What a Starting Method Changes About the Supply

Why this matters

Somebody proposes a soft start because the lights dip, it goes in, the dip improves, and eight weeks later the motor's overload is opening on a schedule nobody can explain. The proposal was not wrong about the dip. It was silent about what the same change does to the motor, and those are two separate numbers that only some starting methods move together. Knowing which methods separate them is the difference between solving a supply complaint and moving it into the machine, where it costs a winding instead of a lightbulb.

The one relationship everything else hangs off

For an induction motor at a given slip, developed torque varies roughly with the square of the terminal voltage. Reduce terminal voltage to 65 percent and torque falls to about 42 percent. That relationship is what every voltage-reducing starting method is exploiting and what every one of them is limited by, and it applies to the motor terminals only. It says nothing about what the line supplying the starter is doing.

Two more relationships get used below, and both are stated here so they are not smuggled in later.

Acceleration time varies inversely with the net accelerating torque, which is the motor's torque minus the load's torque, for a given total inertia. Real torque curves vary with speed, so treating both as averages across the acceleration is an approximation that is good enough to compare methods and not good enough to set a protection setting.

Winding heating during a start goes as current squared times time. That is why a longer start at lower current is not automatically gentler.

What the line sees is not what the motor sees

A method that merely drops voltage puts the same current through the line that it puts through the motor. A method that transforms voltage does not, because a transformer takes less current on its primary than it delivers on its secondary. That single distinction sorts the whole field.

Read the autotransformer row against the solid-state row. At the same tap the motor is in an identical situation in both, so it produces identical torque, accelerates in identical time and heats identically. The line is not in an identical situation: the autotransformer draws k squared where the solid-state starter draws k. At a tap of 0.65 that is 42 percent of locked-rotor current on the line against 65 percent, for exactly the same result at the shaft.

The variable frequency drive is the only row that escapes the square law, and it escapes it by moving a second variable. Reducing frequency along with voltage keeps the magnetic flux in the machine roughly constant, so the motor develops useful torque at low speed with current near its full-load value rather than near locked rotor. Everything else in the table is trading torque for current on the same curve; the drive is changing the curve.

Part-winding is the row with no numbers, deliberately. The split between the winding sections is a property of that specific motor's design, the motor must be built for it, and the resulting current and torque fractions come from the motor manufacturer's data rather than from any rule of thumb.

Four things no starting method changes

The energy the rotor must absorb to accelerate the inertia. Getting a given rotating mass to a given speed takes a fixed amount of kinetic energy, and for an induction machine accelerating from rest a comparable amount is dissipated in the rotor along the way. Slowing the acceleration spreads that energy over more time, which helps a rotor conduct some of it away, but it does not reduce it.

Whether the load will break away at all. If the reduced starting torque is below the load's breakaway torque, the machine sits at zero speed drawing its reduced starting current until something opens. That is worse than not fitting the starter, because a stalled motor makes no torque and gets no cooling from a shaft-mounted fan.

Running current, running power factor or running voltage. Every method in the table is out of circuit or at full output once the motor is at speed. A supply that sags at running load has a running-load problem.

The overcurrent protection of the branch circuit. The conductors and their protection are sized to the running load and the motor's characteristics under the NEC in the edition your authority having jurisdiction has adopted, and a starting method does not license a change to either.

Worked example: the dip got better and the motor got worse

A high-inertia centrifugal fan. Values here come from the motor and equipment data for this machine, not from typicals.

• Across-the-line acceleration time: 12 seconds.
• Load torque, averaged across the acceleration: 20 percent of the motor's across-the-line starting torque.
• Proposed remedy for the dip: a solid-state voltage ramp holding 65 percent of terminal voltage.

At full voltage the net accelerating torque is 100 minus 20, which is 80 percent. At 65 percent terminal voltage the motor makes 0.65 squared, which is 42.25 percent, so the net accelerating torque is 42.25 minus 20, which is 22.25 percent.

Acceleration time scales inversely with net accelerating torque, so 12 seconds times 80 divided by 22.25 gives about 43 seconds.

Now put that through the heating relationship. At full voltage, current squared times time is 1.0 squared times 12, which is 12 units. At 65 percent, it is 0.4225 times 43, which is about 18 units. Winding heating rose by roughly half again, from 12 units to 18, while line current fell by 35 percent.

That is the trade, stated honestly: the supply got quieter and the motor got hotter. If the motor's data sheet gives a maximum permitted accelerating time shorter than 43 seconds, and for a machine whose across-the-line time is 12 seconds it very likely does, the starter as configured has put the motor outside its own limit on every start, and the overload opening two months later is the protection working rather than failing.

