May_EDFA_Digital

edfas.org 5 ELECTRONIC DEV ICE FA I LURE ANALYSIS | VOLUME 24 NO . 2 materials with narrow frequency-windowing harmonic- balance (FWHB) spectrum, KSF phosphors [3] are widely adapted for displays that target DCI-P3 or REC2020 solu- tions. However, the overall external quantum efficiency (EQE) with KSF WLEDs is slightly lower than conventional YAGWLEDs, thus to achieve the same brightness it’s neces- sary to either increase power, or have a special design on the GaN die to increase power efficiency. Expanding the color gamut and improving power efficiency also brings risks. In this article, a novel failure mode encountered in mass production is discussed, which includes the charac- terization and root cause analysis of the failure. FAILURE BACKGROUND AND ISOLATION IN T1 SUPPLIER LCDs are assembled into touchdisplaymodules (TDM), andTDMareassembled into finishedgoods (portable com- puter devices). During the assembly and testing process in T1 supplier, a hotspot issue was reported. After final assembly, products are sampled for ongoing-reliability test (ORT), and hotspot issues also reported in a high- temperature/humidity-bias test (THB). Figure 2 shows failing samples with the hotspot at the bottom (image enhanced to show hotspot clearly). Failures can be detected by visual inspection. The symptom ismost visible at lowbrightness (< 5%) and least visible at high brightness. The A-B-A swap at T1 supplier confirmed that the failure follows the LED bar. To investigate the efficiency drop, IV curve measure- ment was carried out for leakage current characterization. A total of four failing LEDs are measured with a source meter including production failures and reliability test failures, along with three good LEDs as references, as shown in Figs. 3 and 4. Based on IV curve data, high leakage current is present on the failing samples. The leakage at lower voltage is cor- related to EQE drop in the previous measurement. FAILURE NOT DUPLICATED IN T2 BELOW SUPPLIERS However, this failure is not observed at the LCM module tester, and also not shown in the LED sup- plier’s reliability tests reports. Initial investigation found themodule tester andLEDbar tester areusing pulse width modulation (PWM) dimming, but the motherboard of finished goods is using DC (analog) dimming design. The difference can be explained with a simple model below. Considering a simplified model of an LED with leakage as shown in Fig. 5: I Device = I LED + I leakage (Eq 1) If r = , r represents effectiveness of the cur- rent going through the device, and (1— r) represents wasted current going through the leakage path. Fig. 3 IV curve comparison between four failing samples and good reference samples. Fig. 4 Zoom-in of low voltage region of Fig. 3. Fig. 5 Simplified model of failing LED with high leakage current. I LED I Device