Electronics Guide

Voltage Dip on the Supply

Inrush current flows through the impedance of everything upstream, and the drop it produces appears at every other load on the same bus. On a stiff utility feeder the dip may be negligible; on a lightly loaded distribution transformer, at the end of a long cable run, or on a standby generator, it is not. Generators are the hardest case, because a synchronous generator's transient reactance far exceeds that of an equivalent-rated transformer, so a set sized comfortably for the running load will often stall or trip when the same motor starts across the line.

The consequences cascade. Contactor and relay coils drop out below roughly 65 to 85 percent of rated voltage, so a motor start on one machine stops another; switch-mode supplies feeding programmable controllers reset; drives elsewhere on the bus trip on undervoltage; and discharge lighting extinguishes and needs a restrike delay. None of these failures announces its cause. Where several large machines restart in sequence after an outage, the recorded peak on a tariff billed by interval demand can also rise enough to matter.

Mechanical Shock

A motor connected directly to the line develops its locked-rotor torque immediately, roughly 1.5 to 2.5 times rated torque for a Design B machine, applied as a step to a driveline that was at rest a millisecond earlier. The same flux transient that inflates the first-cycle current also produces a violently oscillatory electromagnetic torque during the first few cycles, alternating in sign at line frequency with peaks well above the steady locked-rotor value. Drivelines absorb this imperfectly: belts slip, glaze, and stretch; keyways elongate; gear teeth fatigue at the pitch line; and coupling elements degrade at a rate set by the number of starts. In pumping systems the abruptness reaches the fluid as well, because the pump goes from rest to full flow faster than the discharge column can accelerate, slamming check valves and shocking pipe supports.

Thermal Stress in the Motor

The rotor has no cooling path fast enough to matter during a start, so heating there is effectively adiabatic. For a load with no opposing torque, the energy dissipated in the rotor while accelerating from rest to synchronous speed equals the kinetic energy finally stored in the rotating mass, a classical result that makes high-inertia loads the limiting case, and adding load torque increases the loss further. Manufacturers publish a safe stall time, often between 10 and 20 seconds from cold for a general-purpose machine and considerably less from hot, and nameplates limit starts per hour accordingly. A soft starter changes the shape of the current curve but not this arithmetic; because a soft start lasts longer than a direct-on-line start, it frequently puts more total energy into the rotor, not less. The soft starter protects the supply and the mechanics, not the rotor.

Star-Delta Starting

The star-delta, or wye-delta, starter uses a motor wound for delta connection at line voltage and brought out to six terminals. Starting in star places line voltage across two windings in series, so each sees about 57.7 percent of line voltage and both torque and line current fall to one third of their direct-on-line values. The hardware is three contactors and a timer. Two limitations kept it from being universal: the reduction is fixed with no adjustment, and the common open-transition arrangement disconnects the motor before reconnecting it in delta, so reconnecting out of phase with the retained rotor flux produces a transient that can exceed the inrush the starter was installed to avoid. Closed-transition versions add resistors and a fourth contactor to bridge the gap. Star-delta remains reasonable where a six-lead motor already exists, the load is light, and cost dominates.

Autotransformer Starting

An autotransformer starter applies a tapped fraction of line voltage, conventionally 50, 65, or 80 percent, then transfers the motor to full voltage. Because it reduces line current by the square of the tap ratio, it delivers more starting torque per ampere drawn from the supply than any series method, and the Korndorfer connection sequences the contactors so the motor is never disconnected, using part of the winding as a series reactor during changeover. The autotransformer is bulky, expensive, and thermally rated for a duty cycle that restricts starts per hour, but it survives where its one advantage is decisive: a very large motor on a weak supply where the required torque cannot be obtained within the available line current by any series method.

Primary Resistance and Reactance Starting

Inserting resistors or reactors in series with the stator drops voltage at the motor terminals, and shorting them out after a timed interval completes the start. This is the simplest series method and the worst in torque terms, because line current falls in direct proportion to motor voltage while torque falls with the square. It has one attractive property no fixed-tap starter shares: as the motor accelerates and current falls, the drop across the series element falls with it, so terminal voltage rises automatically and the acceleration is inherently smooth with no control at all. Series reactors remain common on medium-voltage motors, where a medium-voltage semiconductor starter is expensive and a reactor is simple and rugged.

