September 2022. The phone rang at 4:30 PM. A maintenance manager from a food packaging plant had a case packer that kept losing position on one axis. Not every cycle. Just often enough to wreck a case and stop the line. The OEM quoted a ten-week lead time for a replacement actuator. He had eleven days before the next scheduled downtime. I said, "We can do that."
I've been handling linear motion and drive component orders for nine years. I've documented 31 significant mistakes in that time—roughly $47,000 of reusable lessons. That September, I made mistake #32.
The call I almost answered too fast
Here's the thing: when you get a call like that, your first instinct is to be the hero. You jump straight to a solution. My solution was to blame the old stepper motor controller. It was a legacy unit, the kind with jumpers and a cable that looked like it had been through a war. I'd seen it happen before—or thought I had. The machine was losing steps, so the controller must be losing its mind. That was my theory.
I was wrong.
Look, I'm not saying the old controller was perfect. It probably needed replacing anyway. But the real problem was mechanical. And because I skipped the mechanical check, I sent a client on a $3,200 detour.
When I first started doing field diagnostics, I assumed a failing bearing always makes noise. It doesn't. This bearing was silent. No grinding. No heat. The only sign was a slow, intermittent positioning error that got worse over two weeks.
What happens if a ball bearing goes out? Not what I assumed
Everyone talks about the dramatic bearing failure—the screech, the vibration, the seized shaft. Those happen. But there's a quieter failure mode, and it's the one that fools people.
A ball bearing doesn't always "go out" like a light switch. A small spall forms in the raceway. The balls get slightly less round. The cage begins to drag. The torque on the ball screw starts fluctuating. Not enough to stop the motor—just enough to make a stepper lose a few steps when it hits that rough spot. The controller then shows an error. If all you're looking at is the control cabinet, you'll chase firmware while the bearing keeps getting worse.
So what happens if a ball bearing goes out? Sometimes it becomes a "stepper motor controller" problem. That's what we called it. It wasn't.
In our case, the old bearing was a support bearing on the ball screw. When we finally pulled the screw out, I rotated it by hand. There was a rough spot, same angle every revolution. You could feel it clearly. No noise. No heat. Just drag. That drag was enough to trip the stepper's following error.
The first fix: a new stepper motor controller
I ordered a new stepper motor controller and a matching power supply. The install took a day. The machine ran smoothly for about a day and a half. Then the error came back, worse than before. Because now the bearing had more current pushing against it.
Even after I hit "order" on that controller, I second-guessed myself. What if the motor encoder was the real problem? I didn't relax until the part arrived. And then I relaxed for the wrong reason.
Looking back, that's the most embarrassing part. Not the wrong diagnosis—that happens. The problem was I doubled down. I increased current. I adjusted acceleration. I tried to make the drive train behave through programming. It was like changing tires to fix a bad wheel bearing.
The client's maintenance manager didn't call me out. He just said, "Maybe we should look at the screw." I should have looked at it three days earlier.
I also had that moment of doubt after the first replacement parts arrived. What if the ball screw was bent? What if our bearings were fine and this was all a waste of time? Those worries stayed with me until we had the old screw in my hand and could feel the rough spot.
The real problem was mechanical
We disassembled the axis and found the bearing damage. Then I made the second mistake.
I ordered the right replacement components from the Thomson linear bearings catalog. Open-style linear bearings, guide rails, a new ball screw, and a set of support bearings. The Thomson linear bearings catalog has all the sizing data you need—load ratings, shaft specs, installation notes. I read the load ratings. I skimmed the installation notes.
On reassembly, I used a rigid coupling between the stepper motor and the ball screw. The old coupling was rigid, and I decided to keep the design simple. I told myself there was nothing wrong with that. But there was.
Mounting tolerances on that machine were not perfect. The motor plate was off by about six thousandths of an inch. The old bearing housing had some wear. Those differences don't matter much to a flexible shaft coupling—they get absorbed. A rigid coupling transmits every one of them straight into the bearings.
Eleven days after the second repair, the axis started losing position again.
Worse than expected.
A flexible shaft coupling was the actual fix
During the third repair—yes, there was a third—I stopped guessing. I measured the alignment. I checked the machine's existing wear. I ordered a flexible shaft coupling. Not a fancy one, but the right type for a stepper application: torsionally stiff enough to maintain accurate motion, flexible enough to handle small misalignment without loading the bearings.
The difference was immediate. The axis moved smoother. The following error dropped. And it stayed fixed.
A flexible shaft coupling costs more than a rigid coupling. It also costs a lot less than a second service call and a weekend of downtime. That's the kind of math I wish I'd done earlier.
I also went back to the catalog and found the section I'd skimmed. According to the Thomson linear bearings catalog (thomson-linear.com), bearing life depends on more than load. Shaft hardness, straightness, and alignment all affect performance. If you skip those factors, you're not selecting a bearing—you're just guessing.
What I would tell someone going through the same thing
If you're trying to fix a stepper axis that's losing position, do this first:
- Check the mechanical path by hand. Turn the screw or push the carriage. Feel for rough spots, hard spots, inconsistent drag. Do this before you touch the electronics.
- Ask what happens if a ball bearing goes out in your application. The answer is not always noise. It may be a following error on the controller.
- Use a flexible shaft coupling on any motor-to-screw connection where there's even a chance of misalignment. The extra cost is tiny. The insurance is huge.
- Read the catalog notes. The dimensions are only half the story.
I'd also recommend considering a complete actuator module. In another area of the same plant, they had Thomson linear actuators on a different line, and those had run for years without a bearing failure. A finished module eliminates a lot of alignment variables. But if you're doing a retrofit with a stepper motor controller and an existing base, you have to handle alignment yourself.
Bottom line
I'd rather spend ten minutes explaining what a flexible shaft coupling does than drive back for a second call. An informed customer asks better questions and makes faster decisions. That's true for me too.
That September order taught me more than any training class. I still maintain the checklist. Mistake #32 is now in it, next to a note that says: check the bearing first.
The machine has been running ever since. The maintenance manager sends me a text every so often. It just says, "Still running."
Best message I get all year.