In plants with harmonics problems, filter choice often comes down to comparing quotations. But choosing between an active vs passive harmonic filter is a measurement decision, not a catalogue one. For the same group of drives, the right answer may be a detuned capacitor bank in one plant, an active filter in another and a combination in a third.
This guide skips the basics of what harmonics are; for those, see our overview of harmonics and reactive power monitoring with energy analyzers. Here the question is: with measurement data in hand, which solution should you choose and how should you size it? Atasayın does not sell harmonic filters; our role is measuring before and after the decision, so the comparison below favours no technology.
Harmonic Filter Types: A Short Map
1. Detuned capacitor banks
A reactor is connected in series with each power factor correction capacitor step. The goal is not to eliminate harmonics but to protect the capacitors from harmonics and resonance. The LC circuit’s resonant frequency is pulled below the lowest dominant harmonic, so the bank looks inductive to every harmonic above it and cannot resonate with the network impedance.
The reactor is specified by its detuning factor p, and the resonant frequency is fr = fsupply / √p. Common values on a 50 Hz network:
- p = 7% → about 189 Hz (3.78th harmonic): the default for drive-heavy plants where the 5th harmonic dominates.
- p = 5.67% → about 210 Hz (4.2nd): absorbs more 5th harmonic but loads the capacitors harder.
- p = 14% → about 134 Hz (2.67th): for networks with a significant 3rd harmonic, such as UPS, single-phase electronics or LED-heavy buildings.
A detuned bank lowers THDi slightly, but that is a side effect, not its job.
2. Tuned passive harmonic filter
A passive harmonic filter is an LC branch tuned close to one harmonic, giving that current a low-impedance path. Because of component tolerances and the risk of absorbing harmonics from the utility side, it is tuned slightly below the target, typically around 4.7 for the 5th. Each harmonic needs its own branch.
It works well on steady, predictable loads. On the downside, it supplies capacitive reactive power at the fundamental, can be overloaded by harmonics from elsewhere on the network, and drifts when network impedance changes.
3. Active harmonic filter
An active harmonic filter reads the load current through CTs, extracts the harmonic components and injects equal and opposite currents through an IGBT inverter. It is not tuned; it covers a wide band (typically up to about the 50th harmonic, depending on the manufacturer) and follows load changes dynamically. Many models also correct power factor and balance phases. It is sized in amperes of compensation current, with modules paralleled to build capacity.
4. Hybrid solution
A detuned bank covers the base reactive power, and a smaller active filter handles the residual harmonic current. In drive-heavy plants that also need a lot of kVAr, this is the usual way to reduce active filter capacity and total investment.
Active vs Passive Harmonic Filter: Comparison Table
| Criterion | Detuned bank | Tuned passive filter | Active harmonic filter |
|---|---|---|---|
| Primary purpose | Protect capacitors, correct power factor | Absorb specific harmonics | Cancel harmonic current across a wide band |
| Harmonic reduction | Limited | High at the tuned order | High, across many orders |
| Variable load | Stepwise (via controller) | Poor | Dynamic |
| Resonance risk | Low if p is chosen correctly | Sensitive to network changes | None (not a tuned circuit) |
| Sizing unit | kVAr + p (%) | kVAr + tuning frequency | A (harmonic current) |
| Leading power factor at light load | Managed by controller | Yes | No |
| Cost profile | Low capital cost | Medium, project-specific engineering | High capital cost, modular growth |
We give no prices because the cost drivers vary widely: kVAr and harmonic amperes required, voltage level, enclosure and cooling, 3-wire versus 4-wire design and the distortion target. Detuned banks are priced per kVAr and active filters per ampere, so the two are not directly comparable.
The Measurement Campaign: What to Measure, and for How Long
Most selection mistakes trace back to incomplete or badly timed measurement. You need:
- At least one full week of logging, so shifts, weekend base load and peak production appear in the same data set. IEEE 519 also assesses compliance with weekly statistics.
- Logging at the main LV incomer and at the feeders of major harmonic sources: drive panels, UPS, welding lines.
- Where possible, recordings with capacitor steps in and out; resonance only shows up in that comparison.
- THDv, THDi and the individual spectrum (3rd, 5th, 7th, 11th, 13th) in percent and in amperes, plus fundamental current, reactive power, power factor, demand, min/max values and event logs.
The key point: THDi percentage alone is not sizing data. At light load THDi may read 40% on a few amperes; at full load 20% can mean hundreds of amperes. Filters are sized to the highest harmonic current, which is why IEEE 519 uses TDD, referenced to maximum demand current.
IEEE 519-2022 and IEC 61000 Limits
IEEE 519-2022 voltage limits at the point of common coupling (PCC): at 1 kV and below, 5% individual harmonic and 8% THD; from 1 kV to 69 kV, 3% and 5%.
Current limits (120 V to 69 kV) depend on the ratio of short-circuit current to maximum demand current (Isc/IL):
| Isc/IL | 3 ≤ h < 11 | 11 ≤ h < 17 | 17 ≤ h < 23 | 23 ≤ h < 35 | 35 ≤ h ≤ 50 | TDD |
|---|---|---|---|---|---|---|
| < 20 | 4.0 | 2.0 | 1.5 | 0.6 | 0.3 | 5.0 |
| 20–50 | 7.0 | 3.5 | 2.5 | 1.0 | 0.5 | 8.0 |
| 50–100 | 10.0 | 4.5 | 4.0 | 1.5 | 0.7 | 12.0 |
| 100–1000 | 12.0 | 5.5 | 5.0 | 2.0 | 1.0 | 15.0 |
| > 1000 | 15.0 | 7.0 | 6.0 | 2.5 | 1.4 | 20.0 |
Values are percentages of IL. In the 2022 edition only even harmonics up to the 6th are limited, at 50% of the odd limits. Compliance is judged on weekly 95th percentile values. IEC 61000-2-4 sets an 8% voltage THD compatibility level for Class 2 industrial networks.
