SEL characterizations was performed with ambient temperature of 125°C. Ambient temperature of 125°C was achieved with a convection heat gun aimed at the die. The ambient temperature was monitored during the testing using a K-Type thermocouple attached to the heat slug of the package with solder paste. Thermocouple and die correlation was verified by using a thermal IR camera prior to reaching the heavy-ions facility (TAMU). The device was exposed to a Gold (Au) heavy-ion beam incident on the die surface at different angles for a LET_EFF from 85MeV·cm²/mg up to 121MeV·cm²/mg across multiple runs. Flux of approximately 10⁵ ions/cm² ·s and fluence of 1×10⁷ ions/cm² were used. Run duration to achieve this fluence was approximately two minutes. As mentioned in Section 6, all supplies were set to the maximum allowable value.

For SEL characterization, the PHY was first initialized for communication at the proper speed, and to configure the LEDs for link indication. Link status was monitored throughout the beam run. PHY register values were recorded prior to the run. During the run, link was lost, as expected, and was recovered post run. Link recovery did not require a power cycle, but sometimes required a soft reset through registers, a hard reset through registers, or a hard reset through grounding the reset pin through hardware. Under this condition (no power cycle required), no incidences of SEL were detected on any ion runs, indicating that the DP83561-SP is SEL-immune up to LET_EFF = 121 MeV × cm²/ mg at 125°C. A single high current event was observed during one of the runs while testing 100Mbps mode, though RX_CLK was also observed to produce a clock frequency of 125MHz. The high current value observed matched the expected current for 1G communication, and the RX_CLK frequency of 125MHz confirm that the device had been momentarily reset for 1G mode, before recovering to a 100Mbps mode of operation. The device could be re-programmed back to the 100Mbps mode and the device currents were normal Indicating the current fluctuation is caused by register change.

Table 7-2 summarizes the SEL test conditions, including LET_EFF, MAC interface mode, and speed, as well as the overall results. To reiterate, no incidences of SEL were detected on any ion runs, indicating that the DP83561-SP is SEL-immune up to LET_EFF = 121MeV·cm²/mg at 125°C. The SEL cross section was calculated based on zero events observed using a 95% (2σ) confidence interval. Figure 7-1 shows a typical current plot.

σ_SEL ≤ 7.38 x 10^-8 cm²/device LET_EFF = 121 MeV × cm²/ mg, at 125°C, 95% confidence.

MII is characterized up-to 85 LET due to limited time at Lab. We expect it to meet similar levels as 1G