Selection Guide · 10 min read · Updated 2026-10-12
VFD and Servo Cable Selection: Why Screened Cable Matters More Than You Think
The cable between a drive and a motor is not carrying a sine wave. It is carrying a train of fast-edged pulses, and that changes what the cable has to do: contain the common-mode current, survive the reflected-wave voltage, and give the high-frequency current a return path close to the conductors that produce it. This article explains the three mechanisms, the arithmetic behind the critical cable length, and the earthing detail that decides whether a screen works at all.
The drive output is not a sine wave
A voltage source inverter rectifies the supply into a DC link, then chops that DC into a pulse-width modulated train. On a 400 V drive the DC link sits at about 540 V, and the insulated gate bipolar transistors switch it with a rise time of roughly 0.1 to 0.2 µs. Divide the voltage by the rise time and you have the number that matters: 540 V in 0.15 µs is a dv/dt of about 3.6 kV/µs, and fast devices exceed 5 kV/µs.
Three consequences follow, and they are the reason this cable is specified differently from a motor feeder. First, the fast edge behaves as a travelling wave and reflects at the motor. Second, the three phases switching together create a common-mode voltage that drives current through stray capacitance. Third, the cable radiates, because at the frequencies contained in a 0.15 µs edge a 20 m cable is a significant fraction of a wavelength and makes an effective antenna. IEC 61800-3 is the standard that sets the emission and immunity requirements for the drive and its installation.
- DC link voltage on a 400 V drive: about 540 V; on a 690 V drive: about 930 V
- IGBT rise time: typically 0.1 to 0.2 µs, giving dv/dt of 3 kV/µs and above
- Switching frequency: commonly 2 to 16 kHz, which sets how often the edge occurs
- The edge, not the fundamental, is what the cable has to cope with
dv/dt and the reflected wave: when cable length matters
A pulse travels along the cable at about half the speed of light, so roughly 150 m/µs in an XLPE insulated cable. If the cable is long enough for the pulse to reach the motor and come back while the edge is still rising, the reflection arrives while the incident wave is still there, and the two add. The motor has a much higher surge impedance than the cable, so the reflection coefficient is close to +1 and the voltage at the motor terminals can approach twice the DC link voltage.
The critical length is L = v × t_r / 2. With v = 150 m/µs and a rise time of 0.1 µs, L = 7.5 m; with the more forgiving 0.5 µs edges of an older or slowed device, L = 37.5 m. This is why modern drives with fast IGBTs are more sensitive to cable length than the ones they replaced, and why the maximum motor cable length is a number printed in every drive manual. On a 400 V drive the doubling gives about 1080 V at the motor terminals, which a 0.6/1 kV cable handles; on a 690 V drive the same mechanism gives about 1860 V, which is why the cable for a 690 V drive on a long run is specified at a higher voltage rating.
| Motor cable length | What happens | Usual measure |
|---|---|---|
| Up to about 30 m | Reflections add little; peak voltage stays within cable rating | Standard 0.6/1 kV motor cable, screened |
| 30 m to 100 m | Voltage doubling becomes possible with fast devices | Check the drive manual; a dv/dt output reactor is common |
| 100 m to 300 m | Reflected wave overvoltage and bearing current risk are both significant | dv/dt reactor or sine-wave filter, VFD rated cable |
| Above 300 m | Cable capacitance and charging current become the limit | Sine-wave filter, and a separate review of the drive rating |
- Peak voltage at the motor can approach 2 × the DC link voltage
- Reflections are worse with fast devices, not with high power
- The drive manufacturer's stated maximum cable length is the requirement to follow
- A sine-wave filter also reduces the audible noise and the cable-borne emission
Common-mode current and the path it takes back
Because the three output phases switch between the two DC rails, the sum of the three phase voltages is not zero at any instant: there is a common-mode voltage with respect to earth. That voltage drives current through the only capacitance available, which is the cable capacitance to its screen and to earth, plus the capacitance inside the motor between the windings and the frame.
Do the arithmetic for a 50 m run. Take a cable capacitance to the screen of 150 pF/m, which is a typical value for a screened power cable. Over 50 m that is 7.5 nF. To displace 7.5 nF at 5 kV/µs needs I = C × dv/dt, which is 7.5 × 10⁻⁹ × 5 × 10⁹ = 37 A. That is an instantaneous current flowing only for the duration of the edge, but it repeats at the switching frequency several thousand times a second, and it has to flow somewhere. Whatever path it finds is the path it takes: through the screen if the screen is there and properly earthed, through the parallel earth conductor, through neighbouring cable, through the motor bearings, or through the building steel.
