This article focuses on the latest technology, called the insulated gate bipolar transistor that has increased switching frequencies almost twentyfold from the last few years. It turns out that the newer variable-frequency drives, while working full-time shifts as speed controllers of ac induction motors, may be moonlighting with a little part-time electrical discharge machining (EDM). The new drives are suspected of setting up bearing currents. These currents, acting in a fashion identical to the mechanism of EDM—the manufacturing technique by which metal is removed from a workpiece through the use of high-energy sparks—have been eroding the material from the races of motor bearings. In some cases, stray currents are destroying bearings a few months after startup. ABB recently patented a motor winding designed to eliminate circulating bearing currents. The design divides the stator winding into an even number of equal parts per phase. The groups are then distributed uniformly between ac supply connections at both ends of the stator.


The Introduction of silicon control rectifiers to motor control in the early 1980s, with their 360-hertz switching frequencies, brought a clicking sound to the plant floor that hadn’t been heard before. As the technology behind adjustable frequency drives developed, and faster Darlington bipolar transistors replaced SCRs, switching frequencies jumped from 360 to 3,000 Hz. A click was swapped for a buzz.


According to ABB product marketing manager Mark Kenyon, the latest technology, called the insulated gate bipolar transistor, has increased switching frequencies almost twentyfold from what they were a decade ago.

It turns out that the newer variable-frequency drives, while working full-time shifts as speed controllers of ac induction motors, may be moonlighting with a little part-time electrical discharge machining. The new drives are suspected of setting up bearing currents. These currents, acting in a fashion identical to the mechanism of EDM— the manufacturing technique by which metal is removed from a workpiece through the use of high-energy sparks—have been eroding the material from the races of motor bearings. In some cases, stray currents are destroying bearings a few months after startup.

Don Macdonald, a sales representative with Toshiba International Corp. of Houston, said, “It is the fast rise times of IGBTs that has led to bearing problems. The ability of IGBTs to switch at very fast rise times has also increased their ability to switch at higher carrier frequencies. A combination of high carrier frequencies and fast rise times dramatically increases both insulation stress and the likelihood of common bearing currents.”

Recipe for Erosion

Bearing currents are not new. Also called shaft currents, the phenomenon has been around since electric motors were invented, said Thomas Lipo, a professor of electrical engineering at the University of Wisconsin. They have previously been attributed to rotor eccentricity, homopolar flux effects, and electrostatic discharge. As the use of adjustable speed drives has increased, however, observers have detected a corresponding rise in bearing failures, leading many experts to suspect a corollary. “Whether an inverter can, in fact, cause damaging bearing currents must remain an unresolved mystery before an acceptable theory and more solid evidence is available,” write Lipo and Shaotang Chen, of the GM Research and Development Center in Warren, Mich., in a paper on the topic.

Designers of adjustable-frequency drives (also called PWM, for pulse width modulated drives) describe common mode voltage as one source of bearing currents. Common mode voltage develops in a PWM three-phase power supply as the voltage pulses fail to add vectorially to zero. This differs from a typical sinusoidal three-phase power supply whose three legs remain balanced and symmetrical under normal conditions.

The source of the imbalance stems from the operation of the PWM drive itself. Essentially, a PWM drive runs three-phase ac power through individual diode bridge converters, transforming each of the three legs to dc power. The dc voltage feeds onto a dc bus, where it is filtered and smoothed. Then, the dc voltage moves onto the inverter, which converts the dc voltage back to ac. The voltage and frequency of the inverter output can be manipulated to vary motor speed.

The output from the inverter does not actually produce the sinusoidal shape of the ac power waveform feeding into the drive. Instead, the inverter approximates the waveform with a series of voltage pulses. The pulses have constant magnitudes, but vary in duration. Insulated-gate bipolar transistors in the drive switch on and off to modulate the width of each pulse.

Unlike the three phases of sinusoidal power supply that always add to zero, the three phases of the PWM drive, although they balance in peak amplitude, do not balance between phases instantaneously because the pulses are of different widths. The resulting common mode voltage is a source of bearing currents.

Acccording to Lipo and Chen, “There are at least three mechanisms of bearing current generation discovered so far.” The three can occur alone, together, or not at all. It depends on the electrical characteristics of a particular bearing.


Lipo said that two of the three known kinds of bearing currents arise through capacitive discharge of voltage built up on the motor shaft and stator frame. The third kind arises from a different mechanism altogether, he said. It causes a current to circulate through the shaft, bearings, and motor housing. Any explanation of the three varieties of bearing currents should start with the two simpler cases of capacitive discharge.

The first of the bearing current generators develops with the discharge of the airgap capacitor, Lipo said. In motors where the bearings are uninsulated, a charge builds up on the rotor shaft due to the common mode voltage. The balls in the motor bearings ride upon a thin oil film that acts to insulate the rotor shaft from ground.

Every so often, write Lipo and Chen, “The shaft voltage will discharge to its only load—the bearings—and produce a bearing current spike. This occurs when bearings exhibit high internal impedance for a certain period and then suddenly become short-circuited with a low impedance by touching the bearing race.” Breakdown of the lubricant can lead to short-circuiting as well, they add.

