INVT VFD: Why Does a 7.5 kW Motor Vibrate at 15-24 Hz?

A 7.5 kW four-pole motor spinning uncoupled on a 55 kW inverter, oscillating hard between 15 and 24 Hz and stable everywhere else, is not a mechanical fault and not a tuning miss. It is a control-loop mode that nothing in the motor data group can damp. The same motor on a 22 kW frame of the same family shows only trace vibration, which already points the finger at how the drive senses and regulates current rather than at the machine.

Fixes That Fail on This Symptom

The short list below is what gets tried first on a test bench, and why each one leaves the 15-24 Hz band untouched.

Attempted fix	What it actually changes	Why the oscillation survives
Raise carrier to 15 kHz	PWM switching period, audible noise, current ripple in the kHz range	The instability sits at 15-24 Hz. A 15 kHz carrier is three orders of magnitude above it and has no authority over an electromechanical mode.
Repeat the dynamic autotune	Stator/rotor resistance, leakage and magnetizing inductance stored in the motor group	In SVPWM mode those identified values feed only the V/f curve and slip compensation. There is no current regulator to use them.
Re-enter nameplate data	Rated power, current, voltage, frequency, speed	Necessary, but a correct model driving an open-loop voltage output still has no damping term.
Balance the rotor, re-align the coupling	Mechanical unbalance forces, which scale with speed squared	A mechanical resonance is a narrow peak fixed in rpm and present on any drive. This band is wide, load-dependent, and disappears with the 22 kW frame.
Program skip frequencies across 15-24 Hz	Forbids steady-state operation in the band	Hides the symptom. On a test bench that must characterize motors across the whole speed range, a 9 Hz hole is not acceptable.

Raising the carrier is the most expensive of these. On a 55 kW frame, 15 kHz drives switching losses up steeply, forces output-current derating, and increases dv/dt at the motor terminals along with common-mode bearing current. This is heat, not logic.

Current Sensing Resolution on an Oversized Frame

The number that matters is the ratio of frame rating to actual load current. A 55 kW inverter carries current transducers, burden scaling and an ADC input range sized for the full-scale current of that frame. The least significant bit of that channel is fixed by the scaling, whatever motor is connected.

Hang a 7.5 kW machine on it and two things shrink at once. The motor's rated current is a small fraction of the frame's full scale, and at no load the torque-producing component of that current is only friction and windage — a few percent of the motor's own rating. What the drive has to resolve is therefore a fraction of a percent of its measurement span, sitting on top of PWM ripple and offset drift.

Quantization noise and offset error at that level propagate straight into whatever regulator uses the current signal, and into slip estimation. The 22 kW frame on the same motor works at a power ratio near 2.9:1 instead of 7.3:1, which puts the same physical current roughly two and a half times higher up its measurement span. That difference alone explains why one drive is marginal and the other is unusable in the same band.

The 15-24 Hz Window

Assuming a 50 Hz base, 15-24 Hz is 30-48 % of base speed, or 450-720 rpm synchronous on a four-pole machine. That window is the classic open-loop instability region for a voltage-source inverter running V/f or SVPWM control into a lightly loaded induction motor.

Below base speed the back-EMF is small and the stator resistance drop is a large share of the applied voltage, so any error in the applied volts translates into a large flux error. The rotor, the inverter output and the DC-link capacitance form a lightly damped electromechanical loop, and an unloaded rotor contributes almost no mechanical damping — there is no load torque to absorb energy. Excess torque boost makes it worse: boost defaults on a 55 kW frame are scaled for a 55 kW motor, so a 7.5 kW machine gets over-fluxed at low frequency, saturates, and draws additional magnetizing current that the drive then reacts to. The result is a self-sustaining low-frequency oscillation in speed and current: the motor "jumps" rather than turns.

Vector control breaks the loop because it regulates current, not voltage. The flux and torque current components are held to a reference through a fast inner loop, which introduces the damping term that V/f lacks, and the speed observer filters the noisy feedback discussed above.

P00.00 Control Mode Selection

The parameter that decides all of this is P00.00, the speed control mode. Factory default is 2.

P00.00	Mode	Manual description	Motor required	Autotune required
0	Vector	Low frequency, high torque, high speed and torque accuracy; stated as more suitable for small-power applications than mode 1	Yes	Yes
1	Vector	High performance, high speed and torque accuracy	Yes	Yes
2	SVPWM	Where high control accuracy is not needed — fans, pumps; also the mode for multiple motors in parallel	No	No

In modes 0 and 1 the drive closes the loop on the identified motor model and continuously corrects to keep rotation right; the motor must be connected. In mode 2 the drive computes a voltage/frequency curve from the entered motor data and outputs it whether a motor is present or not. That indifference is exactly why the default suits a bench where motors get swapped — and exactly why it leaves the 15-24 Hz mode undamped.

