The over-temperature drift specifications of the RES21A-Q1 assume a steady environmental relative humidity, with characterization testing performed in a low-humidity environment. The qualification testing of the RES21A-Q1 shows the device remains within the specified limits after biased testing at high humidity levels (85% relative humidity, 85°C ambient, 1000hrs). While the device mold compound protects the SiCr resistors from corrosion due to high humidity, moisture egress into the mold compound potentially causes swelling of the mold compound, which manifests as a mechanical stress on the die. This additional stress has the potential to cause minor parametric shifts, depending on the severity and duration of the exposure conditions.

Additional extended validation testing was performed on a subset of units to characterize how the device matching specifications drift when temperature is held constant, but humidity is swept. Humidity was first purged from the oven and the units baked at 70°C, before settling to 25°C and recording a 0% RH measurement. The humidity was then swept from low to high, with three minutes of soak time at each level before the measurement is recorded and the humidity levels increased. Plots presented are normalized to the respective values at 50% relative humidity, to better illustrate the shift behavior.

The results show that the ratio tolerance (t_Dx) and divider matching (t_M) are quite robust to humidity from 20% to 60%. From 0% to 20% or from 60% to 70%, higher ratios such as the RES21A90-Q1 showed a more significant shift as the humidity increased, with lower ratios such as the RES21A10-Q1 showing minimal shifts. From 70% to 90% relative humidity, more significant shifts are observed, with the higher-ratio devices again showing the strongest correlated shifts.

Absolute or end-to-end measurements such as t_abs showed minimal shifts with humidity. Additional experiments suggest that increasing the soak time beyond three minutes improves the settling of the device, leading to reduced parametric mismatch. Note that measurement error sources such as oven condensation at high humidity can impact board cleanliness and exaggerate parametric shifts, especially for higher-ratio devices that can require multimeter range-switching for precise measurements. The humidity results suggest that if using the RES21A-Q1 with a high-humidity mission profile, or in situations where environmental humidity is not well controlled and prone to rapid fluctuation, consider incorporating additional design margin to account for potential shifts with humidity.