“Capacitor discharge seems to be the dominant phenomenon when the motor is supplied with a frequency in the range of 2 Hz to 55 Hz,” they write. “Since motors are mostly operated in this frequency range, the discharge mechanism has been believed to be the major cause of bearing damage.”

This kind of discharge, Lipo said, happens quickly, often within a few microseconds. Thus, this particular mechanism represents the greatest threat to bearing life. High-intensity discharges damage bearings the most.

A second capacitive mechanism stems from the formation of what Lipo called “parasitic capacitances” between the stator windings and the rotor.

“If the effective bearing impedance becomes very small, the airgap capacitor will be short-circuited by the bearings and all currents in winding parasitic capacitors will flow into the bearings,” write Lipo and Chen.

The third mechanism discovered so far is more complex, said Lipo. In the first two mechanisms, the currents flow only one way: from the shaft through the bearings to the stator frame, then to ground. In the third mechanism, the currents actually circulate. It is an inductive rather than a capacitive effect.

“As common mode voltages produce coupling currents to the rotor, they also supply much higher coupling currents to the stator,” write Lipo and Chen, “since the winding capacitance to the stator is much larger than to the rotor. All common mode currents come from the three motor input terminals and they never flow back to the terminals.”

Although the voltage in the three phases of the motor current will not add up to zero, Lipo said, they will total to the value of the common mode current. Thus, a net magnetic flux encloses the motor shaft. “Consequently, a back EMF will be induced in the conductive loop formed by the shaft, the bearings, and the stator enclosure,” he said.

“The EMF is usually very small, in the millivolt range,” Lipo said. “Its contribution to the shaft voltage can be ignored. However, when the impedance of this loop is sufficiently low, a circulating current will pass through the bearings. This becomes the circulating-type bearing current caused by the inverter.”

Bearing Autopsy

According to Macdonald, fluting damage is a telltale sign for bearings undergoing electrical discharge machining through an oil film. In advanced cases, deep flutings space out evenly along the outer races.

The initial EDM events leave only microscopic evidence behind, Macdonald said. When the oil film thins locally from variations in temperature or viscosity—or from changes in radial loading or vibration— and voltage on the shaft exceeds the dielectric strength of the oil film, electrical energy discharges, he said. If the energy is great enough, the discharge melts a tiny pit in the surface of the race.

Bearings in the initial stages of EDM destruction exhibit more of a “satiny finish distributed fairly evenly, depending on how a particular system is operated,” Macdonald said. “If you’re varying the speed all the time, then the discharges tend to be a bit more random than if you are running at one continuous speed.” The earliest bearing damage cannot be detected through vibration monitoring, he added.


Machines operating at slow speeds, where the balls have not risen onto an oil film but are in direct contact with the race, sustain little EDM damage from bearing currents. The voltage does not build to a level sufficient to melt the race surface. However, even low-voltage current passing between a baE and race in contact can cause rapid heating and lead to false Brinelling, Macdonald said.

An interesting aside to this phenomenon is that higher-quality bearings, with their smoother raceways and balls, exhibit fewer irregularities than their lower grade counterparts. Shaft voltage discharges less often with the better bearings. Consequently, voltage builds up to a higher level than it would in rougher bearings. In high-quality bearings, shaft voltage can be charging as much as 80 percent of the time, Macdonald said. Lower-quality bearings charge as little as 5 percent of the time. A high-quality bearing will see fewer, yet stronger, discharges than a low-grade bearing, and, as a result, will sustain deeper damage.

Mitigation Techniques

Annette von Jouanne, a professor of electrical engineering at Oregon State University in Corvallis, recently compiled a list of mitigation techniques for dealing with PWM-generated bearing currents. Along with shaft grounding, she lists bearing insulation and the use of ceramic elements or conductive grease as possible remedies. A Faraday shield, she said, is another method of attacking the problem of bearing currents; so, too, is the use of a new dual bridge inverter designed to eliminate the common mode voltage.

Von Jouanne said that a shaft grounding brush can provide a low-impedance path from the rotor shaft to the motor frame and so eliminate any shaft voltage. A brush wears, however, so this method calls for maintenance and, eventually, replacement of the brush.

Bearing insulation is a second approach. Insulating the outboard motor bearing, von Jouanne said, interrupts the path of circulating bearing currents (which Lipo and Chen said are generated by induction). A second insulating layer, on the inboard bearing, can prevent the noncirculating currents from flowing through the bearings. The voltage on the shaft remains, however, to seek another path to ground. The path it finds can lead through the bearings of the driven equipment or through the bearings of a shaft-mounted tachometer. Von Jouanne said the motor buyer can request insulated bearings from a manufacturer, although she thought the cost of this might be prohibitive for motors under 200 horsepower.

