Cables, bearings and protection around the drive

The drive's output side has its own traps. The inverter does not deliver a sine wave but a train of steep-edged PWM pulses, and the motor cable behaves as a transmission line. This lesson is course sections 6.2.2 and 6.2.3.

Voltage reflection on the motor cable

When the pulse rise time is short compared with the cable's propagation delay, the impedance mismatch between cable and motor causes a reflection, and the reflected wave adds to the incident one. In the limit, the motor terminal voltage reaches twice the DC-link voltage: a drive on a 400 V supply, with a link at about 565 V, can put over 1 100 V across the first turns of the stator winding.

The reflected wave returns to the drive after a round trip of $2L/v_p$, where $v_p \approx 1.5 \times 10^8$ m/s is the propagation speed in a typical motor cable. Full reflection therefore occurs once

$$L \ge L_c = \frac{t_r\, v_p}{2} \tag{6.4}$$

so the critical length falls in direct proportion to the switching time. An older device with a 400 ns rise time tolerates 30 m of cable; a modern IGBT switching in 50 ns reaches full reflection after less than 4 m. Faster switching improves efficiency and reduces audible noise, and it makes this problem dramatically worse, which is why it appears on new installations that replace apparently identical older drives.

Switching rise time t_r

100 ns

About 400 ns for an older device, 50 to 100 ns for a modern IGBT.

Cable length L

20 m

Fit a dv/dt filter

Beyond the critical length: the reflection nearly doubles the pulse at the motor. Fit a filter or an inverter-duty motor.

Critical length L_c

7.5 m

Peak at the motor

1,096 V

Overvoltage

1.94 × V_dc

L_c = t_r v_p / 2 , v_p ≈ 1.5 × 10⁸ m/s ; V_dc ≈ 565 V (400 V supply)

Predict first

An old drive switching in 400 ns fed a general-purpose motor over 10 m of cable for years. It is replaced by a modern drive switching in 50 ns, with the same cable and motor. What does the winding now see?

The same peak: the supply and the cable have not changedA peak about 50 % higher, close to twice the DC linkA lower peak, because the new drive is more efficient

Bearing currents

A second, subtler effect comes from the same steep edges. The three PWM phase voltages do not add to zero at every instant, so a common-mode voltage appears between the motor's star point and earth. It drives a capacitive current through the parasitic capacitance of the bearing's lubricant film. When the film breaks down, the discharge pits the race and the balls, electrical discharge machining inside the bearing, which progresses to fluting and failure, typically within one to two years and usually misdiagnosed as a mechanical problem.

Countermeasures: an insulated non-drive-end bearing, a shaft earthing brush or ring, a symmetrical shielded cable with a low-impedance high-frequency earth path, or a common-mode choke. Larger machines and longer cables raise the risk.

Protection coordination

A drive does not look like a motor to the protection system, and protection designed for direct-on-line starting is wrong for a drive.

• The drive protects the motor. It limits its output current electronically, typically to 110 to 150 % of rating, within microseconds: the fastest and most selective protection for the motor and its cable. A conventional thermal overload relay between drive and motor is redundant, and worse, it measures a variable-frequency current it cannot interpret.
• Temperature, not only current. Motor thermal protection comes from the drive's internal thermal model, supplemented by direct winding temperature measurement (PTC thermistors or Pt100 sensors). That is the only valid method at low speed, where a self-ventilated motor loses most of its cooling while a model based on current alone cannot know it.
• Upstream discrimination. The supply-side device sees the rectifier's inrush when the DC link charges, large but short, and must ride through it without losing sensitivity to a real fault. Semiconductor fuses protect the rectifier bridge: ordinary breakers act far too slowly to save a diode or IGBT, and the fuse is chosen with an $I^2t$ below the device's withstand.
• Earth faults. The drive's common-mode currents and the cable's capacitance to earth produce a continuous leakage current at switching frequency, entirely normal and possibly tens or hundreds of milliamperes. A standard residual-current device trips on it. Where an RCD is required, use a type B device, responsive to smooth DC residual current, with a suitably raised threshold and the cable capacitance included in the calculation.

FoundationStart here if this is new to you

Flick one end of a long rope tied to a wall: the wave runs to the wall, bounces and comes back. If you flick fast enough, the returning bounce meets the next flick and the rope jumps twice as high. A motor cable is that rope, the motor is the wall, and a modern drive flicks very fast.

ExplorerGo deeper: derivations and open questions

Filter. A dv/dt filter slows the edge at the motor to about 2 µs. What critical length does that give? Check it in the widget with 50 m of cable, and explain the cost the filter adds in losses.

Retrofit. List what you would check before connecting a new VFD to a 20-year-old motor on its existing cable.