AKKN ElectronicsField Symptom: Random Overvoltage Faults at Full Load
A 45 kW inverter in a conveyor drive began tripping on DC-bus overvoltage after roughly 18 months of operation. The fault appeared only at full load, and only when the motor was regenerating during deceleration. The control board read a bus voltage 12% above the nominal threshold for 200 ms before shutdown. The original capacitor bank — four 4700 µF, 450 V electrolytics in parallel — showed no visual bulge, no vent leakage, and no measurable capacitance drop when checked with a handheld LCR meter. Yet the inverter continued to trip.
This pattern is common. The capacitors often look healthy, and capacitance checks pass, because the real failure mode is a rise in equivalent series resistance (ESR) that degrades the capacitor's ability to absorb ripple current and stabilize bus voltage during transient load steps. The overvoltage trip is the system-level consequence of a component-level change that a simple capacitance reading will not reveal.
Root-Cause Check 1: Separate ESR from Capacitance
Start by disconnecting the bank and measuring each capacitor individually. A capacitance reading within 10% of the rated value rules out a dry-out or open-circuit failure, but it says almost nothing about the ESR condition. For a 4700 µF, 450 V electrolytic at 20 °C and 100 kHz, a fresh part typically shows ESR between 12 mΩ and 30 mΩ. When ESR exceeds roughly 40 mΩ, the capacitor's internal losses become a significant thermal load and its ripple absorption capability drops disproportionately.
Measure ESR at the capacitor's actual operating frequency if possible, not just the standard 100 kHz. In a DC-link position, the dominant ripple is the PWM switching frequency (typically 4 kHz to 16 kHz) plus its sidebands. ESR at 10 kHz is usually 1.3 to 1.8 times the 100 kHz value, so a part that looks marginal at 100 kHz may be clearly failed at the real switching frequency.
| Parameter | Healthy Range | Degraded Range | Action Threshold |
|---|---|---|---|
| Capacitance (at 120 Hz) | ±10% of rated | −15% to −20% | Replace if below −20% |
| ESR (at 100 kHz, 20 °C) | 12–30 mΩ | 35–50 mΩ | Replace if above 40 mΩ |
| ESR (at switching frequency) | 18–50 mΩ | 55–90 mΩ | Replace if above 80 mΩ |
| Leakage current (after 5 min at rated V) | < 1.5 mA | > 3 mA | Replace |
Use a four-wire Kelvin measurement to avoid lead resistance masking the ESR value. A clamp-on AC current probe on the bus bar, combined with an oscilloscope, gives a direct reading of ripple current. Compare this against the capacitor's rated ripple current (usually specified at 105 °C and 100 kHz). If the measured ripple exceeds the rating by more than 20%, the capacitor is being overstressed regardless of its measured ESR.
Root-Cause Check 2: Verify the Thermal Path and Ripple Budget
ESR rise and ripple current interact in a feedback loop. Higher ESR generates more heat for the same ripple; higher temperature accelerates electrolyte evaporation and further raises ESR. Before ordering replacements, measure the capacitor case temperature with a thermocouple or infrared probe after one hour of continuous operation at rated load. The limit is a core temperature of 105 °C, but case temperature at the bottom can run 10–15 °C cooler than the core. If case temperature exceeds 85 °C under normal ventilation, the capacitor is running dangerously close to its internal limit.
Also check the ripple current distribution across parallel capacitors. Paralleled electrolytics share ripple inversely to their impedance, so a bank with one aging capacitor (higher ESR) forces more ripple through the remaining healthy units. This accelerates their degradation, making the entire bank fail sooner than a single weak cell would suggest. In the conveyor drive case, the original bank had one capacitor with ESR at 52 mΩ; the other three measured between 22 and 28 mΩ. The healthy units were absorbing roughly 30% more ripple than their individual ratings.
Prevention Checklist: Sourcing and Installation Practices
When selecting a replacement bank, focus on parameters that directly affect the failure mode you just diagnosed:
- Ripple current rating: Choose a capacitor with a ripple rating at least 125% of the measured worst-case ripple, not just the calculated average.
- ESR at switching frequency: Prefer parts with published ESR at 10 kHz or 20 kHz, and keep total bank ESR under 10 mΩ for a 45 kW-class drive.
- Voltage margin: A 450 V rated capacitor on a 400 V DC bus provides about 12% margin. If the drive regenerates frequently, consider a 500 V rating to reduce derating stress during transient spikes.
- Thermal placement: Maintain at least 10 mm clearance between capacitor bodies and adjacent heat-generating components. Verify that airflow path is not blocked by bus bars or wiring harnesses.
- Cross-reference verification: When sourcing a drop-in replacement, compare the candidate's ripple current rating, ESR at the relevant frequency, and case dimensions against the original specification sheet, not just the capacitance and voltage.
For the conveyor drive, the fix was a bank of five 3900 µF, 500 V capacitors with a ripple rating 35% above the measured worst-case ripple at the switching frequency. Total bank ESR dropped below 8 mΩ, and the overvoltage trip ceased. The same diagnostic sequence applies to servo drives, UPS systems, and solar inverters where DC-link failures appear as seemingly unrelated control faults.
The key takeaway: when an inverter trips on overvoltage, measure ESR and ripple current before assuming the control board is faulty. Capacitance alone is an insufficient health indicator for DC-link electrolytics under ripple stress.
