by Tektronix Advanced Semiconductor Laboratory

To more thoroughly assess the aging characteristics of power devices, high-temperature operating life (HTOL) tests have increasingly gained attention from power device test engineers. HTOL replicates the operational conditions of power devices by integrating them into actual power circuits. By subjecting devices to stress through continuous hard or soft switching circuits, HTOL delivers aging effects that closely mimic real-world usage, providing a more accurate evaluation of device performance under diverse stress conditions.

For instance, in hard-switch aging, based on the topologies outlined in the JEP182 standard, the device under test can be incorporated into various circuit configurations, including hard switch, soft switch, or resistive load switch setups. This approach enables a more comprehensive simulation of the device’s operational state in real applications, facilitating a more effective assessment of its long-term reliability.

Using figure (a) as an example, the circuit configuration resembles the double pulse test setup. A power resistor is incorporated into the upper tube loop to dissipate the inductor’s energy during the freewheeling phase stage to ensure the current balance during the continuous switching process. As the inductor L’s energy is dissipated through resistor R, significant heat is generated, necessitating a large heat sink for high-power testing. This increases the circuit’s size and restricts its total power capacity, preventing the chip from operating under high voltage and high current conditions. To enhance aging efficiency, intensify stress conditions during testing, and better replicate real-world scenarios, the test circuit can be optimized. By eliminating the resistive load and employing an inductive load, the circuit allows power devices to operate under hard switching conditions while minimizing energy loss as heat, thereby improving efficiency and reducing heat dissipation.

We can adjust the switching sequence of the four power devices to enable Q1/Q4 to operate under continuous hard switching conditions while Q2/Q3 function in freewheeling mode, achieving a continuous hard switching circuit without a resistive load. Using GaNPower’s 650V GaN HEMT device as an example, we conduct HTOL testing with a controlled switching frequency of 100 kHz. During continuous switching, we employ an oscilloscope (Tektronix), a high-voltage power supply (EA 1500V DC power supply), and a clamp test probe (Hunan Lanhai Electric, stabilization time under 100 ns) to monitor the on-resistance trend of the power device during the turn-on phase, providing insights into the device’s aging process.

Using GaNPower’s 650V GaN HEMT device as a case study, we conducted an HTOL test. The test measured the on-resistance variation of the power device during switching using a Tektronix oscilloscope, an EA 1500V high-voltage DC power supply, and a Hunan Lanhai Electric clamp probe.

By optimizing the switching sequence of the power devices, we achieved a continuous hard switching circuit without a resistive load. Test results indicate that over a 61-hour period, the dynamic on-resistance of both GaN devices and CREE’s SiC MOSFET remained largely stable with no significant increase. When the test voltage was raised to 520V and extended to 245 hours, the dynamic on-resistance exhibited a gradual increase but stayed within acceptable limits. Through linear fitting, the device’s continuous operating lifetime under specific conditions can be estimated.

The figure below displays the dynamic on-resistance waveform captured using a Tektronix MSO58B series oscilloscope. We calculate the dynamic on-resistance value at a specific point by dividing the on-voltage by the on-current. After extended testing, the relative drift in dynamic on-resistance becomes evident.

During testing, continuous hard switching of the power device generates switching power losses, leading to device heating. To prevent junction temperature variations from affecting device characteristics, we monitor the case temperature using external infrared measurements and implement a closed-loop temperature control system with fan cooling to maintain stable junction temperatures during extended operation.

In the initial test, we evaluated GaNPower’s TO-247-4 packaged GaN power device alongside CREE’s C3M0040120D (1200V/66A) SiC MOSFET under identical conditions: a case temperature of 80°C, a switching frequency of 100 kHz, an operating current of 15A, and an operating voltage of 400V. Over a 61-hour test period, we compared the dynamic on-resistance trends of both devices, as shown in the figure below (vertical axis in milliohms). The blue curve represents the CREE SiC device’s dynamic on-resistance, approximately 110 milliohms, while the red curve shows the GaN device’s result, around 54 milliohms. A pulse spike at the curve’s onset reflects adjustments to test parameters. Throughout the test, both devices’ dynamic on-resistance remained stable, with no significant increase observed.

During the aging process described above, the DC power supply provided an output voltage of 400V and a DC current below 100mA, resulting in a system DC power consumption of less than 40W per device during testing. This approach significantly reduces power consumption and testing costs compared to traditional HTOL aging currents.

To accelerate aging and observe more pronounced effects, a second test was conducted with the operating voltage as the aging acceleration factor. The test voltage was increased from 400V to 520V, while other conditions remained unchanged. The HTOL operation extended to 245 hours, with a total of 29,100 dynamic on-resistance measurements were recorded at 30-second intervals.

The figure above illustrates the dynamic on-resistance variation curve of the tested power device over approximately 10 days of continuous testing. The vertical axis represents the dynamic on-resistance values, while the horizontal axis indicates the test sampling number. Periodic fluctuations in the test results are observed, attributed to day-night temperature variations in the test environment affecting the circuit. However, the long-term trend reveals a gradual increase in dynamic on-resistance. By comparing these results with Beijing weather data from the same period, as shown in the figure below, the fluctuation patterns align closely, confirming that ambient temperature impacts the test outcomes to some extent.

By applying linear fitting to the test data, we determined the slope of the dynamic on-resistance increase to be approximately 6.93×10⁻⁵. Assuming the device reaches its lifetime limit when the dynamic on-resistance rises by 30%, under test conditions of 520V, 15A, a device case temperature of 80°C, and a 50% duty cycle, the continuous operating life of the device is estimated at 1,724 hours. Given that the device’s typical operating voltage is 400V, its operational lifespan under standard conditions is expected to significantly exceed this duration.

Analysis of the test data indicates that GaNPower’s 650V high-voltage GaN device exhibits aging characteristics comparable to SiC devices. At an elevated operating voltage of 520V, the on-resistance shows a gradual increase but maintains a substantial operational lifespan. Aging tests like HTOL enable a faster understanding of the aging process and performance degradation in new power devices, aiding R&D and design engineers in refining designs and improving product performance efficiently.

Tektronix Innovation Lab V2.0 has undergone a comprehensive upgrade, incorporating advanced equipment and enhanced testing capabilities to address the diverse requirements of third-generation semiconductor power devices. The upgrade includes expanded switching and dynamic on-resistance testing for GaN devices, short-circuit testing and avalanche testing for SiC power devices, alongside a more robust system for static and capacitance parameter testing. Additionally, the lab has introduced a new reliability testing system focusing on evaluating the performance of third-generation semiconductor power devices.

A key addition is the high-temperature operating life (HTOL) test method, which replicates the aging process of devices in real-world conditions, accelerating degradation through elevated temperatures to rapidly acquire aging characteristic data. These results are highly reliable and offer critical insights for determining product warranty periods and maintenance strategies. HTOL testing also enables prediction of device lifespan under specific operating conditions, empowering designers to make informed decisions during product development and optimization.

NOTE: This is a translated version of the original article, which can be accessed in Chinese at the following link: https://mp.weixin.qq.com/s/KsgUFUet3pS3ebmCTwN6nQ