The variable-speed drive is one of the most effective energy-saving devices ever deployed in industry. By matching a motor's speed to the actual demand on it, rather than running flat out and throttling the output mechanically, a drive on a centrifugal pump or fan can cut that machine's energy use by 30–50%. The physics is the affinity law: power varies with the cube of speed, so a fan slowed to 80% draws barely half the power. The business case is unanswerable, which is why drives have spread into virtually every pump, fan, compressor, conveyor, and HVAC system in the modern plant.
That same drive is, on its input side, the single largest source of harmonic distortion in most industrial facilities. The harmonics article on this site establishes that VFDs dominate the harmonic spectrum of a typical site. This article looks at why — what is happening inside the box, why the line-side mitigations differ so much in cost and effect, and the entirely separate set of problems the drive creates on its output side, between the drive and the motor, that rarely get discussed at all.
The uncomfortable shape of it: the energy a drive saves shows up at the motor. The distortion it creates shows up everywhere else — in the cables, transformers, and other equipment across the network. The saving is local; the cost is distributed. That is exactly why the cost stays hidden.
1. What is actually inside a drive
Almost every standard VFD follows the same three-stage architecture, called a voltage-source inverter:
- The rectifier (front end). A bridge of six diodes converts the incoming three-phase AC into DC. This is the line-side stage, and it is where current harmonics are born.
- The DC bus. A large capacitor (often with a smoothing inductor) holds the rectified voltage at a roughly constant DC level — an energy reservoir between the two conversion stages.
- The inverter (output stage). Six fast electronic switches — IGBTs — chop the DC bus voltage back into a three-phase AC output of variable frequency and voltage, which is what lets the motor run at any speed. This is the motor-side stage, and it is where the output voltage problems are born.
The two ends are almost independent. The rectifier's behaviour is set by the diodes and the DC bus; the inverter's behaviour is set by the IGBT switching. Understanding the drive means looking at each end separately.
2. The line side: why the rectifier makes harmonics
The DC-bus capacitor is the key to the line-side problem. It charges up to near the peak of the AC supply voltage and holds there. The diodes can only conduct — and therefore can only draw current from the supply — during the brief moments when the instantaneous supply voltage is higher than the voltage already sitting on the capacitor. That happens only near the very top of each voltage half-cycle.
So instead of drawing a smooth sinusoid across the whole cycle, the drive draws current in two narrow, tall spikes per cycle — one near each voltage peak — topping the capacitor back up. The motor receives smooth power from the DC bus; the supply sees a violently pulsed current. That pulsed current waveform is the harmonic distortion.
Why these specific harmonics
A six-pulse rectifier produces a mathematically predictable set of harmonics called the characteristic harmonics, given by the order h = 6k ± 1 for k = 1, 2, 3… That yields the 5th, 7th, 11th, 13th, 17th, 19th, and so on. The 5th and 7th are by far the largest. The triplen harmonics (3rd, 9th, 15th) are largely absent from a balanced three-phase rectifier — a useful contrast with the triplen-rich signature of single-phase electronics like LED drivers and computer supplies. If you see a current spectrum dominated by the 5th and 7th, you are almost certainly looking at drives.
The magnitude depends on what, if anything, has been fitted to soften the current pulse. A bare six-pulse drive with no line reactor and no DC choke can pull 80–100% current THD at its own terminals — a current waveform that is barely recognisable as alternating. Add the standard mitigations and that figure falls dramatically, which is the subject of the next section.
3. The line side: the ladder of mitigations
There is a well-defined ladder of line-side fixes, rising in cost and effectiveness. Knowing where a given drive sits on this ladder tells you most of what you need to know about its harmonic behaviour.
AC line reactor or DC link choke
A series inductance — either on the AC input (line reactor) or in the DC bus (choke) — resists the sudden current spike, spreading the draw over a wider slice of the cycle. Cheap, compact, and standard on better drives. It brings a bare drive from 80–100% THD down to roughly 35–45%. It is a major improvement and nowhere near a solution — 40% distortion is still well outside any connection standard.
Twelve-pulse and eighteen-pulse rectifiers
Feeding the drive through a phase-shifting transformer with two (12-pulse) or three (18-pulse) sets of diodes arranged so their harmonics partially cancel. A 12-pulse front end removes the 5th and 7th — the two largest — leaving the 11th and 13th as the new lowest characteristic harmonics, for roughly 10–15% THD. An 18-pulse design pushes the lowest harmonic higher still, reaching 5–8%. The cost is a bulky, expensive phase-shifting transformer and a larger footprint, and the cancellation degrades if the supply voltage is unbalanced.
Active front end (AFE)
Replacing the passive diode bridge with a second bank of actively switched IGBTs that draw a near-sinusoidal input current and can even return braking energy to the supply. An AFE can hold input THD to under 5% and is the cleanest front end available — but it is the most expensive option, adds its own high-frequency switching noise, and is found on only a small minority of installed drives.
How drives combine at the bus
A plant rarely has one drive. It has dozens, and their harmonics do not simply add arithmetically. Because each drive's current pulses sit at slightly different phase angles — depending on cable lengths, loading, and exactly when each one happens to be drawing — there is partial cancellation, an effect called diversity. The combined THD at the main bus is therefore usually lower than the THD of any single drive measured alone. This is genuinely good news, but it is also why bench figures and nameplate ratings cannot tell you a site's real distortion. Diversity depends on the specific layout and the specific load pattern. The only way to know the number is to measure the actual bus.
