Improving HALT Testing and Quantification with FDS Analysis 2023-01-1075

Highly accelerated life testing (HALT) identifies failure modes and operating limits in a short period. Generally, test engineers control a HALT test using overall energy (GRMS) as the sole metric. GRMS is a broad-spectrum intensity measurement and does not account for the spectrum’s overall shape, differences in peak acceleration or kurtosis, or variation in vibration across the table. It forces engineers to make decisions with a simplified version of the test item’s response. The HALT method is often preferred in commercial industries because shaker test durations can be time-consuming. However, the distribution of potential damage across the machine's vibration table is not linear, and the table's fundamental frequencies shift during testing. Additionally, the excitation is not truly random, and there can be large valleys or holes in the spectrum, particularly at the lower frequencies. These issues and the ambiguity of the GRMS control prompt engineers to question the validity and repeatability after generating vibration with a HALT machine. They require a method of estimating a HALT table’s relative fatigue-producing levels to more accurately set test durations and determine when to remove a test item from the table or stop the test. This paper presents a novel approach that improves the validation of the vibration portion of HALT using fatigue damage spectrum (FDS) analysis. The primary validation metric is an FDS generated from the combined step stresses. It allows technicians to analyze the vibratory environment of a HALT machine and compare the accrued fatigue levels to a desired level. Applications of the new approach include analyzing variations due to a change in mass, table location, or orientation, comparing the vibration exposure of multiple samples, and monitoring critical components and known or suspected failure locations. This paper provides examples of end-use applications that support the proposed method compared to GRMS control.