What the Soft Starter Changed

The thyristor starter offered neither better torque per line ampere than an autotransformer nor lower cost than a star-delta contactor set. It offered adjustability, stepless transition, and integration. The starting profile became a set of parameters rather than a tap or a winding ratio, so one product served a range of loads and could be retuned when the load changed. There is no transition event at all, because conduction rises continuously to full and the bypass then closes across an already-conducting device. And because the controller must measure three-phase current and voltage to do its job, motor protection, metering, and fieldbus communication came almost free.

Voltage Ratings, Rate-of-Rise Limits, and Snubbers

Blocking voltage is selected with generous margin over peak line-to-line voltage, because the starter sits directly on a distribution system carrying switching and lightning surges. Devices rated 1,200 volts are typical for 208 to 480 volt systems and 1,600 volts or higher for 600 and 690 volt systems, with metal-oxide varistors across each device pair and line to line clamping what remains.

A thyristor also turns on spuriously if forward voltage rises faster than its critical rate of rise, because displacement current through the device's internal capacitance acts like gate current. Phase control creates exactly this stress, since a device holds off most of the line voltage during its delay interval and then sees a step when another phase fires. The traditional remedy is a resistor-capacitor snubber across each pair, sized so the capacitor absorbs the step while the resistor limits the capacitor's discharge current into the thyristor at turn-on; modern high rate-of-rise devices allow smaller snubbers, and some designs omit them where line inductance suffices. The complementary stress appears at turn-on, since conduction begins near the gate and spreads across the die at finite velocity, so a hard gate drive with fast rise spreads conduction before the current can concentrate.

Gate Drive

Every thyristor cathode floats at line potential, and the six cathodes sit at different potentials, so gate drive must be isolated. Pulse transformers remain the standard solution because they provide isolation and energy in one rugged component; optically coupled thyristor drivers appear in smaller ratings.

The gating strategy deserves more attention than its apparent simplicity suggests. The load is strongly inductive, so at large delay angles current is discontinuous and each device conducts for only part of its half cycle. A single narrow gate pulse can fail to latch, because the anode current available at that instant may fall below the device's latching current, particularly early in a start when the motor's phase angle is largest. Designs answer this with a long gate pulse held until anode current is safely above latching, or with a train of narrow pulses issued through the conduction window. Pulse-train firing also handles the three-phase interaction, in which a device cannot conduct until a device in another phase provides a return path, so firing signals must overlap in the correct pattern.

Fusing, Coordination, and Isolation

Thyristors fail faster than ordinary fuses clear, so protecting them against a downstream short circuit requires high-speed semiconductor fuses, designated aR or gR in the IEC scheme, chosen so the fuse's total clearing energy stays below the device's withstand energy. IEC 60947-4-2 defines two coordination types: Type 1 permits damage to the starter provided no hazard escapes the enclosure, while Type 2 requires the starter to remain suitable for further use. Achieving Type 2 requires the exact protective device the manufacturer tested, and substituting an equivalently rated device from another family invalidates the coordination and, in the ordinary case, destroys the thyristors.

A point that costs equipment and occasionally injures people: a thyristor pair is not an isolating device. It leaks in the off state, it can fail short, and it provides no verified break. Every installation therefore needs an upstream isolating contactor, switch, or circuit breaker for maintenance, and the lockout procedure must operate that device rather than relying on the starter's stop command. Practice for this class of equipment is covered under Electrical Safety.

Zero-Crossing Detection and Synchronization

Every firing instant is measured from a reference derived from the supply, with resistive dividers scaling the line voltages to logic levels and comparators marking the crossings. Two problems complicate this. The supply is not clean: notching from other converters, ringing from capacitor switching, and ordinary distortion all produce spurious crossings, while the filtering that suppresses them introduces a phase shift that must be compensated as a function of line frequency. And once the motor turns, its own back-electromotive force appears at the terminals during the non-conducting interval, so terminal voltage measured downstream of the thyristors is not a reliable phase reference. Designs therefore synchronize to the line side, upstream of the power devices.

