When the Line Stops, the First Diagnosis Is Almost Always Wrong
In Q3 2023, we had 620 linear actuators fail in the field within the first nine months of service. Not the whole batch—about 14%—but that's enough to trigger a customer escalation call. The customer's engineering team sent us a failure report with one line at the top: "Motor quality issue." I read that line and sighed, because in my four years of reviewing actuator and motor assemblies, that's the most expensive sentence you can write before you actually understand what went wrong.
Here's what I mean. We pulled the failed units back. We bench-tested the motors themselves. Out of 620 units, 41 had actual motor defects—burned windings, one seized bearing, a handful of encoder dropouts. The rest? The motors were fine. The failures clustered around three things: lead screw wear from misalignment, thermal cutoff trips masking as "dead motors," and controller firmware mismatches that were never caught at incoming inspection because nobody flagged the revision numbers.
That's roughly 6.6% genuine motor failure and 93% something else. If we had accepted the initial diagnosis, we would have re-spec'd motors, negotiated a claim with the motor supplier, and shipped the same failure back to the customer six months later.
The Deeper Problem: Actuators Are Assemblies, Not Components
Most engineers I work with—and I include myself—started by treating the linear actuator as a motor with a screw attached. Spec the motor, spec the stroke, spec the load. Done. I used to think that way. I don't anymore.
The reality is that a linear actuator is a stack of five or six subsystems that each degrade at a different rate, and failure usually happens at the weakest interface, not the most expensive component. The motor might be rated for 10,000 hours. The lead screw nut might be good for 3 million cycles under ideal alignment. But the coupling between them—that's where tolerance stack-up, thermal expansion, and lubrication migration quietly do their damage.
What we found in our teardown data
- Lead screw misalignment: 290 units (47%) showed uneven wear patterns on the nut threads, consistent with lateral offset introduced during assembly. The motors were healthy.
- Thermal cutoff trips: 180 units (29%) triggered over-temperature protection under duty cycles the customer had described as "continuous." Their definition of continuous was not ours. Peak-to-average current draw was 2.7× higher than the datasheet curve we'd used for sizing.
- Firmware mismatch: 109 units (18%) had controller boards two revisions behind the actuator spec, producing current-limit behavior that mimicked motor burnout.
- Genuine motor defects: 41 units (6.6%).
Notice what's missing from that list: nothing about "cheap motors." Every one of these failure categories would have occurred even if we had specified a top-tier motor. Which raises the question every procurement team eventually has to answer honestly—when does motor quality actually move the failure rate, and when are you buying premium components to compensate for a system you didn't specify correctly?
Where Motor Choice Actually Matters
I went back and forth for two weeks on whether to move our sewing machine servo motor line to a higher-cost supplier after a separate quality event in 2022. The budget unit met the nameplate specs—750W, 3,000 rpm, reasonable torque curve. On paper, it was fine. Our pilot run said otherwise.
Sewing machine servo motors live in a specific kind of hell. They start and stop tens of times per minute, the load is highly variable as the needle penetrates fabric of different thicknesses, and the response time has to be tight enough that stitch quality doesn't drift. That's an application where peak torque, not continuous torque, governs the sizing. The budget motor hit its continuous rating fine. At 2.8× continuous during rapid acceleration events, its torque dropped off faster than the curve suggested—not catastrophically, but enough that our stitch consistency tests started showing drift at 8,000 rpm equivalent spindle speed.
We tested a maxon motor in the same frame size. Same nominal story on the datasheet. But the torque hold at peak was noticeably flatter, and—this is the part that's hard to put in a procurement spreadsheet—the batch-to-batch consistency was tighter. We measured torque ripple across 30 units from three production lots. The maxon lot-to-lot variance was under 3%. The budget supplier's was 12%. When you're tuning a servo loop around a motor, that variance is the difference between one tuning profile and three.
On maxon motor price, plainly
I get why maxon motor price becomes a running joke in procurement meetings. A comparable frame size can run 2.5 to 4× the cost of an Asian-manufactured equivalent, and on a line where you're buying 5,000 units a year, that's real money. To be fair to the skeptics, plenty of applications don't need that delta. A conveyor indexing actuator running at constant load, in a climate-controlled plant, with a two-week maintenance window? Buy the budget unit. It'll serve you fine.
