Variable-frequency-drive- (VFD-) generated harmonics largely is a perceived, rather than real, issue. In 27 years of applying VFDs in HVAC and other applications, this author has experienced only a handful of actual harmonics problems, with all but one stemming from high levels of voltage distortion, not the current distortion that has been getting so much attention lately.

Most of the VFD-interference problems this author has encountered have been the result of poor installation — particularly, poor wiring and grounding. In the majority of cases, radio-frequency interference (RFI) or electromagnetic interference (EMI), not harmonics, was the culprit. RFI/EMI issues stem from noise in the 50-Khz-to-low-megahertz range, not the 300-Hz fifth or 420-Hz seventh harmonic range.


In 1981, ANSI/IEEE Standard 519, IEEE Guide for Harmonic Control and Reactive Compensation of Static Power Converters, was published. It included maximum total-harmonic-voltage-distortion (THDV) recommendations.

In the extreme, voltage distortion can cause flat-topping of power-system voltage waveforms (Figure 1), which can cause sensitive electronic processors to become confused and malfunction.

In 1992, ANSI/IEEE Standard 519 was revised. Renamed IEEE Recommended Practices and Requirements for Harmonic Control in Electrical Power Systems, it now concentrates more on total harmonic current distortion (THD1) than voltage distortion.

THD1 can propagate through utility step-down/step-up transformers and make its way from one facility to another. For example, several years ago, a VFD manufacturer was creating high amounts of current distortion during its burn-in testing operation. The current distortion traveled through the utility transformers at the VFD manufacturer's plant to the utility feed at a neighboring printing plant, corrupting the logic circuits in the controls and direct-current (DC) drives running the printing plant's printing press and causing the printing-press registration to malfunction.

THD1 results in additional heat in the distribution transformers typically provided by utilities, as well as the power-feeder cables of the equipment from which it originates. Basically, THD1 is current that a utility has to generate and source to a facility, but that brings no revenue to the utility. While it is a real issue for utilities, THDI largely is a perceived problem from a facility manager's point of view.

ANSI/IEEE Standard 519-1992 addresses the system-issue nature of THD1 by introducing total demand distortion (TDD), which can be calculated as follows:


Ihe = total harmonic current as measured by system

Ihc = total harmonic current contributed by VFDs

IL = maximum demand-load current (fundamental frequency component) (15- or 30-min demand) at utility point of common coupling (PCC) as measured in system

IC = fundamental frequency component contributed by vfds (included only if vfds are an addition to existing loads)

(All quantities are in amperes root mean square.)

Page 2 of 5

ANSI/IEEE Standard 519-1992 states, “Within an industrial plant, the PCC is the point between the nonlinear load and other loads.” Many consulting engineers have interpreted this to mean that THD1 is to be measured at VFD input-power connections (PCC2, instead of PCC1, in Figure 2). This misapplication of ANSI/IEEE Standard 519-1992 has contributed to the overuse of multipulse drives in the HVAC industry. Many millions of facility-equipment dollars have been squandered through the specification and installation of 12- and 18-pulse drives in commercial office buildings and other environments in which a standard six-pulse drive would have done the same job for substantially less upfront cost.

Also unfortunate is the fact ANSI/IEEE Standard 519-1992 has five different levels of acceptable maximum TDD, which depend on the ratio of maximum short-circuit current (ISC) to maximum IL at a PCC. The ISC-to-IL ratios in Table 1 are functions of the strength of a utility's feed to a facility and the size of the substation transformer.


Many specifications simply state, “VFDs shall meet ANSI/IEEE Standard 519.” Such a statement is meaningless without the information needed to perform harmonic calculations:

  • Transformer kilovolt-amperes and percent impedance.

  • Total linear connected-load amperage or total expected linear connected amperage.

  • The number and sizes of VFDs.

  • Utility ISC available.

Calculations are even more accurate when manufacturers have additional information, such as facility total current, existing harmonic content, and wire sizes and lengths.

Some engineers have taken to writing hardware specifications based on horsepower size requirements. For example: “All VFDs 100 hp and up shall be 18-pulse designs.” At 100 hp, an 18-pulse drive easily can cost four times as much as a six-pulse drive with no improvement in energy savings.

Page 3 of 5

That is not to say there are no applications for which a 12- or 18-pulse drive is appropriate. Take, for instance, a cinderblock pump station in a residential neighborhood. This author observed one in which there were three 300-hp VFDs, overhead fluorescent lighting, and a wall-mounted programmable logic controller (PLC). The pump station was fed by a dedicated 480-v transformer. Virtually the entire load on the transformer was non-linear. The VFD non-linear load represented approximately 1,100 amps. The PLC and fluorescent-light loads totaled a couple of amps. That was an ideal application for 18-pulse or other ultralow-harmonic VFD technology.

In a commercial office building, if VFDs are installed on every fan and pump, they typically will use less than 20 percent of the electrical demand load. In almost all such cases, standard six-pulse drives are a good choice.

Contrary to popular belief, ANSI/IEEE Standard 519 is not a law or government/utility regulation; it is a “recommended practice.” It states that strict adherence to its recommended harmonic limits “will not always prevent problems from arising.” The contrary also is true: A facility may have harmonics in excess of the standard's maximum recommended limits and not experience difficulties.


The simplest and least-expensive method of mitigating VFD-generated harmonics is adding impedance at a VFD. This can be accomplished with an input line reactor (Figure 3) or a DC link reactor (bus choke) (Figure 4). In a 1-percent-source-impedance system, a 3-percent line reactor can reduce harmonic-current content at the input to a VFD to about 40 percent at full-load output.

The next-most-common type of harmonic-mitigation technology is the 12-pulse VFD (Figure 5). A 12-pulse VFD reduces harmonic-current content to about 10 percent.

