When I first started specifying motion components for factory automation lines, I assumed every application needed the heaviest, fastest, most rigid part in the Bosch Rexroth catalog. Heaviest bearing. Fastest motor. Maximum rigidity.
Seven years as an applications engineer handling motion control selections, and I've made (and documented) eleven significant specification mistakes totalling roughly $63,000 in wasted budget. This article is the itemized bill — read it so you don't have to pay it.
There's No Universal "Right" Motion Component
There's no single "best bearing" or "best motor" in factory automation. There's only the right choice for your load profile, speed range, and duty cycle. Before you order anything, you need to know which of three situations you're in:
- High-speed, moderate load — radial ball bearings are your workhorse, paired with a servo motor if you need sustained RPM.
- Heavy load, shock, or shaft misalignment — spherical roller bearings are the only sensible answer, even with their higher cost and internal friction.
- Sustained speed above roughly 600 RPM with load attached — a stepper motor will let you down. This is servo territory.
Getting the scenario wrong is where the money disappears. Let me show you how I know.
Scenario A: Radial Ball Bearings for Speed-Focused Lines
Radial ball bearings are the workhorses of factory automation. If you're building an assembly conveyor, a packaging line, or a light pick-and-place axis, these are usually the right call. They run quiet, handle high speeds, and take moderate combined radial-axial loads without drama.
The mistake I made here was over-specifying tolerance classes.
Back in Q1 2022, a vertical lift mechanism on a Bosch Rexroth automation line needed a replacement bearing. Had 48 hours to decide before the scheduled line restart. Normally I'd run a load calculation, compare tolerance classes, verify the speed factor. But with the plant manager standing by, I took the shortcut: higher precision equals better. Paid roughly $620 per bearing for P4 precision components to support a machine that ran at 1,400 RPM on a 40 mm shaft. The application didn't need P4.
ISO 492 defines bearing tolerance classes. P0/P6 for standard industrial motion. P4/P5 for spindle-class work. I should've known the difference. That mis-spec added about $1,800 in unnecessary cost across the full shaft assembly.
For most automation applications, radial ball bearings need:
- Deep groove design for combined loads
- Light or medium series — not extra-heavy unless load calculations say otherwise
- Pressed steel or polyamide cages, unless speeds are extreme
They do not need machined brass cages, exotic lubricants, or the highest precision class. But that's what I ordered. (The original spec was fine, honestly. My "upgrade" created the problem.)
Scenario B: Spherical Roller Bearings for Shock-Loaded Systems
September 2022. A pallet transfer station in a packaging facility started making a noise no engineer wants to hear. Metal grinding. Rhythmic. Growing louder over two shifts. We shut the line, pulled the bearing, and found spalling on the inner ring and ball tracks.
Here's what I'd missed: the pallet stops created shock loads. Every pallet slammed into the stop, transmitting impact through the shaft into the bearing. Radial ball bearings don't like repeated impact loads. The steady-state load looked fine on paper. The dynamic reality was different.
We replaced it with a spherical roller bearing. The difference made sense immediately:
- Higher radial load capacity — two rows of rollers spread the load over a far larger contact area than a single row of balls
- Self-aligning capability — compensates for shaft deflection and housing misalignment, both of which existed in that old machine frame
- Better shock resistance — line-to-line roller contact handles sudden load spikes far better than point contact of balls
The tradeoff? More internal friction. Higher operating temperature. Not a high-speed bearing, and you won't win any efficiency awards with them. But for heavy, shock-loaded applications, the choice is clear.
The replacement ran 18 months with zero unplanned downtime. The radial ball bearing that failed cost $38. The spherical roller bearing cost $94. Line downtime was $4,200/hour. Do the math.
Even after approving the replacement, I kept second-guessing. What if the extra friction slowed the cycle? I watched the first day's production run like a hawk. It didn't slow down. The noise didn't come back.
One more note: this station had a Bosch Rexroth ball screw assembly driving the lift. Misalignment at the bearing support translates directly to misalignment at the screw — which destroys screw accuracy and life. The spherical roller bearing's self-aligning ability saved that screw from an early grave.
Scenario C: How Fast Can a Stepper Motor Actually Turn?
Short answer: depends on what you mean by "turn." A stepper will spin far faster empty than it will under load — and the "max speed" on a spec sheet is usually the empty number.
Here's the thing about stepper motors: torque drops off a cliff as speed increases. A typical NEMA 23 stepper with 1.2 Nm holding torque might deliver roughly:
- ~0.9 Nm at 300 RPM
- ~0.6 Nm at 600 RPM
- ~0.3 Nm at 900 RPM
- Almost nothing usable at 1,500 RPM
Ballpark figures, not universal. Different motors, drivers, and supply voltages shift the curve. But the shape is always the same: flat, then a cliff.
The mistake that still haunts me: in 2024, I sized a NEMA 34 stepper for a linear transfer axis based on the brochure's "maximum 1,200 RPM." With the actual load attached and the axis ramping up, the driver started dropping steps above 700 RPM. The motor wasn't broken. It was out of torque.
We stopped the line and installed a planetary gearbox — 5:1 ratio — which gave the axis the torque it needed at the designed output speed. A correct fix, but a retrofit is never as clean as a correct initial design.
So, the practical answer to "how fast can a stepper motor turn" for automation work:
- Unloaded, many hybrid steppers will spin 2,000–3,000 RPM.
- Under load with useful torque, most are realistically operating between 300 and 800 RPM.
- If your design requires sustained speed above 600 RPM with meaningful torque, you need a servo motor, not a stepper.
Bosch Rexroth factory automation lines typically pair servos with ball screw assemblies when the spec demands both speed and position accuracy. A stepper running near its torque limit is a missed-step disaster waiting to happen.
Which Scenario Are You In? A Decision Guide
You don't have to repeat my tuition payments. Here's how to figure out your scenario before you order anything.
Step 1: Classify your load
Is the load primarily radial, primarily axial, or a heavy combination?
- Radial with moderate axial → radial ball bearing (Scenario A)
- Heavy radial plus axial, with shock or shaft deflection → spherical roller bearing (Scenario B)
- Primarily axial, like a vertical ball screw lift → angular contact or thrust bearing. Don't use either bearing type above for this.
Step 2: Be honest about speed
What's the sustained operating speed — not the peak?
- Below 600 RPM with moderate torque → stepper with a gearbox is acceptable
- 600–1,000 RPM sustained → servo motor territory
- Above 1,000 RPM → servo, and pay close attention to coupling rigidity
Step 3: Confirm the precision class you actually need
- ±0.01 mm or tighter → ball screw + servo, precision angular contact bearings at the fixed end
- ±0.1 mm or looser → belt drive or standard ball screw, radial ball bearing support is plenty
Step 4: Prove it on paper before you order
An hour with the datasheet beats a week of downtime. In the Bosch Rexroth linear motion technology R165139420 ball screw assembly documentation, you can find load charts, axial clearance options, and critical speed curves. The ball screw, the motor, and the bearings are one system. Spec them as a system.
So glad I finally adopted that rule. Almost repeated my stepper mistake on a second axis before I forced myself to walk through the numbers before clicking "order." It caught a 30% speed shortfall in the design phase.
Final Thought
Efficiency in automation isn't just about servo tuning and cycle-time optimization. It's about not designing failures in from the start. The right bearing, the right motor, the right screw — they depend entirely on what your machine actually does. Classify your scenario early, do the math honestly, and you'll avoid most of the pain I documented.
I didn't have this field guide when I started. Now you do.