You bought VFDs to improve operations and processes while saving energy. Yet, you're seeing many costs going up and motor failures doubling. What's happening? It's almost certain you have a mismatch between your drives and motors. However, such a mismatch doesn't have to be.

Our motors lasted for years when we had gear drive motors operating from line power. We put in a variable speed drive, and now we have failures." Despite coincidences, the drive is not the culprit. Drives don't make motors fail; not matching the drive and the motor does. Simply put, there is no reason for you to fear embracing modern drive technology.

Costs of variable frequency drives (VFDs) continue to drop as their advantages go up. As the technology evolves, these drives are becoming smaller in size. They are also employing faster switching, which reduces the switching losses in solid-state devices called Insulated Gate Bipolar Transistors or IGBTs. With fewer losses, efficiency also goes up.

This evolution of new drives also involves the development of switching algorithms that pulse width modulate (PWM) the voltage to the motor in such a way as to cause the motor's current to be nearly sinusoidal. This results in fewer losses in the motor, especially at lower speeds. This particular development vastly improves the losses and reduces harmonics that were higher in the older type of drives.

This positive set of circumstances has made VFDs much more of a "must have" item for new installations of standard AC induction motors. Parallel to this increase in demand, there's a skyrocketing demand for VFDs in existing installations. Herein lies the beginning of a new set of problems: motor/drive incompatibility. This incompatibility can defeat the purpose of using a VFD in the first place.

We can segment these problems into two areas: winding stresses and bearing failures. Winding stresses can occur if the motor doesn't have sufficient insulation to accept the high-voltage peaks that result from the drive's pulses interacting with the motor and motor/drive leads. Let's look at the simple equivalent circuit shown in Fig. 1. (in original article). Here, the leads act as inductors, and the motor acts as a capacitor.

A rectangular pulse from the VFD starts a current in the inductor (leads). The motor capacitance then charges to a voltage equal to the peak of the rectangular pulse. When this happens, the inductor has no voltage across it-but still has current flowing in it. The energy stored in the lead's inductance will force the current to continue. The capacitor voltage will then increase until the current stops.

Once the current stops, the capacitor (motor) voltage is above the peak of the VFD's pulse. The voltage across the leads reverses, and current flows (from the motor) backinto the VFD. This "re-energizes" the leads. Consequently, the capacitor voltage goes below the peak of the rectangular pulse until it stops; current now flows forward in the leads again! This cycle will continue until the resistance of the leads finally dissipates the energy originally stored in the leads' inductance. The ringing can go on for many cycles at a very high frequency, often at 0.5 to 4 MHz.

What drives this overshoot and ringing? The driver is the energy stored in the leads during the initial build up. Longer leads have higher inductance. Their effect is to increase the time required to charge the capacitance of the motor. This results in more energy stored in the leads and more overshoot. Generally, longer leads will cause more overshoot (up to some maximum value). See the Table (in original article) for typical test data. Fig. 2 (in original article)gives the relationship of overshoot versus time for various lengths of leads. It doesn't take much lead length to cause problems. Even 10 ft of lead can create a 43% overshoot.

Modern VFDs have rise times of .025 to 1 microsecond. This quick transition has two effects. It allows the leads to store energy, generating the overshoot. It also causes the voltage within the motor to distribute unevenly.

A faster rise time creates more stored energy and more overshoot. By design, newer VFDs have faster rise times. Substantial overshoots now occur on short lead lengths that were not a problem in the past. See Fig. 3 (in original article, page 38)for the relationship of overshoot voltage versus lead length for slow and fast rise times.

The rise time at the motor also affects the distribution of voltage within the motor. The first coil in the motor acts as a filter for the rest of the winding, and it will have more voltage across it than the rest of the coils. The percentage of peak terminal voltage on the first coil will be greater if the rise time is faster.

A typical dual voltage 3-phase motor has two sets of coils to allow external series/parallel connection. Each set may have four or more coils. Such a motor running across the line would have an equal voltage across each set and nearly equal voltage across each coil.

This will not be the case when operating from a VFD. Experiments on motors built with taps after each coil have shown that the first coil in each phase has much more peak voltage across it than the rest. The first coil may have 10 times as much voltage as when operating across the line. It's remotely possible for the first turn to touch the last turn, which results in having the whole coil voltage between the two magnet wires. If the first coil has many turns, the voltage across the first turn itself is very low. Fig. 4 (in original article)shows the results of voltage distribution tests on a 1 hp motor.

It's important you know the peak voltage at the motor. This information allows you to select a motor that can handle this voltage. The peak voltage at the terminals of a motor under VFD operation depends on the input voltage to the VFD and the amount of overshoot at the motor. We can show mathematically that the overshoot will not exceed twice the peak of the rectangular input, no mater how fast the rise time or how big the inductance. Some rare situations can cause the multiple to exceed two.

All modern PWM VFDs simply rectify the mains' input, resulting in a DC bus equal to the square root of 2 times the input rms voltage. The sidebar "Determining Peak Voltage, " on page 76, shows how to find the size of the voltage spike (peak voltage) the motor will see.

NEMA Standard MG1, in section 31.40.4.2, states that motors rated less than or equal to 600V rms shall withstand 1600V peak. Actually, a 575V rms motor could see peaks above 1600V (and a 230V rms motor would see peaks much lower). The NEMA Standard is in the revision process now.

Failure Mechanism-corona. When two conductors have potential between them, they develop an electric field within, and outside, the insulation on the conductors (see Fig. 5 in original article). When the field strength in the air is high enough, the air breaks down. We call this breakdown corona. The discharge has several bad effects. First, it causes oxygen to recombine into O3 (ozone). Ozone is very reactive and quickly combines with chemicals in the insulation. The result: The additional oxygen in the insulation causes it to disintegrate. Second, corona forms NO2 , nitrogen oxide, with similar corrosive properties. Finally, corona accelerates the charged air particles, which hit the insulation with enough velocity to cause mechanical damage.

Corona activity has historical limitations. The corona will start when the voltage reaches a certain level, which is the corona inception voltage (CIV). The CIV depends on the spacing and type of insulation, temperature, surface features, and humidity. Once you pass the CIV threshold, you must see a substantial reduction in voltage before the discharges will stop. Sometimes during a hi-pot test, you can hear a buzzing sound or smell ozone. This indicates corona is occurring.

Corona has always been the limiting factor for the voltage you can reliably apply to a winding. By design, high-voltage motors eliminate corona during operation. Formed coils, layered insulation, semiconductive layers, mica tapes, etc. are part of the designer's methodology. The design of high-voltage transformers in TV and ignition systems render corona-free operation. You can easily measure the CIV of magnet wire by forming samples of twisted pairs of wire. The CIV will range from 500V to 1000V. Large diameter wire normally has a thicker film and higher CIV. For any system, the CIV drops as temperature goes up. Different materials (types of film, nylon, polyester, etc.) have different CIVs- even if their thickness is the same. Also, the CIV will be different on the same type and gauge of wire from different wire manufacturers. The manufacturer should obtain a specification from the wire vendor for the exact wire to use in a given motor. Fig 6, on page 39(in original article), shows the relationship of CIV versus gauge for two different manufacturers at two different temperatures.

The sidebar "Possible Solutions" gives an overview of different solutions manufacturers are employing. Whatever solution or combination of solutions you choose, remember-keep your lead length as short as possible, and account for your peak voltage.