SDAA249 January 2026 AMC3330-Q1 , INA148-Q1 , TPS61170-Q1 , TPSI2240-Q1
The accuracy of the IMD is related to many factors, such as the precision of resistors, the value of Y capacitors, the deviation of the DC bias supply, the error on the amplifier, and the settling time. The most reliable way to assess the accuracy of the IMD is through hardware testing. The simulation is only used to verify whether insulation failures can be detected under several typical operating conditions.
For the case of an asymmetric insulation failure, as can be seen from Table 4-2, when the insulation resistance drops below 90kΩ, VIN_AVG exceeds the 1.8V threshold, and the insulation failure is detected. When the insulation resistance is greater than 200kΩ, even if VIN_AVG exceeds 1.8V due to various errors, the peak value of VIN is still below the 2.5V threshold, which means a plausibility check prevents false reporting of an insulation failure.
For the case of a symmetric insulation failure, the VIN peak voltage and VIN_AVG can be found in Table 4-3. When the insulation resistance drops below 90kΩ, VIN_AVG exceeds the 1.8V threshold, and the insulation failure is detected. When the insulation resistance is greater than 500kΩ, the VIN_AVG is 60mV lower than the 1.8V threshold, so an insulation failure is not falsely triggered.
| RisoL | RisoN | VIN Peak | VIN_AVG | Plausibility Check Applicable | Insulation Failure |
|---|---|---|---|---|---|
| 10MΩ | 10MΩ | 1.51V | 1.51V | No | No |
| 1MΩ | 1MΩ | 1.69V | 1.69V | No | No |
| 500kΩ | 500kΩ | 1.74V | 1.74V | No | No |
| 200kΩ | 200kΩ | 1.79V | 1.79V | No | No |
| 90kΩ | 90kΩ | 1.81V | 1.81V | No | Yes |
| 0kΩ | 0kΩ | 1.83V | 1.83V | No | Yes |
The trickiest situation occurs when a symmetric insulation failure happens, and the insulation resistances are both around 200kΩ. This is because 1.79V VIN_AVG is very close to the 1.8V threshold, so making an insulation failure falsely trigger can easily occur.
A false trigger at 200kΩ is not so bad, since the insulation resistance is expected to be megaohms and the threshold is 110kΩ, so if the insulation is 200kΩ there is already something wrong in the system. As mentioned in Section 3.5, for this kind of scenario, the proposal is to sample a longer time and take the average to minimize the impact of dynamic errors on sampling accuracy.