Stoklink Technical Articles

VFD Carrier Frequency and Switching Losses

What is carrier frequency on a VFD? Carrier frequency is the switching rate at which the inverter's IGBT bridge chops the DC bus into a pulse-width-modulated (PWM) output waveform, normally set between 2 kHz and 16 kHz on low-voltage drives per IEC 61800-2 rated data. Push it higher and motor current gets smoother, audible whine drops, but IGBT switching losses climb and the drive derates its continuous output current to hold junction temperature. This article covers what generates the loss, how the noise-versus-derating trade-off actually plays out, and the secondary effects on cable length limits and bearing currents.

What Sets Carrier Frequency on a VFD

Every PWM inverter has two frequencies running at once. The fundamental frequency is the sine-equivalent output that sets motor speed — 0 to 60 Hz or higher for a 50/60 Hz motor. The carrier frequency is how fast the IGBTs turn on and off to build that sine-equivalent waveform out of voltage pulses. A 4 kHz carrier means each IGBT switches roughly 4,000 times per second regardless of what the fundamental output frequency is doing.

Factory defaults on general-purpose drives sit low, usually 4 kHz on drives like the ABB ACS580, Siemens SINAMICS G120C and Schneider Altivar 340 (part of the variable frequency drives range Stoklink stocks). That default balances current rating against noise for the majority of pump and fan installs. The parameter is user-adjustable, usually in 1 kHz steps, and some drives auto-derate once you move past the base value.

Carrier frequency is the switching rate of the IGBT bridge that synthesizes the PWM output waveform, typically 2-16 kHz on low-voltage drives (per IEC 61800-2 rated data).

Why Every Switching Event Costs Energy

An IGBT is not a perfect switch. Each time it turns on or off, it passes through a brief region of simultaneous voltage and current, and that overlap dissipates energy as heat. Manufacturers publish this as Eon and Eoff per pulse, measured at rated bus voltage and collector current. Multiply that per-pulse energy by how often it happens per second — the carrier frequency — and you get average switching power.

Formula: IGBT Switching Loss per Device — Source: IGBT manufacturer application note (thermal design method referenced in IEC 61800-2)

Psw = fc × (Eon + Eoff)

Symbol Description Unit
Psw Average switching power dissipated per IGBT W
fc Carrier (switching) frequency Hz
Eon Turn-on energy loss per pulse, from datasheet at rated Vdc and Ic J
Eoff Turn-off energy loss per pulse, from datasheet at rated Vdc and Ic J

This is per device. A three-phase inverter has six switching positions, so total inverter switching loss is roughly six times this figure, on top of conduction loss that does not scale with carrier frequency at all. Double the carrier frequency and you roughly double the switching-loss term — conduction loss stays put. That is the whole mechanism behind derating tables: nothing exotic, just more heat pulses per second in the same heatsink.

Key takeaway: Switching loss is not fixed — it scales close to linearly with carrier frequency, so a drive rated for full current at 4 kHz may need meaningful current derating at 12-16 kHz. Check the manufacturer's derating curve before raising the setting, not after.

Carrier Frequency vs Audible Motor Noise

Motor noise from PWM comes from magnetostriction in the stator laminations, excited at the carrier frequency and its multiples. At 2-4 kHz that tone sits squarely in the most sensitive part of human hearing — the classic VFD whine. Push the carrier to 8 kHz or above and the fundamental tone moves toward the edge of typical hearing sensitivity; many people stop noticing it, even though the drive is switching twice as often.

What we see in the field: office-adjacent HVAC installs and hospital corridors almost always get bumped to 8-12 kHz for this reason, accepting the current derating because the load is rarely near nameplate anyway. A conveyor line running at full load in a noisy plant gets left at 4 kHz — nobody's listening for the whine, and the amps matter more.

Current Derating Above the Rated Carrier Frequency

Derating tables published alongside a drive's normal duty (ND) and heavy duty (HD) ratings assume a base carrier frequency, usually the factory default. Move above it and rated output current drops — sometimes gradually, sometimes in a step at specific thresholds — because the same heatsink and fan now has more switching-loss watts to remove for the same conduction-loss watts. See the normal duty vs heavy duty overload ratings for how ND/HD figures are built from a base frame rating; carrier frequency sits underneath both.

Ambient temperature compounds this. A drive already derated for 45°C cabinet ambient and then pushed to a high carrier frequency stacks two derating factors on the same frame size. This is the derating error we see most often: someone raises carrier frequency to quiet a motor, doesn't re-check the current table, and the drive trips on overload at a load current it handled fine the week before.

