LED Binning Using a Single Pulse Tester

Wolfgang Daehn and Bob Angelo, Gigahertz-Optik Inc., Newburyport, MA

LEDs are considered to be the light technology of the future due to their low energy consumption and long lifetime compared to traditional light sources. As LED luminous intensity levels increase, the range of applications for LEDs expands as well.

Many LED applications require that optical properties like luminous color and intensity be clearly defined, but in the mass-production of single LEDs, small differences in these properties are inherent in the manufacturing process. As a result, after fabrication all LEDs are classified according to light and color properties. This classification is called “binning” (similar diodes going into the same bin class).

The binning of white LEDs is mainly done according to luminous flux, color (CIE 1931) and color temperature. Color temperature bins are defined by xy coordinates on the CIE 1931 Chromaticity diagram. These bins are grouped as quadrants around the standard chromaticity lines for a specific color temperature. The smaller the quadrants, the tighter the tolerance limits. In single pulse mode binning, this measured data is taken during a few-millisecond-long light pulse at a specified operating current.

This technical article will accomplish three purposes:

• Explain the measurement set-up and use of the BTS25-LED Tester and LPS20 LED Power Supply for short-duration 20ms pulse LED binning.
• Investigate what effects light pulse length have on binning parameters. In addition, binning data of the same LED with and without heat sink are compared.
• Examine the effects of integrating sphere absorption on the measurement results.

A Diamond Dragon white LED, operated at a wide range of operating currents, was used as the test sample, and the following Gigahertz-Optik instrumentation was used to perform the actual measurements:

BTS256-LED Tester

The BTS256-LED Tester is a compact measurement device designed for high-accuracy measurement of luminous flux, spectral, and color data of single assembled and unassembled LEDs in the visible spectrum. The BTS256’s bi-technology light sensor offers a fine photometric response photodiode for accurate wide dynamic range flux detection. Along with the integral detector, a compact low stray light spectrometer is installed for spectral color measurements. The photodiode of the Bi-Tec sensor offers a fast data logger mode with 1ms sampling rate, which can be used for pulse form profiling measurements. To operate the BTS256-LED, a laptop or PC with USB interface is needed. The BTS256-LED is powered through its USB adapter when connected to PC.

LPS-20-1500 LED Power Supply

The LPS-20 is a microprocessor-based current and voltage source especially designed for the operation of LEDs and other semiconductor light sources requiring low noise. The LPS20-1500 operates up to 1500mA with 30μA resolution and up to 24V with 0.5mV resolution. It is set up for full remote control operation in CW or single pulse operation mode. Either current or voltage can be measured with high 16-bit ADC resolution. Its I/O interface supports trigger-in and trigger-out operation in flash mode.

Verification of Thermal Stability

First measurements were taken to help gauge any thermal effects on test parameters during the 20ms pulse. For this investigation the LED tester and LPS-20 were adjusted so that a few milliseconds before the flash, the LED tester started to measure and collect data with its photodiode over the whole pulse width. This measurement was of the Diamond Dragon operated with heat sink at currents of 200mA, 350mA, 600mA, 1000mA and 1400mA. During the pulse, the photodiode and the data logger collected data every 1ms. Figure 3 shows that there are no thermal effects visible at the end of the pulse (degradation of flux).

The differences at the beginning and the end of the pulse are caused by the measurement device electronics. The LED tester operates in a different gain range with different rise times for the two lowest current pulses (200mA and 350mA curves overlap) than that at the higher currents (600mA, 1A and 1.4A curves also overlap).

For comparison, LED operation times were lengthened and results were analyzed. Logically the LED temperature will rise during longer operation times and changes in luminous flux, luminous color, and temperature should develop. While the test device LEDs were operated with a constant current of 350mA, luminous flux and color were measured over a time interval of 60 seconds. This measurement was performed twice, with and without heat sink. Results are shown in Figures 4 and 5. As the LED heats up, luminous flux and color characteristics are affected. The effects are noticeably greater without the heat sink attached. Over sixty seconds, only 0.6% of initial luminous flux is lost for the LED with heat sink.

The color temperature changes by 30K. The LED without heat sink luminous flux drops to 93% of its initial value while color temperature increases 310K. To sum up, thermal effects were shown to cause a 10X difference in flux and color temperature for the test LED with and without a heat sink.