Robust firing control trusts no single crossing. The controller estimates line frequency and phase over many cycles, typically with a software phase-locked loop, so one corrupted crossing perturbs the output negligibly. Automatic detection of 50 and 60 hertz operation is standard, and starters intended for generator supplies must also tolerate frequency excursions during the start itself, since the generator slows under the load it is being asked to accept.

Half-Cycle Symmetry

The two devices of an inverse-parallel pair must be fired at equal delay in their respective half cycles. If they are not, the current acquires a direct-current component, which in the motor produces a stationary field and hence braking torque and heating without contributing to acceleration, and which upstream can drive a supply transformer toward saturation. Controllers enforce symmetric firing in software and monitor the measured current for the asymmetry that indicates a failed device or an open gate lead. The same asymmetry is exploited deliberately, and briefly, when a starter offers direct-current injection braking.

Voltage Ramp

The controller applies an initial voltage, often called the pedestal or initial torque setting and typically adjustable from about 20 to 60 percent, then increases it linearly to full over a programmed time, commonly from one second to thirty seconds or more. The mode is open loop: whatever current the load demands at each voltage is the current the supply provides. Setting the pedestal is the whole art. Too low and the motor does not break away, sitting at zero speed drawing current and heating until the start-time protection intervenes; too high and the start begins with an unnecessary current step. The practical procedure is to set it just above the value at which the shaft reliably begins to turn, then add margin for the worst case the load presents, such as a cold, viscous, or long-idle condition.

Current Limit

The controller closes a loop on measured root-mean-square line current and adjusts the delay angle continuously to hold that current at a set multiple of motor full-load current, typically adjustable from about 150 to 500 percent. This is the mode that actually protects the supply, because it caps demand regardless of what the load does, and it is the correct choice whenever the reason for fitting a soft starter is voltage dip, generator capability, or demand charges. Its failure mode is specific: if the limit falls below the current needed to exceed the load torque at some speed, the motor accelerates to that speed and stops there, holding limit current until the start-time protection trips, and the current record shows the diagnosis immediately as a flat plateau with no final drop to running current. Many products ramp the current limit itself from an initial to a final value.

Torque Control

Because torque follows the square of voltage, applying a voltage profile shaped like a square root produces torque that rises linearly, and the simplest torque modes do exactly that in open loop. Closed-loop torque control goes further, estimating air-gap torque from measured voltage, current, and the phase between them and regulating that estimate against a programmed profile. The application that most rewards it is the belt conveyor, whose belt is elastic: a torque step launches a longitudinal wave that travels to the far end, reflects, and can lift product off the belt or snap a splice, while a linearly rising torque avoids exciting that mode. Vendors publish acceleration traces supporting their particular implementations; those are vendor claims, but the underlying reasoning is sound and the difference on a long conveyor is visible without instrumentation.

Kick Start

Some loads exhibit static friction far above their running friction: a conveyor that has stood loaded overnight, a screw compressor with oil-flooded rotors, a mixer with settled solids. A ramp correct for the rest of the start cannot break these loose. Kick start, also called pulse start or boost, applies a short burst of high or full voltage, typically adjustable between roughly 0.1 and 1 second, then falls back to the programmed ramp. The kick is by construction a brief direct-on-line start, restoring the inrush and torque step the soft starter exists to avoid, so enable it only when the load requires it and review it whenever mechanical condition changes, since a rising breakaway requirement usually means a bearing or a seal is failing.

Dual Ramp

A machine that sometimes starts loaded and sometimes empty needs two profiles, so most mid-range starters hold two or more parameter sets selectable by a digital input or over the fieldbus. Some products also produce a low creep speed through deliberately asymmetric firing, which exists to inch a conveyor into position for maintenance and is not speed control in any useful sense, since nearly all the input energy becomes rotor heat.

Why the Bypass Can Be Lightly Rated

The bypass closes while the thyristors are already conducting fully, so the voltage across the closing contacts is only the thyristor drop and the current transfers with essentially no arc. On a normal stop the controller re-establishes conduction before releasing the bypass coil, so the contactor opens with the thyristors carrying the current. Under those conditions the contactor never makes or breaks motor current and may be rated for utilization category AC-1 rather than AC-3.

