Switching VFD Output Contactors on a Running Multi-Motor Drive

Application Overview: Eight-Motor Yarn Winding Machine

The reference machine is a yarn winding station built around eight (8) single-direction 0.25 kW motors. Originally each spindle was driven by an individual motor started direct-on-line (DOL) from a bank of electromechanical relays. A retrofit is now required: the same eight motors must continue to share a common speed reference (same RPM, same frequency, same acceleration ramp), but the start/stop authority for each spindle must move to a PLC driven by eight independent yarn-break sensors. A single variable-frequency drive (VFD) sitting upstream of eight output contactors is the lowest-cost mechanical topology, but it introduces a switching problem that does not exist with a pure relay panel: opening an output contactor while the VFD is actively producing PWM voltage destroys the drive.

Two facts from the original field experience must be carried into the design:

1. The motors were later re-identified as 3-phase 0.25 kW induction machines (not single phase). The VFD therefore runs all eight stator windings from a 3-phase output bus.
2. A similar machine has been running for years on a Yaskawa VFD with output contactors that open and close under load, with no recorded drive trips. The drive has been oversized for the application so that it never "notices" the inrush of the largest motor it must pick up next.

Why a Single VFD for Parallel Motors

Three manufacturer documents frame the baseline rules for multi-motor drives.

ABB Technical Note 204 – Multiple Motors Controlled by a Single Drive makes the central statement: "Using one drive for multiple motors means the motors connected will all operate at the same speed. System control schemes should not actively disconnect a motor from the drive output while the drive is running." ABB's guidance is conservative and assumes the output contactor is opened while the drive is producing PWM. The note further requires that every motor in the parallel group carry its own thermal overload protection because the drive's single electronic thermal model cannot discriminate between eight parallel branches.

Rockwell Automation – Control Multiple Motors with Just 1 VFD lists the four conditions that must be true for the topology to be safe: (1) all motors must be the same NEMA design, (2) cable lengths from the drive to each motor must be similar, (3) each branch must have individual overload and short-circuit protection sized to the motor, and (4) the drive must be rated for the sum of the motor full-load currents plus the inrush of the largest motor being re-started.

Yaskawa PR.AC.02 – VFDs Can Control Multiple Motors explicitly addresses the parallel topology and lists the conditions under which motors can be connected and disconnected from a running VFD output without damage. The Yaskawa document is the most permissive of the three: it accepts output contactor switching provided the drive is large enough that the loss or addition of one motor stays within the drive's continuous current rating and within its DC bus regulation window, and provided the contactor never interrupts the motor current at the moment the drive's IGBTs are commutating that phase.

All three documents converge on the same set of sizing rules:

The Core Engineering Problem: Output Contactor Switching Under Load

When a contactor on the output of a live PWM VFD opens while the motor is drawing magnetising current, two destructive events happen simultaneously.

1. Voltage transient on the motor side. The motor stator is an inductance. The drive's DC bus capacitors and the dV/dt filter (if any) hold the line at the drive's PWM pole voltage. When the contactor opens, the magnetising current is forced to zero in a few hundred microseconds. di/dt on the leakage inductance generates a voltage spike V = L · di/dt that easily reaches 1.5–2.5× the DC bus voltage. A 400 V drive can produce a 1,500 V spike on the contactor load side. This spike stresses the contactor, the cable insulation and the IGBTs that are still in the circuit on the line side.
2. DC bus over-voltage or shoot-through. If the IGBTs of the drive happen to be in the commutation dead-time when the contactor opens, the inductive kick can charge the DC bus through the free-wheeling diodes. The drive either trips on DC bus over-voltage (OV) or, if the spike exceeds the diode reverse recovery, destroys the IGBT module.

Failure Mechanisms You Must Design Out

Solution 1: Independent VFD per Motor

The textbook answer, and the safest one, is to give each of the eight spindles its own VFD. The eight drives all receive the same frequency reference from the PLC (a single 0–10 V or Modbus broadcast), so the spindle speeds remain locked.

