Brand Logo - Power Transmission OEM
Engineering note

Don’t Use Ball Bearings For This: What A $3,200 Servo Motor Mistake Taught Me

Conclusion first: If you are using a standard ball bearing in a servo motor application where encoder feedback is critical, you are likely introducing unacceptable error.

The short version: For most closed-loop servo systems requiring precise encoder feedback, do not use standard deep-groove ball bearings. Use angular contact bearings or preloaded sets. I learned this the hard way on a $3,200 order.


Why you should believe me (or at least take this seriously)

I've been handling motion control component orders for eight years—since 2017. I personally made the mistake I'm about to describe. It cost about $3,200 in scrapped parts, plus a one-week production delay that angered my biggest client at the time.

Since then, I've maintained our team's pre-order checklist. In the past 18 months, that checklist has caught 47 potential specification errors. Probably 40 of those were ball-bearing-related misunderstandings.

I'm not claiming to be the world's greatest engineer. But I am the guy who has documented his failures so you don't have to repeat them.


The mistake: What actually happened

It was September 2022. We had a rush order for 12 servo motor assemblies—the kind used in a pick-and-place robot for a packaging line. The customer's spec sheet called for an 'encoder on the back end, precision class bearing.'

I assumed 'precision class' meant a standard high-grade ball bearing. Didn't verify. Ordered 12 units with a standard deep-groove ball bearing—ABEC 5, good enough for general use, right?

Turned out the system required zero axial play under load. The encoder was directly coupled to the shaft. Any minute axial movement—which a standard ball bearing inherently has—shifted the encoder disc relative to the read head. We got position feedback errors of 0.2° to 0.5° depending on load.

The result: 12 units, each with a $270 bearing, plus machined housings, labor, and testing—straight to scrap. Total: $3,200. Plus the client's production line was down for an extra week while we built new assemblies with angular contact bearings.

"That $200 savings on bearings turned into a $3,200 problem plus a week of angry phone calls."

Why ball bearings fail in servo encoder applications

This isn't a knock on ball bearings—they're fine for countless applications. The issue is what you need from a bearing when your control loop is depending on accurate encoder feedback.

1. Axial play kills encoder accuracy

Standard deep-groove ball bearings have inherent axial clearance. It's part of the design—the balls need some room to roll. But in a servo motor where the encoder disc is mounted directly to the shaft, that axial movement becomes position error. The encoder doesn't know the shaft is moving axially—it just sees an unexpected rotation. The system compensates, oscillates, or rejects the feedback entirely.

2. Preload is not optional

For servo applications, you need a bearing configuration that maintains consistent contact under axial load. This usually means:

  • Angular contact bearings — designed for combined radial and axial loads, with contact angles that manage thrust
  • Paired angular contact sets — back-to-back (DB) or face-to-face (DF) arrangements that preload against each other
  • Preloaded deep-groove bearings — possible but requires matched sets and precise shimming

Key difference: Standard ball bearings typically have 5–20 microns of axial clearance. For a small encoder disc (say 50mm diameter), that 10 microns of axial movement can translate to a 0.3° position error at the encoder. That's enough to make a servo motor oscillate or overshoot.


The checklist item that saved us

After the September 2022 failure, I added one line to our pre-order checklist: "Does the encoder require zero axial play under load? If yes, specify angular contact or preloaded bearing option."

That's it. One line. And it's caught about 40 potential errors since. Not because our team is careless—but because the default assumption for most engineers is 'ball bearing = correct.' It's not, in this specific context.


When ball bearings are fine

I don't want you to walk away thinking ball bearings are useless. They're great for:

  • Simple rotary applications — fans, pumps, conveyors with no encoder feedback
  • Open-loop steppers — where position feedback isn't real-time
  • Low-speed indexing — where small position errors don't matter
  • Applications with secondary support — like a separate thrust bearing

But don't use them when: your encoder is directly coupled to the shaft and you need sub-0.1° accuracy under varying loads. That's where the angular contact bearings earn their keep.


Bottom line

In my experience, the extra $50–$150 for a proper angular contact bearing set is nothing compared to the cost of rework. I'd rather spend $100 upfront than face a $3,200 scrapped batch.

If you're designing a servo motor assembly with encoder feedback, ask yourself: Is there any axial load on the shaft? If yes—get specific about your bearing. Don't just write 'ball bearing.' Write 'angular contact bearing, DB pair, preloaded.'

Simple. Cost-effective. Nobody calls you at 2am to ask why the robot isn't doing what it's told.

Ask about this topic