-
How I ended up with a six-year spreadsheet of motor failures
-
The unit price trap: where the 34% hides
-
The bearing factor: why "how ball bearings are made" matters more than torque curves
-
Not every application needs a maxon (and that's fine)
-
What changed in the last two years (and what didn't)
-
Where the cheap option still wins
After six years and $180,000 in motor purchases, I can tell you the single biggest cost mistake I made wasn't choosing the wrong vendor. It was choosing the lowest unit price. Across 214 orders, the two vendors with the lowest initial quotes ended up with the highest total cost of ownership—34% higher than the mid-priced option when I factored in replacement rates, downtime, and engineering time. That's not a rounding error. That's a full quarter of my annual budget I could've spent elsewhere.
Here's the part that stings: I knew better. I'd been tracking these numbers since 2019, and the spreadsheet was telling me this story for at least two years before I actually changed our procurement policy. I kept rationalizing. "This batch will be different." It wasn't.
How I ended up with a six-year spreadsheet of motor failures
I'm the procurement manager at a 140-person automation integrator. I've managed our component budget—roughly $180,000 annually across motors, drives, and controllers—for seven years. In 2019, after a particularly nasty quarter where three customer projects got delayed because of motor failures, I started logging every motor we bought: unit price, vendor, application, failure date, replacement cost, and hours of engineering time spent on the fix.
What started as a frustration project became the most useful procurement tool I have. When a sales rep tells me their motor "lasts 20,000 hours," I can check what that actually meant for the last three motors I bought in that class.
The unit price trap: where the 34% hides
Take one specific case. In 2022, we were sourcing DC motors for a medical device customer's pump system. Vendor A quoted $187 per unit. Vendor B quoted $142. On a 200-unit order, that's a $9,000 difference on paper. My boss wanted Vendor B. I almost agreed.
Then I pulled up the failure log. The previous batch of Vendor B motors—same model, similar application—had a 19% failure rate within 14 months. Each failure cost us roughly $310 in replacement labor, plus an average of 4.5 hours of engineering time to re-qualify the replacement unit. Vendor A's motor, running in a comparable application, had a 3% failure rate over the same period.
When I ran the math, Vendor A's true cost per operating unit came to $221. Vendor B's came to $296. That's a 34% premium hidden in the "cheaper" option. We went with A. Two years later, that batch has had exactly one failure.
It's tempting to think you can just compare datasheets and pick the one with the better specs per dollar. But identical torque and speed ratings from different vendors can result in wildly different outcomes—because the specs don't tell you anything about manufacturing consistency or bearing quality.
The bearing factor: why "how ball bearings are made" matters more than torque curves
I didn't fully understand motor reliability until I spent a day at a bearing supplier's plant in 2023. If you've ever searched "how ball bearings are made" out of curiosity, you probably saw the basics: steel wire drawn into balls, ground, polished, heat-treated. What the videos don't show is the tolerance stack-up—and that's where the real cost lives.
ABEC-1 bearings have looser tolerances than ABEC-7. For a hobby servo running at low duty cycle, the difference is academic. For an industrial motor running 16 hours a day, it's the difference between 8,000 hours and 20,000 hours of service life. The maxon DC motor we now use in that pump application runs ABEC-7 bearings. The Vendor B motor used economy-grade bearings—I checked. That single component choice explains most of the failure gap.
The surprise wasn't the price difference on the motor. It was how much hidden value came with the "expensive" option—bearing quality, documented tolerance ranges, and a datasheet that actually reflected production reality rather than a cherry-picked test unit.
Not every application needs a maxon (and that's fine)
I want to be careful here, because I've watched engineers over-spec motors just as often as procurement under-specs them. We still buy cheap servos for prototyping. The MG996R servo motor—about $6 a piece—is perfectly fine for bench testing a gripper mechanism or a first-article prototype. It doesn't have the torque consistency or the lifespan of an industrial servo, but for a 40-hour test, neither matters.
Similarly, we use A4988 stepper motor drivers on our in-house 3D printers and small CNC fixtures. The A4988 is a $2 chip. It's not going into a customer-facing medical device, but it's the right tool for open-loop positioning where the load is predictable and a missed step just means re-homing the axis.
The calculus changes when the motor is in a customer's product or a 24/7 production line. That's where paying for maxon-grade components—or specifying a maxon e-bike motor for an e-bike drivetrain application, where weight and reliability both matter—starts to look less like a premium and more like insurance.
What changed in the last two years (and what didn't)
Three years ago, I would've told you the mid-tier motor market was a wasteland. Vendors either sold premium with the documentation to back it up, or they sold commodity motors with a spec sheet and a prayer. That's changed. Since 2023, we've qualified two mid-priced vendors with real documentation and consistent quality. What was best practice in 2020—"always buy the premium brand for industrial applications"—isn't the whole story anymore.
But the fundamentals haven't changed. A motor is a mechanical system, and mechanical systems fail at the bearing, the winding, or the seal. If a vendor can't tell you the bearing grade and the insulation class, the rest of the datasheet is decoration.
Granted, our approach requires more upfront work—vendor audits, sample testing, keeping a failure log. I get why most teams skip it. The transaction cost is real. But the spreadsheet I built in 2019 has paid for itself roughly eleven times over in avoided replacements and prevented project delays.
Where the cheap option still wins
I can only speak to our context—mid-size B2B integrator, predictable order volumes, customer-facing industrial products. If you're running a university lab or a startup building ten prototypes, the TCO math looks different. Replacement labor is cheaper when your engineers are already hands-on. Downtime doesn't carry the same customer penalty. For a breadboard-stage project, the MG996R is still the smart buy.
But when the motor goes into something that ships, or something that runs unattended, the unit price stops being the interesting number. The interesting number is the one that shows up 14 months later when the "cheap" motor fails and you're explaining to a customer why their line is down.
I still kick myself for the 2022 batch we bought from Vendor C before I'd built the failure log. If I'd applied the same TCO spreadsheet to that order, we'd have avoided $11,000 in rework and six weeks of schedule slippage. The data was there. I just didn't look at it.
"The numbers don't lie, but they don't volunteer either. You have to go looking for them."
That's the whole lesson, really. The cheapest motor isn't the one with the lowest price tag. It's the one you only have to buy once.