When a 15kW frequency converter is started under load, and the operating frequency rises above 30Hz, most of the lighting fixtures, computers, washing machines, and other devices on the same power supply line smoke and burn out. The 380V power supply voltage was measured at this time, rising from the normal value to approximately 500V; the 220V power supply also increased accordingly.

The frequency converter was disassembled and inspected, and a 3.7kW small-power motor was connected for testing, but no abnormalities were found. The student contacted me, also seemingly confused: unable to clearly explain the cause of the fault, only suspecting the DC circuit's energy storage capacitor might be the culprit. The student checked the capacitor's capacitance and found it to be normal, and thoroughly checked all related circuits, but found no problem. Unable to find the issue, a test was conducted on-site, and the power supply voltage was monitored, confirming again that the frequency converter's operation was causing the grid voltage increase.

The frequency converter was disassembled and inspected again, but the problem was still not found. When measuring the internal resistance of an energy storage capacitor using an ESR (Electronic Resistance Ratio) meter, I found that one of the two 3900uF capacitors connected in series had an internal resistance exceeding 8Ω. After replacement, the machine functioned normally.

The capacitor's capacitance was normal, and it handled the load adequately without exhibiting any undervoltage alarms. I've always believed that since energy storage capacitors operate in a low-frequency pulsating DC environment, a small internal resistance wouldn't affect their charging and discharging performance. Furthermore, I've tested multiple internal resistances, all of which were 0Ω; there's no reason for it not to be 0Ω.

Converting the two 3900uF capacitors in series to a single 2000uF capacitor, the integration time constant is approximately 8ms. Based on 1F 1Ω ≈ 1s; 1000uF 1Ω ≈ 1ms, the period of a 300Hz pulsating DC is 3.3ms. In this case, the time constant has reached 8ms, which is indeed significant! This is the crux of the problem. It seems necessary to measure the AC internal resistance of energy storage capacitors as well.