Rockwell Automation states the point directly in its white paper on soft starter bypass technology: internal bypass contactors "are typically not fully rated (AC-3), because they are typically designed to not make or break load current." The same document warns that where the bypass may in fact make or break motor current, AC-3 is the minimum, and gives an explicit sizing method. For a 200 ampere motor whose starting current reaches ten times full-load amperes, the AC-3 contactor is chosen from 200 times 10 divided by 6, or about 333 amperes, the divisor being the six-times make characteristic that defines category AC-3. Its example application is a rock crusher, where a jam during running spikes the current enough that an internally bypassed starter drops out of bypass and back to thyristor control, cycling the contactor and shortening its life.

Hybrid and Fully Solid-State Starters

IEC 60947-4-2 distinguishes the two architectures through its utilization categories. Category AC-53a covers squirrel-cage motor starting, running, and stopping with an electronic switching device, that is, a fully solid-state starter whose thyristors are rated to conduct continuously, while category AC-53b covers a starter bypassed during runtime, whose thyristors carry only intermittent duty. The rating strings encode the duty explicitly, and Rockwell's published examples make the fields concrete: AC-53a:3.5-30:99-1 denotes a 350 percent current limit, a 30 second start, a 99 percent on-load factor, and one start per hour, while AC-53b:3.0-50:1750 denotes a 300 percent current limit, a 50 second start, and an off time of not less than 1,750 seconds between starts. A specification naming only horsepower has not specified the starter.

Each architecture has a domain. The hybrid starter with an internal bypass is smaller, cooler, cheaper, and needs no extra wiring, which suits pumps, fans, and conveyors running at constant load. The fully solid-state starter has no contacts to bounce, weld, or attract dust, which Rockwell cites as the reason to prefer it in high-vibration and dirty environments, and it tolerates far higher operating rates because it has no mechanical life limit. It pays for those advantages with a larger footprint, bigger heat sinks and fans, and more heat in the enclosure.

Transition, Measurement, and Failure

The controller declares the motor at speed either when the ramp completes or, better, when measured current falls below a threshold indicating the machine has pulled into its normal running slip. It then energizes the bypass coil while holding the thyristors in full conduction for at least the contactor's closing time, so the transfer is bumpless. Where the current transformers sit determines what protection remains available afterward: with an internal bypass, or an external one wired so the starter's transformers still see motor current, the thermal model and the fault functions keep running, while with some external arrangements the starter loses that visibility and a separate overload relay becomes mandatory. Rockwell notes explicitly that mounting and wiring determine whether the starter can read current and voltage while bypassed, and specifying an external bypass without checking this is a common way to leave a motor unprotected while believing otherwise.

A welded bypass contact is the failure that matters most, because it leaves the motor connected directly to the line with no way for the starter to stop it. The controller detects the condition by comparing what it commanded with what it measures, since current flowing while the bypass is commanded open and the thyristors are unfired can only mean a welded contact or a shorted thyristor. Detection alone is insufficient, because the starter cannot clear the fault it has found. The only correct response is to trip something upstream, which is why a properly engineered installation includes either a separate line contactor ahead of the starter or a circuit breaker with a shunt trip, operated by the starter's fault relay. An installation with no upstream interrupting means cannot guarantee that a stop command stops the motor, and that is a machine-safety deficiency rather than a reliability nuisance. The opposite failure, a bypass that does not close, is less dangerous but still damaging, since the thyristors then carry continuous current under an AC-53b rating that assumes they will not, so starters confirm bypass closure and trip on the absence of confirmation. Critical processes add a further emergency-bypass contactor, fully rated to at least category AC-3 and wired with its own overload relay, that runs the motor direct on line while the starter is out of service; municipal pumping is the archetypal application, and the separate overload relay is not optional because the starter has no control over that contactor.