Topology. One Yaskawa GA500 or V1000 micro-drive per motor, rated ~0.4 kW (one frame size above 0.25 kW to absorb the DOL inrush of the 0.25 kW motor on a running system). The PLC writes the frequency reference and the run command to each drive over Modbus RTU or EtherNet/IP. The thread sensor input goes directly to a digital input on the drive, and the drive is configured for a freewheel stop on that input. No output contactors are needed.

Cost. Eight 0.4 kW microdrives at the 2024 industrial list price is roughly 5× the cost of one 2.0 kW drive plus eight contactors, but the engineering cost is the lowest because every motor is independently protected and the contactor switching problem disappears.

When to choose it. New installations, plants that have cabinet space, or applications where the cost of a single spindle going down (yarn break) is high enough to justify a per-spindle investment.

Solution 2: Common DC Bus with Synchronization Relay

The second architecture keeps one VFD as the active front-end and adds a synchronisation relay between the drive output and the motor contactor. The relay enforces a strict sequence: the drive must be at 0 Hz and IGBT-blocked before the contactor is allowed to open, and the contactor must be closed and the cable pre-charged before the drive resumes PWM. This is the architecture that the Yaskawa PR.AC.02 application note actually uses in its multi-motor example, and it is what the contributors in the field report were describing as the "safety relay."

Sequence for a stop (yarn break or operator request):

1. PLC receives the stop request (sensor or HMI button).
2. PLC writes the Baseblock Enable (Yaskawa parameter b1-03 = 0, run command source cleared) or sends a controlled ramp-to-zero (C1-02) for the specific drive if multiple drives are present; in the single-VFD case the drive is set to > 30 Hz, then the PLC simply turns off that motor's enable bit.
3. PLC waits for the drive to report “Zero Speed” (Yaskawa digital output set in H2-01 to value 21) AND for the DC bus to settle (100 ms minimum).
4. Safety relay (Pilz PNOZ s5, ABB CM-MSS, or a hard-wired interlock using two contactors) is allowed to drop out, opening the output contactor to the motor.

Sequence for a start (fault reset or operator request):

1. PLC checks the contactor is open (auxiliary contact wired back to a PLC input).
2. PLC closes the output contactor. Drive is still at 0 Hz; IGBTs are blocked.
3. PLC waits one line cycle (20 ms) for the cable and motor capacitance to charge through the pre-charge resistors built into the synchronisation relay.
4. PLC pre-charge resistor is bypassed (contactor closes fully) and the drive is released from baseblock; frequency command is applied; motor accelerates to the running setpoint.

Critical detail: the synchronisation relay must be fail-safe – if the PLC drops out, all output contactors must drop out, and the drive must be baseblocked. A hard-wired interlock using two contactors in series (K1 = line contactor, K2 = drive output contactor, both with mirrored auxiliary contacts) is the simplest approval-friendly implementation.

Solution 3: Single VFD with Output Reactor and Yaskawa-Style Configuration

The third architecture is what the original field machine uses, and the one that will scale to the retrofit in the source problem because the cabinet and the wiring already exist. It is the same topology as Solution 2 but the synchronisation is performed by the drive's own ramp and baseblock logic and the output contactor is a standard IEC contactor without a dedicated safety relay.

Required hardware.

• One VFD, oversized to a minimum 2.2 kW (one frame above the 2.0 kW sum of eight motors). A Yaskawa V1000 (CIMR-VU2A0020FAA, 2.0 kW, 9.2 A) is the closest standard catalog number; the CIMR-VU4A0009FAA is the 400 V class equivalent at 9.0 A. The current headroom is what allows the drive to absorb the inrush of a motor being re-added without an OC trip.
• One VFD-rated output reactor, 5% impedance at the carrier frequency, rated 10 A continuous. For 8 motors the reactor sits between the VFD output and the contactor bus. Example: Yaskawa 3% line reactor (catalog prefix RM) or an ABB output reactor of the same impedance.
• Eight output contactors (IEC AC-3, 9 A, with at least one normally-open auxiliary contact for the interlock feedback). Schneider LC1D09, Eaton XTCE009, or ABB AF09.
• Eight thermal overload relays (range 1.0–1.6 A), one in series with each contactor. ABB TF42-1.6, Schneider LRD08, Eaton XTOB.
• Eight short-circuit protective devices, 3 A aM fuses or 4 A motor-protective circuit breakers (MPCB).

