Motionless system to measure relative angular emission intensity of decaying or modulated light emitting diodes

Nowadays, light emitting diodes (LED) represent a group of optoelectronic devices used in many fields of applications. The device performance of LEDs has to be characterized on the one hand by electrical- and on the other hand by optical quantities. In that sense, current-voltage characteristics, emission spectrum, luminous intensity, and angular dependent emission represent key quantities. Typically, the angular emission is determined by goniometric techniques,¹ providing the angular dependent luminous intensity I(ϕ, θ) of the light source by measuring the illuminance which is received by a detector. Assuming a LED as a point source in distance r from the detector, the relation between measured illuminance (E_V) and luminous intensity (I_V) can be described as

In general, the luminous intensity depends on both polar and azimuthal angles (ϕ, θ). However, it is an industrial standard² to report a relative angular emission I_R by using intensity I_{V(0, 0)} in a specified reference direction corresponding to θ = θ₀ and ϕ = ϕ₀ as a normalizing factor. Moreover, as many light sources exhibit azimuthal rotational symmetry, it is practiced to characterize LEDs in one plane resulting in dependence on only one angle α. In consequence, I_R(α) can be defined by Eq. (2)

A graphical overview of the measurement geometry is shown in Fig. 1, which assumes an ideal Lambert emitter. In that case, the maximum luminous intensity is emitted in forward direction of the device, and is proportional to the signal captured by detector A, whereas detector B provides the luminous intensity at a certain angle α. As indicated by the graph, the emission pattern can be measured by radial scanning using a goniophotometer equipped with a single detector, resulting in a high angular resolution limited only by the angular resolution of the goniometer and desired measurement duration.

Nevertheless, this method is quite time intensive and requires high temporal stability of the light emitting device under test, which is sometimes a limiting factor in the research and development of new materials for light sources (e.g., OLEDS). In order to overcome this limitation, a fixed reference detector (indicated in Fig. 1 as Detector A) can be used to provide constant access to the normalization quantity E_{V(0, 0)}. Such a setup enables measurement of decaying and pulse driven or modulated devices.

Nowadays, light sensors with active amplification circuit are commercially available on the market. In particular, ambient light sensors which are used for automotive applications, display backlighting, and mobile communication are too compact, accurate, and inexpensive devices. One of these sensors is represented by the SFH 5711 (OSRAM). The device is characterized by its compactness (2.35×2.95 mm²), a high accuracy over a wide illumination range, and a good match to human eye sensitivity. Besides these properties, the logarithmic output current represents an exceedingly interesting property being the key to the functionality presented in this work.

The output current response I(E_V) of the detector as a function of illuminance E_V is defined by Eq. (3)³ 

By applying the SFH 5711 as a sensor in a setup sketched in Fig. 1 the relative angular emission I_R(α) can be described by inserting Eq. (3) into Eq. (2), which results in Eq. (4)

As indicated by Eq. (4) the relative emission intensity can be expressed by subtracting the current signals from detectors which can be realized by basic analog electronic circuitry. This paper presents a measurement setup exploiting this concept by usage of 18 fixed detectors and appropriate mixed signal electronic system to enable fast and accurate determination of the relative angular emission without moving parts.

As indicated in Fig. 2, the output currents of all sensors are converted to voltage by mean of a shunt resistors R_L. As the photo detector delivers a maximum current of 50 μA at an illuminance of 10⁵ lx,³ a shunt resistors of 100 kΩ are used, resulting in maximum output voltage described by Eq. (5), where n denotes an index of a sensor

The output of the reference sensor passes through unity gain buffer, and as U_REF connected to the non-inverting input of differential amplifier, and directly to one of the microcontroller's (MC) (PIC 18F2550, Microchip) 10-bit ADC channels. The outputs of the remaining sensors are connected to an analogue multiplexer (MUX, from Analog Devices ADG 732), and similarly its output is passed through unity gain amplifier. The MUX output (U_MUX(n)) is connected to the inverting input of the differential amplifier and to DAC channel. As a result, the output of the differential amplifier U_Diff(x) can be defined by Eq. (6)

The relative illuminance I_R(x) for the detector x can finally be described by Eq. (7)

The selection of a sensor index is done by means of a 4 bit bus driven by the microcontroller, which enables appropriate routing in the MUX. By using the built in Universal Serial Bus (USB) interface of the PIC18F2550 the circuit is connected to a computer providing the user interface.

In order to reliably measure the radial emission of light emitting devices, detectors were located along a semicircle around the emitting sample. This configuration poses an optimization problem for choosing a radius of the semicircle. Shortening the radius increases sensitivity of the individual sensors as their illuminance is inversely proportional to the second power of distance (see Eq. (1)). At the same time it also reduces the angular resolution of the system by limiting the number of sensors positioned at circumference, and introduces errors emerging from spatial extent of a measured device. Detailed analysis of this trade-off, described in Ref. 4, resulted in choosing a 35 mm radius. With such a choice, up to 18 detectors can be located on the detector ring providing a nominal angular resolution of 10°. As indicated in Fig. 1, the typical emission pattern of light emitting devices is symmetric with a mirror axis along the surface normal; emitting the maximum luminous intensity in forward direction. Consequently, the reference detector is located at 0°, as indicated in Fig. 3. The other detectors are mounted in 10° steps along a half circle. The resulting detector positions were chosen in such a way that a symmetric emitting device can be measured with a resolution of 5° (−90°, −80°,..., 0°, 15°, 25°,..., 85°).

