arrangement of the light sources in the LED module. By means of the running index i, the LEDs are numbered consecutively.

The LED module according to the invention is provided in particular for backlighting or illumination. For example, the LED module can be used in a flat screen television for backlighting or as a radiation source for general lighting. Preferably, the radiation homogeneity H _{BLU of} the LED module is evaluated by a scalar quality function. The quality function is calculated as follows: take a number of test points on a radiation exit surface and determine the color location of each test point in the CIE u'v ¹ space or CIE L * a * b * space. Then calculate the mean color location of the LED module and also calculate the distance from the mean color location for each test point. The quality function is preferably defined as twice the maximum resulting distance.

The radiation exit surface is determined to be a surface located in a plane at a distance D from the support and extending over the entire extent of the LED module.

Preferably, the radiation homogeneity H _{BLU of} the LED module is at most half as large as the statistical mean H * of the radiation homogeneity distribution.

In the present case, an advantageous simulation method is used to calculate the radiation homogeneity H _{BLU of} the LED module. For this purpose, the brightness contribution of an LED is considered as a parameterized function of several variables.

The variables are the location of the LED, the location of the considered point on the exit surface, and possibly other variables that cause variations in the LED emission characteristic (for example, the chip position tolerance). The brightness contribution of an LED is considered, for example, as an additive superposition of several two-dimensional Gaussian functions of different strength and width; Edge effects can be, for example, on the edge mirrored further, possibly elliptic Gaussian functions take into account. A continuous transfer function of the light from the single LED, which is located at a certain, but freely selectable location, to another, freely selectable location on the exit surface is thus determined by the parameters of the Gaussian functions used. The adaptation of these parameters can either be done by numerical optimization of the transmittance model to measurements or ray tracing simulations or "by hand".

The quality function is then defined as the maximum resulting distance. The quality function of the radiation homogeneity H _BLU can be determined from simulation results.

In a first approach, it is possible to use identical red, green and blue LEDs for the simulation and thus to investigate the "intrinsic radiation homogeneity" of the LED module at different LED intervals. In further calculations, one can use the limits of the LED bins read from the data sheet to generate random color and brightness values of the individual LEDs and thus to investigate the influence of production-related brightness and color differences of LEDs.

With such a simulation method, it is for example an LED module with 1000 LEDs and about 100 x 100 test points, so 100 x 100 x 1000 = 10 ⁷

Functional evaluations of the transfer function, it is possible to determine the radiation homogeneity H _BLU at an experimentally determined time of about 0.3 μs per function evaluation on a standard PC in about 3 seconds. Advantageously, the results are without any Statistical noise, but contain modeling errors, the size of which depends on the flexibility of the transfer function model used and the accuracy of the parameter adaptation. The speed advantage lies in putting all the optical behavior of a particular LED module design into the transfer function model one single time, and then using this model to calculate a variety of LED module configurations.

Based on the LEDs installed in a light source of the LED module, one can statistically examine the different radiation homogeneity _values H _{BLU /} resulting from the different possibilities of the light source arrangements in the light module.

According to a preferred variant of the LED module, each light source has a light-emitting diode or a plurality of light-emitting diodes. In the present case, a light-emitting diode is preferably to be understood as meaning a radiation-emitting component having at least one semiconductor chip. However, it may also be meant a single radiation emitting semiconductor chip.

According to a further preferred variant, the light-emitting diodes are mounted on a printed circuit board.

In particular, each light source has at least two light-emitting diodes which emit radiation of different wavelengths. By way of example, each light source can have at least one red, one green and one blue light-emitting diode. Preferably, a plurality of RGGB clusters are used for the light sources, that is, clusters having one red, two green and one red include a blue LED. Alternatively, each light source may include at least one light emitting diode emitting white light.

A method according to the invention for producing an LED module as described above has the following steps:

Provision of a plurality of light sources,

Measurement of the color locus Fi and the brightness Hi of the individual light-emitting diodes of each light source,

Calculation of an optimum position for each light source on the support as a function of the values of the color locus Fi and / or the brightness Hi such that the LED module has a radiation homogeneity H _BLU which is smaller than the statistical average H * of one

Radiation homogeneity distribution, which results in a random arrangement of the light sources in the LED module,

- Positioning of the light sources at the calculated positions on the carrier.

Further features, advantages and developments of the invention will become apparent from the following explained in connection with the figures 1 to 3 embodiments.