Also common are broad-band and passive filters (Figure 6). These hybrid filters reduce harmonic-current content to approximately 7 percent.

The next-most-effective technology is the 18-pulse drive (Figure 7), which typically presents approximately 5-percent current distortion at VFD inputs. Compared with a VFD with no impedance, total harmonic reduction is in the range of 93 percent.

Relatively new technologies are the active harmonic filter (Figure 8) and the active-front-end VFD (Figure 9). A single active filter can filter the harmonics of several VFDs or an entire facility. Meanwhile, the THDI content of a VFD with an active front end — measured at the VFD input — typically is less than 4 percent, while the total-harmonic-current-content reduction is 95 percent.

Table 2 lists the expected current distortion, percent current-distortion reduction, and relative cost of the various harmonic-reduction technologies. The estimates are based on a 1-percent-source-impedance system and a perfectly balanced voltage supply.

Page 4 of 5

All hardware-based, “brute-force” methods of harmonic reduction are affected negatively by input-power-system voltage imbalances. Most VFD manufacturers have computer programs that can be used to estimate harmonic distortion from VFDs.

The greater the base load on a substation transformer, the lesser the current distortion at a PCC. Because harmonic-current distortion causes additional transformer heating, utilities often oversize substation transformers relative to the loading expected from a facility. As a result, having the correct maximum transformer load (estimated or measured) is vital. Otherwise, maximum transformer IL must be assumed.


Most harmonic-analysis programs assume available power is a balanced voltage — for example, 480 v each on Phase A, Phase B, and Phase C. In the real world, however, no matter how well-designed a building distribution system is, perfect balance is unobtainable. The best one can hope for is a slight imbalance, such as 478:480:482 v. Most utilities allow power-voltage imbalances of up to 3 percent.

Many years ago, at a large university in the Midwest, the VFDs provided in an energy-saving retrofit project were being blamed for buildings exceeding the distortion levels recommended in ANSI/IEEE Standard 519. Harmonic analysis showed substantial third-harmonic content. In a perfect world, VFDs do not create third harmonics, as third and other triplen harmonics cancel because of the three-phase nature of VFDs. If, however, the voltage relationship between phases A, B, and C is unbalanced, cancelation cannot occur completely, and VFDs can create triplen harmonics. In this case, Phase A was approximately 450 v, while phases B and C were close to 480 v. The university was asked to move loads to get the input voltage to a more balanced condition. Once that was done, the VFDs stopped causing elevated levels of harmonic distortion.

During the mid-1990s, the Power Electronics Applications Center, a subsidiary of the Electric Power Research Institute, tested the drives of 17 manufacturers.1 A 0.2-percent voltage imbalance at the input lugs of a VFD with no input line reactor or DC-bus choke was found to cause up to a 17-percent current imbalance.

With an unbalanced input-power system, all hardware-based harmonic-mitigation technologies are subject to detrimental harmonic-cancelation effects. For example, a 12-pulse phase-shifting transformer has three input leads and six output leads and two components: a delta/delta winding set and a delta-wye winding set (Figure 10). This configuration causes a 30-degree electrical-phase shift in the power being fed into one of the drive's two diode bridges, causing, in a perfect world, fifth and seventh harmonics to be canceled. If input power is unbalanced, however, cancelation will not occur completely.

Some VFD manufacturers supply 18-pulse drives with an additional 5-percent impedance reactor in front of the auto transformer. This helps balance the current draw into the auto transformer's three sets of windings and helps to minimize the effects of unbalanced voltage and source feeds.


The most effective means of obtaining ultralow harmonics at VFD inputs is an active filter or an active front end. An active filter works like an active noise-reduction headset. If, for example, it detects a 30-amp fifth harmonic in Phase A of a power supply, it injects 30-amp fifth harmonic 180 degrees out of phase with the VFD-created harmonic, creating a cancelation effect. This technology is less susceptible to incoming-voltage imbalances because it measures and injects corrective harmonic content automatically.

Page 5 of 5

A few manufacturers make ultralow-harmonic VFD technologies. An ultralow-harmonic VFD has six insulated-gate bipolar transistors (IGBTs), rather than passive diode-bridge components, in its convertor section (Figure 11). These IGBTs control the harmonic current drawn by a VFD. With no harmonic current drawn, no cancelation is required. Ultralow-harmonic technology typically reduces input harmonic currents to 4 percent or less at a VFD input (Table 2).

In one test, a 3-percent voltage imbalance on the input of an 18-pulse transformer/drive caused a 1.5-percent-per-unit increase in current distortion. Thus, if the computer harmonic-analysis estimate had been 4 percent, actual THDI would have been 5.5 percent.

With an ultralow-harmonic or active-filter system, a 3-percent voltage imbalance increases harmonic-current distortion by less than 0.5 percent per unit.


A harmonic analysis should be performed before a design is finalized. The analysis should be conducted at the PCC to determine current distortion at the main utility service entrance to a building. Hardware-based specifications dictating that any drives over a certain horsepower shall be a certain technology should not be utilized.


1) Mansoor, A., Phipps, K., & Ferro, R. (1996). System compatibility research: Five horsepower pwm adjustable-speed drives. Knoxville, TN: Power Electronics Applications Center.

For past HPAC Engineering feature articles, visit

The manager of HVAC applications for ABB Inc. Power & Control Sales, Michael R. Olson has extensive experience in the HVAC, water/wastewater-treatment, and chemical industries. He has written numerous trade-journal articles discussing the application of adjustable-speed drives and been a contributing editor to several books on the subject. He has a bachelor's degree in electrical engineering from the University of Illinois and a master's degree in engineering management from the Milwaukee School of Engineering. He is a member of the American Society of Heating, Refrigerating and Air-Conditioning Engineers and BACnet International. He can be contacted at