Key takeaway: Carrier frequency is a commissioning-time decision, not a run-time tuning knob. Raising it later to answer a noise complaint can silently push a drive into thermal derating if load current is already close to nameplate.

Downstream Effects: Bearing Currents and Cable Length

Higher carrier frequency means more dV/dt transitions per second at the motor terminals, and each transition is a candidate for the two problems covered in motor bearing currents and shaft grounding and dV/dt and sine-wave output filters: reflected-wave voltage spikes on long motor cables, and common-mode discharge currents that pit bearing races. Neither problem is caused by carrier frequency alone — cable length and grounding practice matter more — but a higher switching rate does raise the frequency of the events that trigger both.

This depends on motor cable length: under roughly 15 m the reflected-wave risk is minor at any carrier frequency; past 50 m on an unfiltered installation it becomes a real concern independent of the carrier setting. Carrier frequency changes the noise character, not the underlying insulation stress mechanism.

Key takeaway: If bearing currents already show up on a drive with long motor cables, a shaft grounding ring or insulated bearing fixes the failure mode directly — lowering carrier frequency only reduces switching count, it does not eliminate the mechanism.

Choosing a Carrier Frequency in Practice

For a scalar V/f pump or fan on the ND rating, leave carrier frequency at the factory default. There is no efficiency or noise argument strong enough to justify giving up rated current on a load that runs at partial speed most of the time anyway — see the affinity-law math in VFD energy saving on pumps and fans for why partial-speed running already dominates the savings case.

For vector or DTC control on machinery with fast current-loop bandwidth requirements, carrier frequency interacts with control performance too — a modulator that is not switching often enough limits how tight the current loop can be tuned. See V/f, vector and DTC control methods for how the control method itself constrains the usable carrier range. Some integrators default to the highest carrier frequency the frame supports for a quieter, smoother machine, but that only holds up if the load current has headroom against the derated table — verify it, don't assume it.

Key takeaway: There is no carrier frequency setting that is simultaneously quietest, most efficient, and highest-current-rated. Pick low (2-4 kHz) when output current headroom matters, high (8-16 kHz) when noise or current smoothness matters, and size the frame for whichever you choose.
Switching loss is the energy dissipated in an IGBT during each turn-on and turn-off transition, summed at the carrier rate to give average switching power, per the thermal design method IEC 61800-2 references from IGBT manufacturer application notes.

Background reading on the power stage that does this switching is in the VFD engineering guide, and on the PWM mechanism itself in how PWM and V/f control work.

Frequently Asked Questions

What is a typical VFD carrier frequency range?

Most low-voltage drives switch between 2 kHz and 16 kHz, with 4 kHz the common factory default on general-purpose drives such as the ABB ACS580, Schneider Altivar 340 and Siemens SINAMICS G120C. Machinery drives with fast current loops often run 8-16 kHz.

Does raising carrier frequency reduce motor efficiency?

Not on the motor side — it can actually reduce harmonic heating in the windings. It reduces drive-side efficiency, because the IGBTs dissipate more switching energy per second, which is why most manufacturers apply automatic current derating above the factory default.

Why does my drive derate at high switching frequency?

Each IGBT turn-on and turn-off event dissipates a fixed energy loss. At a higher carrier frequency that energy is dissipated more often per second, raising junction temperature for the same load current. To hold the same junction temperature limit, the drive's continuous output current rating is reduced — check the manufacturer's derating curve for the exact figure at your ambient temperature.

Can I set carrier frequency to any value I want?

Only within the drive's supported range, typically adjustable in 1 kHz steps from about 2 kHz to 16 kHz on general-purpose drives. Some parameter menus include automatic derating or automatic carrier-reduction logic that overrides a manual setting if the drive's thermal model predicts an overtemperature trip.

Does carrier frequency affect bearing currents?

Yes, indirectly. Higher carrier frequency increases the number of common-mode voltage transitions per second, which raises the frequency of discharge-type bearing current events. The fix is a shaft grounding ring, insulated bearings, or symmetrical shielded motor cable — not lowering carrier frequency, which only reduces event count without removing the mechanism.

Conclusion

Carrier frequency is a single parameter with three linked consequences: audible noise, switching loss, and current derating. There is no setting that wins on all three. Start from the factory default, change it only for a specific noise or control-bandwidth reason, and re-check the manufacturer's derating table every time you do — the amp headroom you assume you have may not be there once the carrier frequency moves.

Comments (0)

    Leave a comment