Yaskawa parameter setup.

VFD Sizing Calculation for the 8x 0.25 kW System

Start with the motor nameplate. A 0.25 kW 3-phase 400 V IE2 motor has a typical full-load current of 0.83 A and a locked-rotor current of 5×FLC, i.e. 4.15 A. For eight motors in parallel:

• Sum of FLC: I_total = 8 × 0.83 A = 6.6 A
• Sum of LRC: I_LRC = 8 × 4.15 A = 33.2 A
• Worst case re-start: 7 motors running at FLC + 1 motor starting from cold = 7 × 0.83 + 4.15 = 9.96 A peak

Use three-phase apparent power to cross-check:

S = sqrt(3) × V_LL × I / 1000

S = 1.732 × 400 × 9.96 / 1000 = 6.9 kVA peak

A 2.2 kW (9.6 kVA) drive is therefore correctly sized: the peak re-start current of 6.9 kVA is below the drive's 30-second overload rating of typically 150% (13.5 kVA) and well within the continuous 9.6 kVA. The drive will not notice the re-start.

If the VFD is sized at 1.5 kW (the same sum-of-FLC rating), the re-start inrush will trip OC on the first restart and the architecture will not work. The one-frame-headroom rule is non-negotiable.

PLC Integration: Thread Sensor, Contactor Logic, and Fault Reset

The PLC is the brain of the retrofit. It must (a) read eight thread sensors, (b) drive eight output contactors, (c) communicate with the VFD, and (d) implement a fault-reset cycle that brings a spindle back into the running system after the operator clears the yarn break.

I/O assignment (Yaskawa V1000 plus a generic compact PLC, 24 VDC logic).

Ladder logic (structured text equivalent) for one spindle.

// Inputs
bSensor1  := X0;    // yarn intact = TRUE
bAuxOK1   := X10;   // contactor closed feedback = TRUE
bZeroSpd  := X20;   // VFD at zero speed
bDriveOK  := X21;   // VFD not faulted
bOpReset  := X30;   // operator HMI reset button
bRunReq   := X31;   // operator HMI run command (master)

// Outputs
bCont1    := Y0;    // contactor 1 coil
bRunDrv   := Y10;   // VFD run command
bResetDrv := Y11;   // VFD fault reset pulse

// Internal state
bTripped1 : BOOL;   // latched trip flag for spindle 1

// Yarn break detection - latches trip on a single FALSE
IF bRunReq AND NOT bSensor1 AND NOT bTripped1 THEN
    bTripped1 := TRUE;          // latch
END_IF;

// Output contactor follows the inverse of the trip, gated on zero speed
// This is the synchronisation - drive must be at 0 Hz before contactor opens
IF (bRunReq AND bSensor1 AND NOT bTripped1 AND bDriveOK) THEN
    bCont1 := TRUE;             // close contactor, drive is running
ELSE
    IF bZeroSpd THEN            // wait for drive to stop
        bCont1 := FALSE;        // open contactor only when drive is at 0 Hz
    END_IF;
END_IF;

// Fault reset on operator command - single 100 ms pulse
IF bOpReset AND bTripped1 AND bSensor1 THEN
    bTripped1 := FALSE;         // clear the latch
    bResetDrv := TRUE;          // pulse VFD fault reset
END_IF;

IF bResetDrv THEN
    TON_Reset(IN:=TRUE, PT:=T#100ms);  // 100 ms one-shot
    IF TON_Reset.Q THEN
        bResetDrv := FALSE;
    END_IF;
END_IF;

// Master run command to the drive - held as long as the drive is needed
bRunDrv := bRunReq AND bDriveOK;

Why the zero-speed gate matters. If you simply do bCont1 := NOT bTripped1, the contactor will open the moment a yarn break is detected, while the drive is still producing PWM. The synchronisation block in line 14–19 is what protects the VFD. The drive is commanded to ramp to 0 Hz (Yaskawa decel ramp C1-02, here 2 s), the Zero Speed digital output asserts, and only then the contactor is released.

