AKKN ElectronicsA 100 kW UPS with a 540 V DC-link bus uses a bank of aluminum electrolytic capacitors rated at 10,000 µF total capacitance. At full load the bank carries a ripple current of roughly 35 A rms across a switching frequency band of 2–20 kHz. The capacitors are rated for 5.0 A rms ripple each, so eight parallel units share the load. In practice, the bank operates at 45°C ambient inside the enclosure, and the measured case temperature reaches 82°C under sustained load. Core temperature is higher. That difference—between case and core—is where many thermal design errors originate.
This article walks through the thermal budget for a capacitor bank in forced-air or free-convection duty, the parameters that matter when estimating hot-spot temperature, and a practical verification procedure that does not require thermal mockups.
Ripple-Current Heating and the Core-Temperature Limit
Aluminum electrolytic capacitors dissipate heat mainly through ESR. The power dissipated in each unit is approximately Irms² × ESR at the switching frequency. ESR falls as temperature rises, which partially self-limits heating, but the relationship is not linear and the core still runs well above case temperature. Polypropylene DC-link capacitors behave differently—ESR is low but stable—yet the same thermal-budget logic applies: the core must stay below the manufacturer's hot-spot limit.
For aluminum electrolytics, a typical hot-spot limit is 105°C. Lifetime doubles for roughly every 10°C below that limit, so a bank running at 95°C core temperature has roughly half the predicted lifetime of one running at 85°C. Derating ripple current by 10–15% under high ambient conditions is a standard design practice, but it is only meaningful if the core-temperature estimate is based on the actual ESR at the operating temperature, not on the 20°C datasheet value.
In polypropylene capacitor banks, hot-spot limits run higher—typically 115°C to 125°C—but the failure mode changes: sustained over-temperature accelerates film aging and can lead to abrupt dielectric failure. The practical consequence is the same regardless of technology: cooling is part of the electrical design, not an afterthought.
Airflow and Placement Rules for the Bank
Free-convection cooling works only when the bank has enough surface area per watt and vertical airflow paths are unobstructed. A useful rule of thumb: free-convection cabinets handle roughly 0.5–1.0 W/cm² of exposed capacitor surface area. Beyond that, forced airflow is required.
For forced-air design, the key parameter is volumetric airflow per watt dissipated. A practical starting point for capacitor banks is 0.03–0.06 m³/min per watt of total bank dissipation, depending on duct geometry and inlet temperature. Fans mounted to exhaust upward through a vertical channel are more effective than fans blowing perpendicular to the capacitor axes, because the cylindrical shape leaves open paths for air to pass between units.
Placement rules that matter in practice:
- Leave at least 10 mm between adjacent capacitor cans, and 20 mm between the bank and the nearest enclosure wall.
- Orient longer capacitor axes vertically so air flows along the can surface rather than across the terminals.
- Do not place the bank directly above inductors, transformers, or power resistors; the heat column from those components enters the capacitor air stream.
- When paralleling capacitors, keep the terminal interconnects symmetrical so that ripple current divides evenly. Uneven sharing causes one unit to run hotter than the others.
A Worked Thermal Budget and Component Selection Values
Consider a 200 kW motor drive with a DC-link bank of 12 aluminum electrolytic capacitors, each rated 8,200 µF / 450 V. Total ripple current at full load is 48 A rms. The switching frequency is 4 kHz. Each capacitor has an ESR of approximately 12 mΩ at 20°C, falling to about 8 mΩ at 85°C core temperature. The per-unit dissipation is roughly (4 A)² × 8 mΩ ≈ 0.13 W, and total bank dissipation is near 1.5 W. That seems small, but the core-to-case thermal resistance—typically 3–5 K/W for large screw-terminal capacitors—means the core sits 2–4°C above the case. Under 45°C ambient with restricted airflow, that contribution is manageable; the dominant term is usually ambient temperature rise inside the enclosure.
A more demanding case: a solar inverter with a ripple current of 120 A rms through a polypropylene DC-link bank. The per-unit dissipation is higher because a modest ESR of 3 mΩ still yields 43 W at full current. A bank of six units must dissipate roughly 7 W per unit, and forced airflow of 0.05 m³/min per watt becomes necessary to keep case temperature within 15°C of ambient.
| Parameter | Aluminum electrolytic | Polypropylene film |
|---|---|---|
| Typical hot-spot (core) limit | 105°C | 115–125°C |
| ESR behavior vs temperature | Falls as temperature rises | Nearly constant |
| Lifetime model | Hour-life doubles per ~10°C below core limit | Film aging accelerates above rated temperature |
| Self-heating at rated ripple | 5–15°C case rise | 3–8°C case rise |
| Typical core-to-case thermal resistance | 3–5 K/W (screw-terminal) | 1–3 K/W |
| Recommended airflow (forced) | 0.03–0.05 m³/min per watt | 0.02–0.04 m³/min per watt |
Measurement and Verification Procedure
Case-temperature measurement with a thermocouple is straightforward, but core temperature must be estimated. A practical approach:
- Measure case temperature at the hottest point near the center of the can under full load.
- Record the ambient temperature inside the enclosure, 100 mm upstream of the bank.
- Calculate the core-to-case differential using the manufacturer's thermal resistance value. For an 8,200 µF screw-terminal unit, a typical differential is 2–3°C at rated ripple, but check the datasheet.
- Verify that the estimated core temperature stays below the hot-spot limit with a safety margin of at least 5°C.
- If the margin is insufficient, increase airflow, reduce bank density, or select a capacitor with lower ESR and lower thermal resistance.
For production validation, thermal imaging on the case is useful but does not reveal the core. A thermocouple embedded in the vent plug of a sample unit provides a better estimate, though it modifies the unit and cannot be used for every product. The practical compromise is to measure case temperature in production and correlate it with a one-time lab measurement of core temperature using an embedded sensor.
When reworking an existing bank design, replacing individual capacitors with a different brand or series requires re-validating the thermal budget. ESR differences of ±20% between manufacturers lead to proportional changes in dissipation, and the case temperature at which a bank is deemed acceptable must be re-derived from the new part's thermal resistance and hot-spot limit.
Hot-spot temperature, not case temperature, is the parameter that appears in lifetime models. A bank that looks cool at 70°C case temperature may have a 95°C core if thermal resistance is high. Designs that keep the core below 85°C in 45°C ambient generally achieve acceptable lifetime for electrolytic technology. The same discipline applies to film capacitors, where the limit is higher but the consequence of exceeding it is more abrupt.