Change one input and the recommendation changes. Fit an autotransformer at the same 0.65 tap instead. The motor sees the identical 65 percent, so acceleration is still about 43 seconds and the winding still takes about 18 units of heating; nothing at the shaft improves. But the line draws 0.65 squared, or about 42 percent of locked rotor, rather than 65 percent, so the dip is substantially better than the solid-state option delivered. If the goal is purely the dip, that is the better trade for the same motor penalty. If the motor's accelerating-time limit is the binding constraint, neither method is acceptable and the answer is a drive or a larger motor, because both of those change the torque available rather than rationing it.

Change the load instead. Put the same 0.65 ramp on a loaded conveyor whose breakaway torque is 60 percent of the motor's across-the-line starting torque. The reduced starting torque is 42.25 percent, which is below 60, so the belt never moves. Current sits at its reduced starting value with the rotor stationary, and the ramp's own overload or the motor's protection ends the attempt. Nothing in the fan arithmetic transfers, because the fan case turned on acceleration time and this one turns on whether acceleration begins.

The failure mode in the field. Both of those look like a starter problem. The conveyor gets a bigger starter and the fan gets a replacement overload relay, and the actual constraint, which is the motor's torque at reduced voltage against the load's demand, is never written down anywhere on the ticket.

Testing a start without creating a second problem

Test starts are how this gets confirmed, and they carry hazards your own instruction creates rather than hazards that come at you from the equipment.

Count the starts against the motor's permitted starts per hour and its permitted accelerating time, both from the motor data, and stop at that number rather than at the point where you are satisfied. An extended-acceleration test is thermally worse than a normal start by the ratio the example above computes, so a reduced-voltage test start consumes more of the motor's thermal budget than a full-voltage one does, and repeating it needs the cooling interval from the same data.

Where the driven machine has its own hazards, the extended start is where they show. A fan accelerating for 43 seconds spends that time at low speed, so a motor relying on a shaft-mounted cooling fan is being heated hardest exactly while it is cooled worst. A conveyor that fails to break away is a stalled drivetrain with stored torque in the belt, so keep clear of the return run and the take-up before an attempt.

Where measurement during the start is unavoidable, that is energized work: 29 CFR 1910.333(a)(1) permits it only where the employer can demonstrate that de-energizing introduces additional or increased hazards or is infeasible due to equipment design or operational limitations, and the boundaries and PPE come from the risk assessments at NFPA 70E-2021, 130.5 and 130.7, in the edition your employer's electrical safety program adopts. Isolation for any work inside the starter follows 29 CFR 1910.333(b)(2), not 29 CFR 1910.147, which excludes exposure to electrical hazards from work on conductors and equipment in electric utilization installations at (a)(1)(ii)(C); the construction counterpart is 29 CFR 1926.417.

One boundary that is not a field call: fitting a starting method to a packaged, listed assembly may put the assembly outside the configuration it was evaluated in. Whether that voids the listing is a question for the listing agency and the authority having jurisdiction, and the equipment manufacturer's instructions are enforceable through NEC 110.3(B) in the adopted edition. Write the question up rather than answering it on the roof.

Who owns which number

The motor manufacturer owns locked-rotor current, permitted accelerating time, permitted starts per hour and whether the machine can be started part-winding or run in delta. The driven-equipment manufacturer owns breakaway torque and inertia. The adopted NEC edition and the authority having jurisdiction own conductor sizing and overcurrent protection. The starter manufacturer owns the tap or ramp settings available and the duty rating at each. No one of those four can answer another's question, and a proposal missing any of them is a guess with arithmetic on top.

References

• 29 CFR 1910.333(a)(1) energized-work gate and (b)(2) electrical lockout; 29 CFR 1910.147(a)(1)(ii)(C) exclusion of electric utilization installations; 29 CFR 1926.417 construction counterpart
• NFPA 70E-2021, 130.5 and 130.7 for boundaries and PPE, in the edition adopted by your employer's electrical safety program
• NEC Article 430 (motors, motor circuits and controllers) and 110.3(B), in the edition adopted by your authority having jurisdiction
• Motor and driven-equipment manufacturer data for locked-rotor current, accelerating time, starts per hour, breakaway torque and inertia
• See related: What Inrush Does to Everything Else on the Supply; How to Decide Whether a Supply Can Start What You Are Adding