What the HMI shows. Eight status blocks, one per spindle: a green dot (sensor OK, contactor closed, motor running), a yellow dot (sensor OK, contactor open, motor waiting for reset), a red dot (yarn break latched, contactor open, motor stopped). A master "Reset All" button clears all eight latches in one operation after the operator has threaded all the broken yarns.

Commissioning Procedure

1. Pre-power check. Verify all eight motor nameplates match (V, Hz, kW, FLA, RPM, cos φ). Verify the contactor bus is rated for the drive's PWM output voltage (VFD-rated contactors only – standard AC-1 rated contactors are not approved for VFD outputs).
2. Drive-only test. Disconnect the contactor bus from the VFD. Run the drive into a single 0.25 kW motor directly. Verify direction, current, and the 8 kHz carrier setting does not produce audible nuisance from the output reactor.
3. Synchronisation test. Connect a single contactor and overload. Command a stop from the PLC. Verify the contactor only opens after the Zero Speed digital output has been TRUE for at least 100 ms. Use a scope on the contactor coil and the VFD output terminal to confirm the PWM has collapsed before the contactor pole starts to open (typical 30–50 ms contactor drop time).
4. Eight-motor load test. Connect all eight branches. Start all eight. Run for 30 minutes at rated frequency. Check each branch current with a clamp meter and compare with the motor nameplate FLA. Branch currents must be within ±10 % of each other.
5. Yarn-break simulation. With the machine running, pull thread from sensor 4. Verify spindle 4 ramps to a stop, contactor 4 opens, and the other seven spindles continue running. Reset from the HMI and verify spindle 4 re-starts without the drive tripping.
6. Single-phase loss test. Manually open one pole of contactor 4 while running. Verify the drive trips on output phase loss (L8-07) within 5 seconds. Reset and verify normal operation resumes.
7. Fault reset cycle test. Pull thread from all eight sensors. Verify all eight spindles stop, the drive remains enabled, the HMI shows eight red dots. Press "Reset All". Verify all eight spindles ramp back up in sequence (200 ms stagger between contactors to avoid a summed inrush) without an OC trip.

Verification Checklist

Troubleshooting Matrix

Field-Proven Caveats

• Standard AC-3 contactors are acceptable for VFD outputs in IEC 60947-4-1, but only up to a dV/dt of about 1 kV/µs. Modern Yaskawa and ABB drives produce 5–10 kV/µs at the IGBT output. The contactor manufacturer derates the device for VFD duty, or you fit an output reactor / sine-wave filter to bring dV/dt below 1 kV/µs at the contactor.
• The thermal model inside the VFD is single-channel. A 2.2 kW drive on E2-01 = 9.6 A will protect the sum of eight motors from a stalled-rotor event on the bus, but it will not protect one motor from a single-phase supply loss (a single-pole contactor failure). That is what the per-branch mechanical overloads are for, and the L8-05 / L8-07 output phase loss protection in the drive.
• Output reactors heat up under PWM. A 5% line reactor at 8 kHz carrier dissipates roughly 3× the 50 Hz copper loss. Specify a Class F or Class H reactor for VFD duty; a 50 Hz reactor will fail within months.
• Do not use solid-state contactors on VFD outputs. The leakage current of a back-to-back thyristor is too high for the drive's current transformers and will cause nuisance ground-fault trips.
• The Yaskawa field machine works because the drive is oversized. Replicating the topology on a right-sized drive will fail. The one-frame-headroom rule is the entire reason the source application does not trip.