Decision Flow: From Measurement to Filter Type
Step 1: Calculate the resonance risk
The parallel resonance order of a capacitor bank is approximately hr ≈ √(Ssc / Qc), where Ssc is the short-circuit power at the connection point and Qc the capacitor power in service. A 1600 kVA transformer with 6% impedance gives Ssc of about 26.7 MVA (ignoring upstream impedance). With 1000 kVAr in service, hr ≈ 5.2, about 258 Hz, right on the 5th harmonic; at 600 kVAr it moves to about 6.7, near the 7th. As an automatic bank switches steps, resonance sweeps across harmonic orders. That is why unreactored capacitors are risky wherever harmonic sources exist.
Step 2: Look at the dominant orders
- 5th and 7th dominant, low 3rd → a 7% detuned bank is the first candidate.
- Significant 3rd harmonic → 14% detuning; with high neutral current, consider a 4-wire active filter.
- Broad spectrum including 11th, 13th and above → passive branches multiply and an active filter gains ground.
Step 3: Assess load variability
Narrow-band harmonic current suits passive solutions; fast, wide swings (presses, welding, cranes) favour active or hybrid designs.
Step 4: Size to the target
For a detuned bank, start from the reactive power requirement and apply two corrections. The reactor raises capacitor voltage by about 1/(1−p): roughly 430 V on a 400 V network at 7%, before harmonics, so the capacitor voltage rating must allow for it. And a reactored step’s real kVAr differs from the capacitor nameplate, so ask for the effective step output. Where reactive power charges are the concern, our industrial electricity bill analysis guide shows where they appear.
For an active harmonic filter, a practical approximation is: required current ≈ I1 × (THDimeasured − THDitarget). An incomer with 800 A fundamental current and 30% THDi carries about 240 A of harmonic current; for a 5% target the filter needs roughly 800 × 0.25 = 200 A, plus the manufacturer’s recommended margin. Use the peak harmonic current in the weekly log, not the average.
Step 5: Consider a hybrid
Active filter capacity is expensive kVAr. Leave reactive power to a detuned bank and size the active filter for residual harmonics only.
The Most Common Mistakes in the Field
- Plain capacitors on a harmonic-rich network. Swollen capacitors and blown fuses usually signal resonance.
- Sizing from nameplates. Adding up drive ratings and assuming a “typical THDi” ignores diversity and real loading.
- Measuring for one day or at light load. Missing peak production leads to an undersized active filter.
- Looking only at THDi percentage instead of harmonic amperes and the individual spectrum.
- Misplaced active filter CTs. For closed-loop control they belong on the supply side, with correct polarity.
- Ignoring light-load behaviour of passive filters, which pushes the plant towards a leading power factor at night and weekends.
Verification After Installation
After commissioning, record again at the same points, for the same duration and under similar production. Compare THDv and THDi/TDD against targets, the ampere reduction at dominant orders, power factor and inductive/capacitive reactive energy, and active filter utilisation: if it sits at 100% most of the time, it is undersized.
Performance drifts as loads change and components age, so leave a permanent measurement point on the main incomer. Atasayın’s ATS-11P and other ATS-11 series energy analyzers measure current and voltage harmonics, inductive and capacitive reactive power, power factor, demand, min/max values and event logs. Sent over RS485 Modbus RTU or Modbus TCP to the ATS Energy Monitoring System, this data puts pre- and post-filter trends side by side. See panel-door vs DIN-rail energy analyzers for choosing the format, and what is an energy monitoring system for the wider picture.
Frequently Asked Questions
What is the difference between an active and a passive harmonic filter?
A passive filter is an LC circuit tuned near one harmonic; it suits steady loads and also supplies capacitive reactive power. An active filter measures load current and injects opposing currents across a wide harmonic band, adapting to load changes. It is sized in amperes rather than kVAr.
What is the difference between 7% and 14% detuned reactors?
On a 50 Hz network, 7% gives a resonant frequency of about 189 Hz and 14% about 134 Hz. Use 7% where the 5th harmonic dominates and 14% where the 3rd harmonic is significant, based on the measured spectrum.
How do you size an active harmonic filter?
In amperes: roughly the full-load fundamental current multiplied by the difference between measured and target THDi. For 800 A with 30% measured and 5% target THDi, that is about 200 A plus the manufacturer’s margin.
How long should you measure before selecting a harmonic filter?
At least one full week, so that shifts, weekend base load and peak production appear in the same recording. IEEE 519 compliance is also assessed with weekly statistics.
What harmonic levels does IEEE 519 allow?
Under IEEE 519-2022, systems at 1 kV and below are limited to 8% voltage THD and 5% per individual harmonic. Current TDD limits range from 5% to 20% depending on the short-circuit to demand current ratio.
How do you verify a harmonic filter after installation?
Repeat the measurement at the same points under similar production and compare THDv, THDi/TDD, individual harmonic currents and reactive energy. Keep a permanent energy analyzer on the main incomer, because performance drifts over time.
Conclusion
Neither filter type is universally better; the right one matches the plant’s measured spectrum, load variability and resonance risk. Steady loads with a dominant 5th harmonic often need only a detuned bank; variable loads, broad spectra or strict targets favour active or hybrid solutions. Either way, the decision starts with a week of measurement and is confirmed by measuring again.
To map your plant’s harmonic profile before investing in a filter and track the result afterwards, contact us to plan the measurement infrastructure together.