A spectral match between measured device and detector is not under lesser consideration. According to CIE technical report on measurement of LEDs,² a spectral mismatch f'₁ should not exceed 3%. In order to fulfill this requirement, while still being able to measure a wide variety of light emitting devices, individual sensors are mounted as exchangeable modules (see Figures 2 and 3). Modules can be equipped with sensors of different spectral responsivity, so exchange of a modules set can lead to obtain desired spectral match. Moreover, the detector ring module holder with mounted detectors forms an enclosure which prevents a stray light from reaching sensors.

The MC software module is designed in order to operate the detector ring in 2 different modes: voltmeter/oscilloscope and ring mode. In voltmeter/oscilloscope mode, user can select between modes where the MUX can be switched manually providing a detailed information about one, arbitrarily chosen, detector; this mode is mainly developed for analyzing time dependent changes of the devices due to degradation. In the latter, the system periodically scans the MUX settings in order to probe the complete angular emission. In both cases, the MC stores up to 90 values in its internal memory buffer before the data package is sent to the PC via USB. The basic communication and control via USB is provided by a dynamic link library (DLL) which includes a Matlab Executable (MEX) interface. The acquired data are transferred using this module to Matlab which provides all necessary functions to visualize the measured data.

A Matlab based graphical user interface (GUI) allows the user to select the device mode and parameters (e.g., sampling rate, detector configuration,...), and visualize the emission pattern using a xy and polar plots.

In order to demonstrate the performance of the presented detection system, a benchmark white LED was mounted in the centre of the detector ring. The LED voltage supply and the three output channels of the detector ring were connected to an oscilloscope in order to visualize all values simultaneously as indicated in Figs. 4(a) and 4(b). For a first test, the LED was driven with a constant voltage of 3.48 V. Consequently, the voltage at the reference output is expected to be constant as demonstrated in Fig. 4(a). By switching the MUX with a sampling rate of 180 Hz the MUX output switches every 5.56 ms to the next detector in a defined sequence (0°, 10°, 20°,..., 80°, −15°, −25°,..., −85°, 0°,...). As a result, all detectors are covered by the measuring cycle within λ_Ring = 93 ms as indicated in Figs. 4(a) and 4(b). The voltage probed at the MUX output shows a step like shape decreasing for the detectors located at larger angles caused by the decreasing luminous intensity. The difference output shows a mirrored shape with its minimum located at 0 V. By applying Eq. (7) to the resulting data the angular dependent relative luminous intensity can be calculated. The result is visualized in Fig. 4(c) in a polar plot by open circles. The actual angular distribution of the benchmark LED is plotted for comparison by a solid line. Striking seems that the measured data show a very good agreement with the data reported by the manufacturer. Furthermore, the absolute luminous intensity can be measured which was performed by applying a constant current of 20 mA to the LED. The measured voltage at the reference detector results in 3.07 V which corresponds to a luminous intensity of 1448 mcd in forward direction. As reported in the datasheet a typical luminous intensity at the used driving current is 1300 mcd which is in good agreement with the measured value.⁵ 

In a second experiment, the driving voltage for the LED was modulated by using a function generator. As indicated in Fig. 4(b) a sine shaped signal with a frequency of 5.7 Hz was applied to the probed LED resulting in a reference voltage oscillating with the same frequency. As all detectors are affected by the variation of the light intensity the MUX output is also modulated as indicated in Fig. 4(b). On the contrary, the difference output is not affected and provides the same signal than in the previous situation as demonstrated in Fig. 4(a) and directly includes the relative angular emission intensity without needing additional information.

Using a commercially available light sensor a measurement system was developed which enables fast and accurate acquisition of the angular dependent emission intensity of light emitting devices. Alignment of the detectors on a semicircle around the emitter and application of a multiplexer allows sample characterisation with a resolution of 5° in less than 100 ms. By taking advantage of the logarithmic dependent output signal a voltage difference between individual detectors and the reference detector the presented system provides a direct access to the relative emission intensity. As demonstrated by theory and experiment, the measurement system can be used for characterization of non-stable and modulated emitters, while exchangeable detector modules allow for a good spectral match.

The authors would like to thank H. Peierlberger providing technical drawings for the ring shaped detector holder and J. Jägermüller for the practical realisation of the constructed part. Furthermore, we would like to thank D. Gruber from the Research Institute for Integrated Circuits (Johannes Kepler University) for fruitful discussions and providing the infrastructure to solder and analyse the electronics.

This work has been financially supported by the Austrian Science Fund (FWF): P25154.