But the price gap isn't arbitrary branding. In our tear-down comparisons, the differences concentrated in winding consistency, magnet grading, encoder resolution stability, and—the one I've come to care most about—documentation completeness. Every maxon DC motor shipment we've received has come with traceable lot numbers, torque constant data per batch, and thermal derating curves that match our own test bench results within 5%. Two of our lower-cost suppliers send a certificate that says "conforms to spec" and nothing else. When an actuator fails in a medical or automotive end application, that paperwork gap becomes a liability gap.
Is it worth the premium every time? No. But in actuator designs where the motor is the failure-critical, hard-to-service component, the premium is cheaper than the alternative. We've run the numbers on a liftgate motor program where field replacement cost (labor + customer downtime credit) ran $340 per unit. A motor that cut field failure rate from 4% to 1.2% paid for itself in under 18 months, even at 3× unit cost.
The Stepper Motor Trap Nobody Warns You About
Industrial stepper motors are the seductive choice for linear positioning. Cheap, simple, no encoder needed if the load is predictable. I've specified them. I'll specify them again. But I've also seen them fail in a way that costs more than a servo would have.
Open-loop steppers don't tell you when they've lost steps. The motor keeps running, the controller keeps counting, and the position drifts—quietly—until something jams or the motion profile walks off the end of the stroke. On a light-load indexing application, you may never notice. On a heavy vertical load with a lead screw that's picking up friction over time, you'll notice when the actuator stalls mid-stroke with product on it.
We caught this in 2021 with a high-speed camera setup—not because we suspected the steppers, but because the machine's reject rate kept creeping up and nobody could explain it. At 40% of rated load, one stepper model started losing steps during rapid direction reversals. Not at peak load. At 40%. The datasheet torque curve was accurate; the issue was that the curve didn't describe dynamic behavior during reversals, and nobody on the design team had thought to test for it.
The fix was a hybrid closed-loop stepper—cost increase of about $60 per axis, on 14 axes. Trivial compared to the reject rate we were eating. But the lesson stuck: stepper torque curves describe steady-state capability. They don't describe what happens at the moment of reversal, which is exactly where most stepper-driven actuator failures originate.
The Cost of Getting the Diagnosis Wrong
The 620-unit field failure I opened with cost us more than the actuator replacement. The customer's line was down 11 hours—not because we couldn't ship replacements quickly, but because their maintenance team had to disassemble, diagnose, and reinstall each unit. At their stated line output value, the unplanned downtime came to roughly $180,000. We absorbed the replacement cost and a portion of the downtime credit. The annual supplier scorecard took a hit that took three quarters to recover.
None of that was caused by a bad motor. It was caused by a specification that said "continuous duty" when the customer meant "high-frequency start-stop," a controller firmware mismatch that our incoming inspection missed because we were checking part numbers and not revisions, and an assembly tolerance that allowed lead screw misalignment on roughly one in seven units.
That quality issue cost us $22,000 in rework and delayed a product launch by six weeks. The motor was never the problem. But the motor was the first thing everyone blamed, because motors are the most visible component and the easiest thing to point at.
What Actually Prevents Actuator Failures
Three things, in order of impact—and none of them is "buy more expensive motors."
- Define duty cycle honestly. Peak current, frequency of reversals, thermal environment, expected service life. Get the customer to sign off on the definition. If they say "continuous," ask what percentage of time the actuator is at peak load. The answer is almost never what the word suggests.
- Test the assembly, not just the components. Our incoming inspection used to check motors, screws, and controllers separately. Now we run every actuator through a 72-hour burn-in that includes thermal cycling and reversal stress. The failure rate in the field dropped 61% in the first year after we implemented it.
- Match motor class to failure consequence, not to load. A maxon motor isn't "better" in some absolute sense—it's more consistent, better documented, and more predictable at the edges of its spec. That matters when a failure causes downtime or safety exposure. It matters much less on a prototype or a low-duty indexing application. Knowing the difference is the job.
Last thing. If an actuator fails and the first explanation is "the motor died," slow down before you accept it. Pull the unit apart. Measure the lead screw wear. Check the controller revision against the BOM. Look at the thermal log. The motor is the easiest component to blame and, in my experience, rarely the actual root cause. Diagnosing the failure correctly the first time is usually worth more than any component upgrade you could make.