Setting P00.00 = 1 on the 55 kW drive removes the oscillation and leaves only minor vibration at 18 Hz. There is a decision path here worth walking: the manual positions mode 0 as the better fit for small-power machines, so if residual vibration at any point in the sweep is still outside your acceptance band, test mode 0 with the same identified motor data before touching anything else. Both modes demand a valid autotune; neither will behave with frame default motor parameters left in place.

Reconfiguration Procedure

1. Uncouple the motor shaft or verify the load can free-wheel. Rotational autotune must not be performed against a locked or geared load.
2. Enter the motor nameplate values in the motor parameter group: rated power, rated voltage, rated current, rated frequency, rated speed, and pole count. On an oversized frame these default to the drive's own ratings and must be overwritten for every motor under test.
3. Run the dynamic (rotational) autotune and let it complete without a fault. Record the identified stator resistance, rotor resistance, leakage inductance and mutual inductance; save them per motor type so a bench swap becomes a parameter-set recall rather than a fresh tune.
4. Set P00.00 = 1. Confirm the drive accepts the change with the motor stopped — control mode is normally a stop-only parameter.
5. Return the carrier frequency to the drive's default for the 55 kW frame. Keep it low enough to avoid output derating; 15 kHz on this frame buys audible noise reduction at a real thermal cost and contributes nothing to low-frequency stability.
6. If the V/f group was modified during earlier attempts — torque boost, slip compensation gain, oscillation suppression — restore those to default. They are inactive in vector mode and will confuse the next technician who reads the parameter list.
7. Re-run the speed sweep before recording any acceptance data.

Speed Loop Trim for the Residual Point

A single low-amplitude point left at 18 Hz after switching to vector mode is a speed-regulator bandwidth artifact, not the original instability. Two levers address it, in this order.

First, the speed loop proportional and integral gains. Vector-mode ASR gains are segmented by frequency in INVT drives — a low-speed set, a high-speed set, and switching frequencies between them. An 18 Hz point sitting near a gain switchover boundary produces exactly this kind of localized ripple. Move the switchover frequencies so 18 Hz falls clearly inside one segment, then reduce proportional gain slightly in that segment and check whether the ripple shrinks or grows. Growth means the gain was too low, not too high.

Second, mechanical: at no load, a coupling half, a key, or a fan cowl that is loose will show a fixed-rpm response independent of drive settings. 18 Hz on a four-pole machine is 540 rpm synchronous. If the vibration follows rpm identically when you change base frequency or run the motor from the 22 kW drive, stop tuning and fix the mechanics.

Verification Sweep

Acceptance for a bench drive is a documented sweep, not a spot check.

1. At each step log output current, output voltage, output frequency and estimated torque from the drive's monitor group. A stable point shows current holding within a narrow band; the failure signature is current amplitude walking up and down over roughly one second at fixed frequency reference.
2. Clamp a current probe on one output phase and watch the envelope on a scope with a slow timebase. Modulation of the fundamental envelope at a few hertz is the oscillation; a clean flat envelope is the pass condition.
3. Repeat the sweep loaded, if the bench allows it. Load adds mechanical damping and normally widens the stable region — if a point that is clean at no load misbehaves under load, the speed loop gain is too high, which is the opposite defect.
4. Record the drive's fault history and heatsink temperature after the sweep. Confirm no overcurrent or overload entries were logged during the unstable-band transit and that the carrier setting is not pushing the module temperature toward its derating point.

Quantities to have on paper before starting:

Quantity	Value in this setup	Where to read it
Drive-to-motor power ratio	55 kW : 7.5 kW ≈ 7.3:1	Drive and motor nameplates
Reference ratio with acceptable vibration	22 kW : 7.5 kW ≈ 2.9:1	Same
Unstable band, no load	15-24 Hz (30-48 % of a 50 Hz base)	Speed reference during sweep
Synchronous speed, 4 poles	1500 rpm at 50 Hz; 450-720 rpm across the band	Nameplate pole count
Control mode default	2 (SVPWM)	P00.00
Control mode that stabilizes	1 (vector)	P00.00
Continuous output current derating at raised carrier	Frame-specific	Derating table in the drive manual
Identified motor model values	Per motor under test	Motor parameter group after autotune

Escalation Threshold

Stop adjusting parameters if the oscillation persists in both vector modes with a clean autotune and correct nameplate data, or if the drive logs overcurrent while sweeping the band. At that point the question is whether the 55 kW frame's current measurement chain can resolve the no-load current of the smallest motor in your test range, and that is a hardware limit that only INVT technical support can answer for a specific frame and firmware — bring them the autotune results, the parameter list, and the logged sweep. Where the bench must cover a 7:1 or wider power span, a second smaller drive is usually a cheaper answer than any amount of tuning.

FAQ