How VFDs Control Motor Speed—and the 3-Phase AC Motor Mistakes That Cost Me $127,000

Two weeks before Christmas last year, a customer called me about a 3 phase AC motor that kept tripping its overload relay at 20 Hz. He'd just installed a new VFD, dialed the speed down, and the motor was pulling full current and running hot. His conclusion: the motor was junk.

My conclusion, after an hour on-site with a multimeter and the drive's parameter logs, was different. The motor was fine. The drive parameters were wrong. The fix took eight minutes of setting changes.

I'm a motion control applications engineer, handling specification and troubleshooting orders for industrial customers for 8 years. I've personally made—and documented—31 significant mistakes, totaling roughly $127,000 in wasted budget and rework. That December call was one of the cheap lessons. Some of my earlier ones were not.

So if you're fighting with a motor that won't behave on a VFD, or a linear actuator with encoder that drifts and jitters, this is written from the trenches, not from a marketing department.

The Surface Problem: 'The Motor Is Broken'

Almost every troubleshooting call I take starts with the same conclusion: the hardware failed. Someone installs a VFD, wires it up, sets a frequency—and the motor runs wrong. Too slow. Too hot. Tripping. Squealing. Or the actuator twitches and drifts instead of moving cleanly to position.

It makes sense why people blame the parts. The motor is the thing you can touch. The VFD is a black box. The encoder cable is just a wire. And if it doesn't work, one of these physical things must be defective.

But here's a pattern I noticed when I compared the calls where we swapped components against the calls where we changed settings: roughly 4 out of 5 components our customers returned as 'failed' tested perfectly fine on the bench. The hardware wasn't dead. It was being told to do something wrong.

Deep Reason #1: People Misunderstand How VFDs Control Motor Speed

Let's get the theory out of the way, because it matters. A 3 phase AC motor's synchronous speed is set by the frequency of the power feeding it: N = 120f/P. A 4-pole motor on 60 Hz runs at about 1800 rpm. Feed it 30 Hz and it runs at about 900 rpm. Change the frequency, change the speed. Simple, right?

Not quite. The VFD has to manage voltage at the same time. The motor's magnetic flux depends on the ratio between voltage and frequency. Drop the frequency to 30 Hz but keep the voltage at 460 V, and the flux saturates, current spikes, and the motor overheats or stalls. That's why VFDs maintain a V/f curve—they raise frequency and voltage together, keeping the ratio constant.

What I mean is that the drive isn't a frequency dial. It's a coordinated voltage-and-frequency controller. Most modern drives calculate the correct V/f curve automatically if you enter the motor nameplate data properly. But I've seen more than one installation where someone skipped the nameplate entry, or loaded parameters from a completely different motor, and then concluded the motor couldn't handle VFD duty. The motor was fine. It was just being treated like a different motor.

Deep Reason #2: 'Standard' Motors May Not Survive VFD Duty at All

Here's the mistake that cost me the most, and the one I'm most embarrassed about. In my first year, 2017, I approved a standard 3 phase AC motor for a VFD application because it was cheaper and immediately available. It ran at low speed for about three months before the winding insulation failed. The rewind invoice was $1,900. The production delay was three weeks.

That was my introduction to PWM harmonics. VFDs create an AC waveform by chopping DC voltage into high-frequency pulses—usually at a carrier frequency between 4 and 16 kHz. Those pulses have steep voltage edges, which stress motor winding insulation far more than standard grid power. Add long cable runs, over 50 feet say, and reflected voltage waves can almost double the voltage at the motor terminals. A motor designed for direct-on-line power isn't guaranteed to survive that. It's why inverter-rated motors exist.

So is this a motor problem? Only if you choose the wrong motor. In 2017, I was the wrong chooser.

Deep Reason #3: The Feedback Chain Is the Fragile Part

The two examples above are open-loop induction motors. But a big part of my job these days is closed-loop systems—specifically linear actuator with encoder setups, where position or speed feedback is the entire point.

And the pattern I keep seeing is this: the actuator was fine. The encoder signal was corrupted.

Encoders send A/B quadrature pulses, and usually a Z-index pulse, back to the drive or PLC. If the drive can't see clean edges, it can't know where the actuator is. The two things that corrupt encoder signals are the two things that show up on every VFD job: wiring mistakes and electrical noise.