The bearing path is the one that costs money. Common-mode current passing through the motor bearings discharges through the lubricant film in small sparks, and over time it produces the characteristic fluting pattern on the raceways and eventually bearing failure. It is one of the better-documented causes of premature motor failure on inverter duty, and it is a cable problem as much as a motor problem, because a shorter, better-screened cable reduces the current that reaches the bearings.
Braid, foil, or both: coverage and transfer impedance
A screen is judged by its transfer impedance, measured in milliohms per metre. It is the voltage that appears on the inner surface of the screen per unit of current flowing on it, so it is the coupling between the interference current on the screen and the signal it induces inside. Lower is better, and the two common screen types are good at opposite ends of the frequency range.
A copper braid has real copper in it, so it works well at power frequency and can carry fault current, and it is mechanically tough and flexible. At high frequency its transfer impedance rises, because the braid strands spiral and the resulting inductance appears in series. A foil screen, an aluminium or copper layer on a polyester carrier, has effectively complete coverage and a low transfer impedance at high frequency, but its cross-section is tiny, so it is poor at low frequency and it cannot carry fault current at all. The combination of a foil and a braid is what many VFD and servo cables use, and it is the safe default when the installation has not been analysed.
| Screen type | Optical coverage | Transfer impedance behaviour | What it is best at |
|---|---|---|---|
| Copper wire braid | 70 to 90 % | Low at low frequency, rising above about 1 MHz | Power frequency, fault current, flexibility, mechanical strength |
| Aluminium or copper foil with a drain wire | Close to 100 % | Low at high frequency | High frequency screening where no fault current flows |
| Foil plus braid combined | Close to 100 % plus a conductive path | Low across the whole range | VFD and servo outputs, and mixed installations |
| Copper tape lapped | Close to 100 % | Low at high frequency | Fixed installations; limited flex life |
| Unscreened with a separate earth conductor | None | No screening | Short runs where the drive manual permits it |
Earthing the screen: both ends, 360 degrees, no pigtails
The screen of a VFD cable is the return path for the common-mode current, and for it to work it has to be bonded at both ends, at the drive end to the drive's earthing plate or screen clamp and at the motor end to the motor terminal box with a conductive gland or clamp. At the frequencies involved the connection has to be a circumferential one: a 360 degree clamp that contacts the screen all the way round.
A pigtail does not do this, and the reason is arithmetic. A length of wire has about 1 µH of inductance per metre, so a 100 mm pigtail is about 0.1 µH. Its reactance at 10 MHz is 2π × 10⁷ × 10⁻⁷, which is roughly 6 Ω. That 6 Ω sits in series with a bond that is supposed to be in the milliohm range, at exactly the frequencies the screen exists to shunt. The practical rule is that a pigtail should not be longer than a few tens of millimetres, and the better answer is not to have one: use a screened cable gland, a clamp plate in the drive, and a conductive EMC gland at the motor.
- Bond the screen at both ends, at the drive and at the motor
- Use a 360 degree contact: EMC gland, clamp plate or a metal saddle on the whole circumference
- Keep any drain wire pigtail under about 30 mm, and shorter if possible
- Never leave the screen floating at one end: an unterminated screen does not shunt anything
- Bond the motor frame to the drive's earth, and keep the earth conductor in the same cable as the phases
Separation and routing
The other half of EMC is keeping the noisy cable away from the sensitive ones, and it is cheaper to do this at the design stage than to fix later. A drive output cable lying alongside an analogue signal cable couples into it, and the result is a control loop that misbehaves in ways that look like sensor faults. The values applied in practice, and required by IEC 61800-3 for a compliant installation, separate the two categories physically.