Thermal Modeling

The controller cannot measure winding temperature unless the motor carries thermistors wired back to it, so it estimates temperature from current. The simplest model integrates the square of current against a single thermal time constant; better models keep separate stator and rotor states, because the rotor heats far faster during a start and because rotor heating is what limits starts per hour.

IEC 60947-4-1 defines the trip classes that describe how quickly an overload device responds, referenced to a tripping time at 7.2 times the current setting; class 10, class 20, and class 30 devices trip within approximately 10, 20, and 30 seconds respectively at that current. A soft-started motor almost always needs class 20 or class 30, because the deliberately prolonged start would trip a class 10 device that was perfectly appropriate for the same motor starting across the line. Selecting the class to suit the start and verifying it against the motor's published safe stall time is a required commissioning step, not a default. Thermal memory must also survive a loss of control power, since a starter that forgets its accumulated state on every interruption will permit a fourth start the motor cannot survive.

Phase Imbalance, Phase Loss, and Sequence

Unbalanced supply voltage, a loose connection, or a blown fuse produces a negative-sequence current component that rotates opposite to the rotor. The rotor sees that component at nearly twice line frequency, where the deep-bar skin effect raises the bars' effective resistance well above their direct-current value, so negative-sequence current heats the rotor far more per ampere than positive-sequence current does. Protection algorithms weight it heavily in the thermal model and trip above a set percentage of imbalance.

Loss of one phase is the extreme case: a running motor continues to turn on the remaining two phases at greatly reduced capability and overheats quickly. Detecting it on a delta-connected motor is not as simple as looking for a line with no current, since current redistributes among the windings, which is why imbalance measurement rather than a loss-of-current test is the correct function. Phase sequence detection is a related and easily overlooked feature, since reversing two supply conductors reverses the motor, which for many pumps and compressors means immediate mechanical damage.

Locked Rotor, Stall, and Underload

A start that holds at the current limit past the permitted start time indicates the load has not accelerated, and the starter trips on excessive start time or locked rotor. A jam or stall detector is a distinct function that watches for a sudden current rise while the motor runs at speed, indicating that something has seized or that material has bridged, and it must respond faster than the thermal model would.

An unexpected fall in load is equally diagnostic. A broken V-belt, a sheared coupling, a snapped auger flight, a fan that has lost its impeller, or a centrifugal pump that has lost suction all present the same signature, and underload protection trips when the load falls below a set threshold for a set time. The implementation detail separates a useful function from a decorative one. An induction motor's current changes very little between no load and about half load, since magnetizing current dominates in that region, while real power tracks shaft load closely across the whole range. An underload function watching current is therefore blind to exactly the conditions it exists to catch, whereas one computing real power from measured voltage, current, and phase detects a dry-running pump within seconds. Where a starter offers both, enable the power-based function; dry running destroys the seals of a centrifugal pump in minutes.

Shorted Thyristor Detection

A thyristor that fails, and they usually fail short, leaves one phase permanently connected. The consequence on the next start attempt is single-phase excitation: the motor hums, draws heavy current, develops no starting torque, and heats rapidly. Detecting this beforehand requires a pre-start check, and starters implement one. With line voltage present and no gate drive applied, each device pair should block, so the controller measures the voltage across each pair and looks for the blocking voltage it expects. A pair showing none is shorted, or its bypass pole is welded, and the starter must refuse to start and trip the upstream isolating device. The same check run at shutdown catches a device that failed during the run, which is the more common sequence, and gives maintenance a specific fault to act on rather than a motor that mysteriously will not start the next morning.

Ground Fault, Supply Limits, and Communications

Earth-fault detection works either by summing the three measured phase currents, which is inexpensive but insensitive because it relies on small differences between large numbers, or by a dedicated core-balance current transformer that measures residual current directly. Undervoltage, overvoltage, and frequency limits guard against supply conditions in which a start would be unwise, and a starts-per-hour lockout enforces the duty the AC-53 rating assumes. Because the starter holds current, voltage, power, and remaining thermal capacity, it is also a useful source of process data over Modbus RTU, PROFIBUS DP, PROFINET, EtherNet/IP, or DeviceNet, letting a control system trend starting current in a way that reveals a degrading bearing or a fouling impeller long before a trip occurs.