Show it:

FIG. 1 shows a schematic plan view of an LED module according to the invention,

FIG. 2 shows a histogram representing a frequency distribution W of the radiation homogeneity H _BLU for a LED module with a random arrangement of the light sources, Figure 3 is a schematic representation of a light source arrangement in a conventional and in an LED module according to the invention.

The LED module 10 shown in FIG. 1 has 1152 LEDs 2. The LEDs 2 result in 48 light sources 1 each having 24 LEDs 2 in juxtaposed RGGB clusters 3. The light sources 1 are arranged on a carrier (not shown) in the manner of a matrix having twelve rows and four columns. The 48 light sources 1 of the same design can basically be installed in any order. At 48 light sources

1 result in 48! ≤10 ⁶¹ possibilities, inconceivably more than you could fully investigate. But you can get an idea of the probability distribution by counting a certain number (for example

2000) of arrangements and evaluated as mentioned above. In this way, one can determine the radiation homogeneity distribution and the statistical average H *

Estimate the radiation homogeneity distribution and its variance. The light sources 1 are then arranged on the carrier such that the radiation homogeneity H _{BLU of} the LED module 10 is smaller than the statistical average H *.

To determine the radiation homogeneity distribution, the color locus Fi and the brightness Hi of the LEDs 2 are first determined. For reasons of production control, the color location Fi and the brightness Hi of the individual LEDs 2 of each ready-equipped light source 1 are measured again anyway, although the LEDs 2 are typically already grouped prior to assembly according to color locus F ± and brightness Hi. The measurement for the purpose of determining the radiation homogeneity distribution therefore does not provide any In addition, usually for error traceability each individual light source 1 is marked, such as with a bar code label, so that each light source 1 is identifiable.

In the present case, the so-called "final clearance ^w " measurement data of the individual light sources 1 can be stored, for example, together with the bar code of each light source 2 in the PCB assembler in a database. While the light sources 1 are packaged together in kits for each LED module 10 and transported for final assembly, the optimization calculation can take place advantageously offline, which will be described in more detail below. Their result can be transmitted to the final assembly electronically.

In the final assembly by hand, the worker scans the bar code (he has to do anyway, because of the traceability) and then gets the space provided for the light source 1 on a screen. The space can also be displayed by marking the corresponding location on the carrier with an automatically controlled lamp. The final assembly can also be carried out automatically, for example by means of a robot. In both cases (manual assembly or automated assembly), the sum of the travel paths is not increased by the optimization method.

The information available at the light source level is used for the simulation of the

Radiation homogeneity distribution used. One can try with the available information, by numerical optimization, a best possible arrangement of the light sources 1 to so that the finished LED module has an optimized radiation homogeneity H _BLU .

The problem of finding the best possible arrangement is related to the classic "traveling salesman" problem of optimization theory (in which order should the traveling salesman visit 100 cities, so that his total travel distance is minimal?). It is known that this problem is NP-complete. The search for the global optimum of the more than 10 ⁶¹ options is thus hopeless. However, a very good solution can be found with the presently chosen method of simulated annealing. This process is inspired by the way in which the arrangement of the crystallites in a piece of steel during annealing finds a minimum of free energy.

Starting from a start arrangement, a random interchange of individual light sources or individual disjoint rectangular light source areas is selected here. The radiation homogeneity H of this changed arrangement is calculated. If there is an improvement, the new arrangement is taken in any case. If a deterioration results by ΔH, this new arrangement is accepted with a probability of exp (-ΔH / T). In this case, T is a parameter called "temperature", which should have such a high value at the beginning of the calculation that even many deteriorations are accepted. In the course of the calculation, the parameter T is then gradually reduced until almost only real improvements are selected.

Advantageously, in the present case, the already existing information about the color locus Fi and the brightness Hi of LEDs used and translated into an easily automatable and almost no additional effort implementable manufacturing instructions. Not only bad LED modules are avoided, but the radiation homogeneity H is systematically and considerably improved for practically all LED modules.

FIG. 2 shows the result of a calculation according to the invention. On the abscissa is the

Radiation homogeneity H in JNDs (Just Noticeable Differences). The ordinate indicates the frequency W of the radiation homogeneity H for a given number of random arrangements, in this case for 2000 random arrangements, in the respective interval.