4. The motor side: the problem no one mentions
Everything so far concerns the line side — the harmonics a drive pushes back into the supply. But the output stage creates a completely separate family of problems on the cable and motor it feeds, and these are routinely overlooked because they do not show up on an energy bill at all. They show up as failed motors.
The inverter does not produce a smooth sine wave. It synthesises the output by switching the full DC-bus voltage on and off thousands of times a second — pulse-width modulation (PWM). The motor's inductance smooths the resulting current into something close to sinusoidal, but the voltage at the motor terminals remains a train of extremely steep-edged pulses. That steepness is the root of three distinct failure mechanisms.
dV/dt and voltage stress
Each PWM pulse rises from zero to full DC-bus voltage in tens of nanoseconds — an enormous rate of voltage change, written dV/dt. The first turns of the motor winding absorb a disproportionate share of that steep edge, stressing the inter-turn insulation far harder than a smooth sine wave of the same RMS voltage ever would. Over time this erodes the insulation and brings on premature winding failure.
Reflected-wave voltage doubling
This is the one that surprises people. When the motor cable is long, it behaves as a transmission line, and a mismatch between the cable's impedance and the motor's allows each steep voltage pulse to reflect at the motor terminals and superimpose on the incoming pulse. The result can be a transient peak approaching twice the DC-bus voltage at the motor. The effect becomes significant on cable runs beyond roughly 15–50 metres depending on the switching speed, and longer cables make it worse. A motor perfectly happy on a short cable can have its insulation hammered to death simply because it sits at the far end of a long run from the drive. NEMA MG 1 Part 31 specifies the peak voltage that inverter-duty motors must withstand precisely because of this mechanism.
Bearing currents and shaft damage
The high-frequency switching also induces a voltage on the motor shaft. When that voltage exceeds the breakdown threshold of the thin oil film in the bearing, it discharges through the bearing as a tiny spark — electrical discharge machining (EDM) in miniature. Each discharge pits the bearing race. Over months the pitting develops into a washboard pattern called fluting, the bearing roughens and overheats, and it fails — often long before the motor itself would have. The motor is replaced, the new one fails the same way, and the underlying cause is never identified because nobody connects a mechanical bearing failure to the drive feeding it.
5. Why it stays hidden
Put the two sides together and you can see why the cost of drive distortion goes unnoticed for years.
The drives were never installed as a single project. They arrived incrementally — a pump upgraded here, a new air handler there, a conveyor retrofit the year after. Each addition was justified, correctly, on the energy it would save the motor it controlled. No individual step looked like a power-quality decision, and no one ever stepped back to total up what the accumulated population of drives was doing to the network as a whole.
The line-side harmonics, meanwhile, do their damage distributed across the network — a few percent of extra loss in this transformer, a slightly hotter cable there — never concentrated anywhere obvious. The motor-side damage shows up as bearing and winding failures that read as ordinary mechanical wear, attributed to anything but the drive. And the energy saving from the drives is large, real, and visible, which makes the whole drive population look like an unalloyed good. The saving masks the cost.
None of this is an argument against variable-speed drives. They are indispensable and their energy savings are real. It is an argument for treating the harmonics they create as the predictable, manageable side-effect they are — rather than an invisible tax the plant pays forever without noticing.
6. What to do about it
The sequence is the same as for any power-quality problem, and it begins with measurement because diversity and cable layout make every site genuinely different.
- Measure the real spectrum at the bus and at the larger drives, not just at the incomer. A 5th-and-7th-dominated current signature confirms drives are the source and quantifies how much diversity is helping you.
- Treat the line side with correction sized to the measured spectrum. Active harmonic filtering cancels the 5th, 7th, 11th, and 13th in real time and tracks the spectrum as drives ramp and cycle — which fixed, passive elements tuned for one condition cannot.
- Correct close to the drives, not only at the meter. Because the harmonics are produced deep in the plant, cancelling them near the drives keeps the dirty current out of the internal network entirely — the argument set out in full in our article on correction location.
- Address the motor side on its own terms where long cables or repeat bearing failures point to it — output filters, inverter-duty motors, and shaft grounding. This is a distinct intervention from line-side correction, not a substitute for it.
HarmoniQ's three-component architecture targets the line-side half of this directly: the Filter cancels the drive harmonics actively, the Alpha conditions the line and stabilises voltage as drives switch, and the Booster corrects true power factor across the changing load. Positioned against the plant's real drive population rather than parked at the incomer, it keeps the distortion the drives create from spreading across the network. The detail is in the product documentation.
Summary
Variable-speed drives are the defining efficiency technology of the modern plant and, simultaneously, its dominant source of distorted current. On the line side, the rectifier and DC-bus capacitor draw current in sharp pulses, producing the characteristic 5th, 7th, 11th, and 13th harmonics — softened to 35–45% by the line reactor that most drives carry, but no further unless a 12-pulse, 18-pulse, or active front end has been paid for. On the motor side, PWM switching imposes steep voltage edges that stress winding insulation, double up on long cables through reflected waves, and erode bearings through discharge currents — an entirely separate problem with its own cures.
The reason all this hides so well is that the saving and the cost land in different places. The energy saving is concentrated, visible, and on the bill. The distortion cost is spread thinly across the network and disguised as ordinary equipment wear. The drives arrived one upgrade at a time, and no one ever added up the total.
The fix is not to abandon drives but to measure what their accumulated population is actually doing, correct the line-side harmonics close to where they are produced, and treat the motor-side stresses on their own terms. Done properly, you keep the energy saving the drives were installed for — without paying the hidden distortion tax that came bundled with it.