Three years ago, I signed off on a $3,200 linear actuator with encoder order. Field wiring ran the encoder cable inside the same conduit as the motor power cable. The VFD's PWM pulses induced enough noise on the encoder lines that the drive counted phantom pulses. The actuator drifted, then jumped, then faulted out. The customer replaced the actuator once because I told them it was faulty. It wasn't. The cable routing was the problem—and the actuator itself had passed final test at our facility.

Seeing the failed install vs. the re-done install side by side made me realize something uncomfortable: we'd shipped a working product and then let the application sheet ruin it. No connector, no component, no datasheet can fix a wiring practice that turns a clean digital signal into an antenna for noise.

What This Pattern Actually Costs

Let me put real numbers on this, because 'efficiency' sounds abstract until you see the invoices:

• Motor rewind after VFD-driven insulation failure: $1,900, plus 3 weeks of production downtime.
• Replacing a linear actuator that was never broken: $3,200, plus a $130 rush-shipped replacement cable, plus the awkward conversation when I explained the return label I'd approved was for a product that had passed final test.
• The 2022 parameter mismatch that ran a machine at 70% speed for six days before anyone noticed: about $11,000 in overtime to recover production.

I don't have hard data on industry-wide VFD commissioning failure rates. What I can say anecdotally, from 8 years of calls, is that 30 to 40 percent of the 'motor is dead' reports I investigated turned out to be setup or installation issues. The parts were fine. The configuration—or the lack of it—was the killer.

The worst part is the time-pressure spiral. Last spring, a customer's line was down and they needed a motor in 48 hours. I had a choice between a standard motor on the shelf and an inverter-rated unit that was two days out. Normally I'd hold the line and wait for the right part. But with the plant manager pacing my office, I caved. I signed off on the standard motor with an internal 'it'll be fine for a while.' It ran for 11 days.

Hit 'approve' on that PO and immediately thought—did I just create a second failure? Didn't relax until the inverter-rated motor was installed and running. Then I rewrote our substitution policy from scratch. There is a reason standards exist, and it's usually written in someone else's pain.

Rule I now operate by: if a component 'fails' within the first month, suspect the setup before you suspect the part.

What I Do Now (And What I'd Tell a Younger Engineer)

I'm not going to turn this into a 40-step quality manual. Three things: motor rating. Drive parameters. Feedback integrity. In that order. Concretely, the checks that would have prevented my specific failures:

1. Confirm the motor is inverter-rated before connecting it to a VFD. If it isn't, swap the motor or add a drive output filter. This is not optional, and 'it's a short cable' is not an argument.
2. Program the drive from the motor nameplate. Let the VFD calculate the V/f curve. Don't copy parameters from another machine at the site—I've watched a whole production line run at the wrong speed because of that shortcut.
3. Separate power and feedback cables. Different conduits. Proper shielding, grounded at exactly one end. This single rule would have prevented most of the encoder issues I've debugged.
4. For linear actuator with encoder applications, verify the encoder details before ordering: incremental vs. absolute, supply voltage, output type (push-pull vs. open-collector), and pulses per revolution that match the drive's input. A mismatch usually shows up as 'the actuator moves the wrong distance.'
5. Run a no-load commissioning test at minimum speed, maximum speed, and a few middle setpoints. Watch the actual feedback values, not just the commanded ones. This takes about an hour and catches most of the problems I listed above.

That last one is the no-brainer that keeps saving us. Since we implemented these checks, we've caught 47 potential errors in 18 months. Three of them would have become motor failures. Two would have cost us a customer. I now maintain our team's pre-shipment checklist, so someone else doesn't have to repeat my education.

One more thing on verification. Per the FTC Green Guides (ftc.gov/green-guides), if a supplier slaps 'energy efficient' or 'ECO' on a product, that claim needs actual evidence behind it. I've had to invoke that once when a vendor's spec sheet promised efficiency numbers with no test data. The point applies to all of this: verify, don't assume. The cheapest part of a motion control system is the hour you spend checking assumptions.

Bottom Line

How VFDs control motor speed isn't magic. It's coordinated frequency, voltage, and—in closed-loop systems—clean feedback. The majority of the 'broken' motors and actuators I've documented over 8 years were casualties of assumptions: wrong motor duty, wrong parameters, wrong wiring, wrong signal integrity.

If you're specifying for an industrial motion project, the details matter more than the brand sticker. Johnson Electric industrial products—including 3 phase AC motors and linear actuator with encoder configurations—publish full specifications on the Johnson Electric official website. The nameplate data, duty ratings, and feedback options are there for a reason. Read them before you approve the order, not after the machine stops.

I learned that the expensive way so you don't have to.