The histogram distribution in this case has a statistical mean H * of around 4 JNDs and a total width of around 3 JNDs. The best value of the 2000 random arrays is around 2.5 JNDs. The bar at 1.5 JNDs shows the value of the optimum arrangement determined by the optimization procedure from a total of about 2,000 tested assemblies. It is not a contradiction that the found optimum lies far outside the histogram distribution. This only shows that the actual distribution is significantly wider than you can see with only 2000 random experiments.

FIG. 3 shows in comparison on the left a conventional LED module 10 with a radiation homogeneity H = 4.6 JNDs and on the right an inventive LED module 10 with an optimized radiation homogeneity H _BL u = 1.5 JNDs. In the conventional LED module 10, red, green and blue LEDs are pre-sorted in bins for brightness and color location and mounted "bin-pure" on printed circuit boards, with a light source 1 of 24 LEDs will be produced. The light sources 1 are mounted on the carrier in a random arrangement. The inventive LED module 10 differs in the arrangement of the light sources 1 of the conventional LED module 10, as can be seen due to the numbering of the light sources 10. In a random arrangement of the light sources, it is very unlikely that the optimized radiation homogeneity H _{BLU is} achieved (compare Figure 2 H = 4.6 JNDs is more probable than H _BLU = 1-5 JNDs).

It should be expressly noted that the method can be easily transferred to other evaluation functions that also evaluate the gradients in addition to the color locus and / or the brightness.

The invention is not limited by the description with reference to the embodiments. Rather, the invention encompasses any novel feature as well as any combination of features, including in particular any combination of features in the claims, even if this feature or combination itself is not explicitly stated in the claims or exemplary embodiments.

This patent application claims the priority of German Patent Application 102007011988.9, the disclosure of which is hereby incorporated by reference.

Claims

claims

An LED module (10) comprising a support and a plurality of light sources (1) each comprising a light emitting diode (2) or a plurality of light emitting diodes (2), each light emitting diode (2) having a color locus Fi and a brightness Hi and the light sources (1) depending on the color locations F ₁ and / or the brightnesses Hi are arranged in a predetermined position on the carrier such that the LED module (10) has a radiation homogeneity H _BLU which is smaller than that statistical average H * of a radiation homogeneity distribution resulting from a random arrangement of the light sources (1) in the LED module (10).

2. LED module (10) according to claim 1, wherein the radiation homogeneity H _{BLU of} the LED module (10) is at most half as large as the statistical mean H * of the radiation homogeneity distribution.

3. LED module (10) according to any one of the preceding claims, wherein the light-emitting diodes (2) are mounted on a printed circuit board.

4. LED module (10) according to any one of the preceding claims, wherein each light source (1) has at least two light-emitting diodes (2) which emit radiation of different wavelengths.

5. LED module (10) according to claim 4, wherein each light source (1) has at least one red, one green and one blue light emitting diode (2).

6. LED module (10) according to any one of the preceding claims, wherein each light source (1) has at least one light emitting diode (2) which emits white light.

7. LED module (10) according to any one of the preceding claims, wherein each light source (1) has twenty four light emitting diodes (2).

8. LED module (10) according to any one of the preceding claims, wherein the light sources (1) are arranged in the manner of a matrix on the carrier.

9. LED module (10) according to claim 8, wherein the matrix has twelve rows and four columns.

10. LED module (10) according to one of the preceding claims, which is provided for backlighting or lighting.

11. A method for producing an LED module (10) according to any one of claims 1 to 10 with the following steps:

Providing a plurality of light sources (1),

Measurement of the color locus Fi and the brightness Hi of the individual light-emitting diodes (2) of each light source (1),

Calculation of an optimal position for each light source (1) on the support as a function of the values of the color locus Fi and / or the brightness Hi such that the LED module (10) has a radiation homogeneity H _BLU which is smaller than the statistical one Mean H * of a radiation homogeneity distribution, which is at a random arrangement of the light sources (1) in the LED module (10),

- Positioning of the light sources (1) at the calculated positions on the carrier.

12. The method of claim 11, wherein the calculation of the optimal position by means of "simulated annealing" takes place.

13. The method of claim 11 or 12, wherein a plurality of printed circuit boards each having a plurality of light emitting diodes (2) for producing the light sources (1) are fitted.

14. The method according to claim 13, wherein the light-emitting diodes (2) are grouped before the color point Fi and brightness Hi.