| Situation | Minimum separation applied in practice | Note |
|---|---|---|
| Parallel run, no divider | 300 mm | The common default in industrial specifications |
| Parallel run with an earthed metal divider or tray | 100 mm | The divider must be bonded to earth at intervals |
| Parallel run in the same steel tray with a partition | 150 mm | Partition height should be at least the cable height |
| Signal and power crossing | Cross at 90 degrees | Crossing perpendicular minimises the coupled length |
| Signal cable in its own earthed steel conduit | No separation needed | Keep the conduit earthed at both ends |
| Fieldbus, encoder and 4-20 mA cables | Route on a separate tray | Encoders on servo drives need screened and twisted-pair cable |
Sizing the motor cable
The motor cable carries the motor current plus the harmonic content of the waveform, and it is enclosed in a screen, which slightly reduces its ability to lose heat. Neither effect is large, but together they are the reason many drives suppliers recommend taking the next size up on long runs. The input side is a separate calculation: the input current contains harmonics and, on a four-wire supply to a six-pulse drive, the neutral can carry a substantial third harmonic, so a neutral sized on the assumption of a balanced sinusoidal load will be too small.
Two points are worth stating because they are often missed. The screen or armour should not be the only protective conductor on a drive output, because the high-frequency current it carries has nothing to do with the fault current it may also have to carry. And the cable capacitance itself is a design input on long runs: the drive has a maximum load capacitance, and a very long, heavily screened cable can exceed it and cause the drive to trip on earth fault or overcurrent at startup.
- Size the motor cable for the motor nameplate current, and consider the next size up on long runs
- Use a symmetrical three-phase plus three-earth construction so the return currents are distributed evenly
- Do not rely on the screen or armour as the sole protective conductor
- Check the drive's maximum permitted cable capacitance on runs over about 100 m
- Use a cable with a 0.6/1 kV rating for 400 V drives, and a higher voltage rating for 690 V drives on long runs
What to put in the RFQ
The details below turn a generic power cable into a drive cable, and they are the difference between a quotation that can be compared and one that cannot.
The list is short because each item is a place where two offers can differ honestly. A supplier who receives it can quote a drive cable; one who receives only a size and a length has to guess, and the guess usually shows up later as emission or tripping.
- Construction: symmetrical 3 + 3, or 3 + 1 if the drive supplier accepts it
- Screen: foil plus braid, or braid alone, with the optical coverage stated as a percentage
- Conductor: Class 5 flexible copper for short runs and moving parts, Class 2 for fixed long runs
- Voltage rating: 0.6/1 kV, or higher for 690 V drives on long runs
- Insulation: XLPE at 90 °C, or an electron-beam cross-linked compound for higher thermal margin
- Sheath: PVC for general use, LSZH where the route passes through occupied areas
- Drive data: switching frequency if adjustable, motor cable length, and any output reactor or filter in the circuit
- Documents: transfer impedance data if the installation has an EMC requirement, and the capacitance per metre for long runs
Frequently asked questions
Do I need screened cable between the drive and the motor?
For almost every installation, yes. The screen gives the common-mode current a return path close to the conductors that produced it, which reduces both radiated emission and the current that reaches the motor bearings. A few short runs with a filter at the drive and no sensitive equipment nearby can be installed unscreened, but the drive manual will say so and it is the exception rather than the rule.
How long can the cable between a VFD and a motor be?
Different manufacturers publish different limits, and that number is the one to follow. The physics behind it is the reflected wave: the critical length is half the rise time multiplied by the propagation velocity, about 150 m/µs, so a 0.1 µs edge gives 7.5 m. Typical published limits are 30 m to 100 m without an output reactor, with a dv/dt reactor or sine filter beyond that.
Should the VFD cable screen be earthed at one end or both?
Both ends, for a drive output. A single-ended bond works for low-frequency electrostatic shielding of a signal cable, but a drive output screen carries high-frequency common-mode current, and an unterminated end means that current finds another route. Bond at 360 degrees at the drive and at the motor, and do not use a long drain wire pigtail.
Why is my VFD cable bigger than the motor current needs?
Two reasons. The waveform contains harmonics that add heating without adding useful torque, and the screen reduces the cable's ability to shed heat. Many drive suppliers recommend the next size up on long runs. Also check the drive's maximum load capacitance: a very long screened cable can exceed it and cause tripping at startup.
What is the 3+3 construction for VFD cable?
Three phase conductors plus three protective conductors of equal size, distributed in the spaces between the phases instead of one earth in one position. It makes the cable geometrically symmetrical, so the high-frequency return currents are distributed evenly rather than concentrated, and it lowers both the inductance of the path and the common-mode impedance.
Need this worked out for a specific installation?
Send load current, route length, installation method, ambient temperature and the applicable standard — we will return a sized construction with the volt-drop and derating arithmetic shown.