Ceramic bearings are yet a third way to address the problem of bearing currents. The nonconductive balls block the electrical path from shaft to frame. As with insulated bearings, however, shaft voltage remains to seek another route to ground. Another detriment, said von Jouanne, is that ceramic bearings add cost and manufacturing time to a motor order.

A suitably conductive grease could provide another low-impedance path for shaft voltage, she said. Unfortunately, any grease that suspends enough metal bits to conduct voltage reliably has so far been found to put too many particles in the path of the bearing elements. That leads, mechanically this time, to the same result as bearing currents do, namely, premature wear.

An electrostatic, or Faraday, shield can arrest the formation of bearing currents. “The Faraday shield blocks electrostatic coupling of the stator and rotor,” von Jouanne said. In the form of a thin copper foil, tape, or paint that nests in the airgap between rotor and stator, Faraday shields have been shown experimentally to reduce shaft voltages by up to 98 percent, she reported.

Finally, a dual bridge inverter can prevent development of common mode voltage by balancing the excitation of the motor, von Jouanne said. The use of such an inverter requires that the motor be dual voltage, and that it be run at the lower voltage rating, she added.

ABB recently patented a motor winding designed to eliminate circulating bearing currents. According to Tapio Haring, vice president of technology at ABB Motors, the design divides the stator winding into an even number of equal parts per phase. The groups are then distributed uniformly between ac supply connections at both ends of the stator. This generates a high-frequency net current flowing equally, and in opposite directions, through the windings. “By dividing the windings into two branches, we have a better chance of balancing the high-frequency common mode currents and getting more symmetric flux distribution,” Haring explained.


The ABB design veers away from traditional induction motor construction through this split winding, said Haring. The windings of a conventional motor normally connect to power leads only on one side of the stator. Haring explained that the common-mode voltage induces a current to flow in the circuit formed by the motor winding, stray capacitances from the winding to the stator core and the frame, and the grounding lead back to the frequency converter. This current induces a voltage between the ends of the motor shaft due to the inductive coupling between the stator and the rotor.

The size of this induced voltage, Haring said, depends upon the amplitude of the common mode current and its rate of change. Low-frequency currents are blocked by capacitances within the motor, such as those between varnished windings and the motor frame. But high-frequency currents are not blocked. Instead they form common mode loops that originate at the inverter and follow a path of least impedance back to the dc bus by way of the inverter frame.

Over the last four years, Haring and fellow ABB employee Jarkko Iisakkala conducted a series of experiments on the nature of circulating bearing currents. “The experiments showed that the problems were caused by asymmetric distribution of high-frequency currents in the motor,” Haring said. “Once we established this, finding a solution was relatively straightforward.”

The solution pits one high-frequency net current against a current of equal magnitude flowing in the opposite direction. The currents, in effect, cancel each other, and the bearings roll on, unmolested.

The motor windings are connected so that the direction of the fundamental current through them is identical to that of a motor wound in the conventional way. The solution is especially applicable to NEMA motors produced for the North American market, Haring explained, because the windings already terminate at a central junction box. Other motors will need changes in design before they can be wound in the new style.

Haring said that the design eliminates a motor manufacturer’s need to insulate any bearings during construction. That should lessen expense while shortening the time it takes to fill an order. Although modified windings might add somewhat to the cost of building a motor, Haring said the expense would be substantially below what bearing insulation would add.

“This is a remedy only to the circulating current cases,” said Haring, “which, as a matter of fact, are the most serious cases in big machines. The other types we have to eliminate by using symmetrical cables and by making proper termination and grounding connections.”

ABB’s Kenyon said that European motors have had fewer shaft grounding problems than U.S. models, mainly because of a more elaborate effort put into maintaining the integrity of the path to earth. European grounding systems incorporate copper or aluminum shields that terminate in a full 360 degrees of contact, Kenyon said. The Europeans also use special gland plates in their grounding systems, he added.

Toshiba’s Macdonald said, “Keeping the carrier frequency low is a major step in mitigation of drive-related bearing problems and is a no-cost mitigation technique.” Special windings and grounding methods do not solve the common mode problem, “the largest aspect of bearing-related drive failures,” he said.

“Bearing current is not a rampant problem,” he said, “but it does occur. The problem is that most of the time the end user says, ‘Well, I got two years hfe out of that bearing.’ So, they change the bearing and carry on. They don’t do a bearing analysis to discover what the problem was.

“I’m not sure that, as an industry, we have a really firm grasp on how often this happens,” Macdonald said. “It may be happening a lot more than we think. But it’s not like you put a drive on a motor and three months later the bearings fail because of electric discharge. If things are done properly to help minimize the occurrence, you might go years before that bearing fails. It’s still a premature failure. But, a lot of times, it seems to be an acceptable enough life that people don’t even bother investigating what the problem was.”

Lipo said, “Bearing current is a mechanical problem, but it is caused by an electrical phenomenon. That’s where things got complicated. That’s why it went undetected for so many years. Converters aren’t brand-new. People have been wondering, ‘What’s happening to my machine? I keep burning up my bearings.’ Finally, we’ve been able to figure out what’s going on.”