Verification of the Test Sample Bin Classes

According to the Diamond Dragon data sheet (LUW W5APMYNY-4C8E), at an operating current of 1400mA the typical color temperature should be about 6500K and located in one of the brightness groups from MY to NY (meaning luminous flux between 210 and 390 lm). Also it should fall into one of the xy Chromaticity coordinate bin classes from 4C to 8E (Figure 7). According to the data sheet the LEDs are only classified with binning data for brightness and chromaticity group, not for single bins.

For the binning measurements the BTS256-LED tester and LPS-20 LED power supply was set in 20ms single pulse mode and 1400mA operating current so the measured data could be compared with the datasheet specifications. Also, overall measurement accuracy was checked.

The BTS256-LED tester and the LPS20 were adjusted so that 10ms before the 20ms long flash, the spectrometer started measuring and stopped 10ms after the flash. The photodiode collected data every 1ms starting and finishing 3ms before and after the flash.

First the results show that all of the measurement values fell into the same bin groups as stated in the data sheet. Secondly the higher resolution single bin-class can be specified. So the measured LED is part of brightness group MZ (240 lm -280 lm) and it belongs to Chromaticity coordinate group 5D (Figure 9).

The next question is what impact does varying the current have on the binning results. To find an answer the testing of the LED operated at 200mA, 350mA, 600mA and 1000mA was repeated.

Looking at Figures 7 and 8, you can see that not only does luminous flux change with current, but also color temperature and chromaticity coordinates. As shown in Figure 9, at 1400mA the LED is located in 5D, but the same LED operated at 1000mA and 600mA falls into 6D, whereas operated at 350mA and 200mA it is 6E.

Spectrometer Measurement Plausibility Check

Unlike photodiodes, diode array sensor sensitivity is controlled by integration time. The greater the signal level the shorter the integration time. Longer integration times are required for higher sensitivity. In short-duration binning applications, the integration time — thus sensitivity — is set to a fixed value.

So, the measureable flux range is limited by the dynamic range of the diode array. To avoid inaccuracy due to low signal- to-noise ratio or saturation of the pixel signal, the diode array raw signal ADC output in counts should be checked. A graphical measurement display in the software supplied with the LED tester displays the spectral flux in counts. In our test, the raw data at 46 lm (200mA) and 250 lm (1400mA) are 475 counts and 2850 counts, respectively, at the wavelength of peak signal. On the assumption that the signal-tonoise ratio of the ADC signal should be, at minimum, about 100:1, the measurement range of the LED tester with the white LED measured is about 10 lm to 300 lm. For higher flux levels, the 20ms measurement time needs to be shortened. For lower intensities, the measurement time would need to be longer. However, for longer measurement times the temperature effects must be checked as already described.

LED-Binning with Large Diameter Integrating Sphere

The LED tester can be combined with integrating spheres from 210mm to one meter in diameter (Figure 10). The effects of adding an external integrating sphere on the measurement values was examined. For this test, the BTS256-LED tester was mounted to a Gigahertz-Optik ISD-21-V01 integrating sphere using its standard bayonet type adapter. The sphere is 210mm in diameter with a 50mm diameter measurement port that can be increased to 75mm or reduced using optional port reducers. The detector port is baffled to the measurement port to avoid direct illumination of the detector by the test source. The sphere is set up with an auxiliary lamp to measure and compensate for self-absorption of the test sample.

The luminous flux and spectral flux of the test LED was measured at 200mA, 350mA, 600mA, 1000mA and 1400mA operating current. The measured signal in counts related to the luminous flux confirms attenuation of the sphere by a factor of 53. At this level of attenuation, the available signal to the LED tester is less than 100 counts for all measured intensities and, therefore, less than the recommended minimum signal- to-noise ratio of 100:1 counts. Single LEDs with a maximum luminous flux of 500 lm should, therefore, be measured with the BTS256-LED tester without the sphere. Higher power LEDs, as well as large size, high-power LED matrixes or LED luminaires, can be measured using the large size integrating sphere with LED tester. For flux levels lower than 500 lm, longer diode array spectrometer integration times are recommended. Also, for longer pulse length operating times, the effects of LED operation temperature on measurements must be evaluated as previously described.

For more information, Click Here 

---