Specify the operating condition, not just the part number. That's the sentence I write on more rejection notices than any other. I'm a quality and brand compliance manager at an automation integrator that builds positioning systems around SKF bearings, ball screws, and linear actuators. I review roughly 250 line items every quarter. In 2024, I rejected 12% of first submissions. Most were rejected because the mechanical spec was incomplete, not because SKF made a bad part.

A bearing doesn't fail because of the brand name on the ring. It fails when the load, speed, lubrication, or alignment is outside the design boundary. If you're here for skf bearings as a generic answer, that's understandable. As of March 2025, the better mental model is to start with three questions before the part number: which direction does the load point? Does it move? What happens if the machine stops? What I mean is that the cheapest part number becomes expensive when the operating conditions are not part of the purchase order, because the bearing might be right for the catalogue but wrong for the machine.

Start with load direction, not bearing size

In my experience, bearing selection is a direction problem before it's a size problem. The first question I ask is simple: which way do the forces point? Radial load, axial load, or both? Why does this matter? Because the bearing family that survives radial loads is not always the family that survives axial loads.

  • Deep groove ball bearings: good for radial loads with modest axial loads.
  • Angular contact ball bearings: better for combined radial and axial loads, but the load is directional, so mounting direction matters.
  • SKF thrust bearings: axial-load specialists.
  • Spherical roller bearings: heavy combined loads with shaft misalignment.
  • Spherical plain bearings: slow oscillation, heavy load, misalignment.

That map is simplified by design. I'd rather give you a starting point than a false promise. Every bearing family has limits, and knowing those limits is part of the spec.

SKF thrust bearings are axial-load specialists

It's probably worth pausing on thrust bearings because they get misapplied more often than you'd think. If you search for SKF thrust bearings, you'll find thrust ball bearings, cylindrical roller thrust bearings, tapered roller thrust bearings, and spherical roller thrust bearings. Their shared rule is that they push axial load through the bearing. What they don't do well is handle radial load, unless the surrounding design separates the load paths.

The most frustrating part of reviewing a thrust-bearing spec is seeing a drawing that passes every dimensional check but fails the load-direction check. You'd think written specifications would prevent that, but a note that says axial load often ignores the radial load that the same shaft also needs to transfer. A thrust bearing is not a universal replacement for a radial bearing.

SKF angular contact ball bearings: X, Y, and e are not optional

Now let's address the phrase that probably brought you here: SKF angular contact ball bearings factors X, Y, and e. These letters come from the equivalent dynamic load calculation. When a bearing carries both radial and axial load, you can't simply compare one maximum load rating to one applied load. Instead, you calculate P = X Fr + Y Fa, where Fr is radial load, Fa is axial load, and X and Y are factors.

e is the boundary between two regions. When Fa/Fr is below e, the radial load dominates and the axial term is relatively small. When Fa/Fr crosses e, the axial load becomes significant enough that the Y factor must be applied. The exact values of X, Y, and e change with the bearing series and contact angle. They are not universal constants, so copying a number from a different catalogue is a classic way to create a quietly wrong bearing selection.

As of March 2025, SKF publishes the current X, Y, and e factors in its rolling bearings catalogue and online tools. The equivalent dynamic load method is consistent with ISO 281. Verify the data for the exact bearing series you are specifying, because factors vary by design and contact angle.

SKF's online calculation tools will do the arithmetic, but the output is only as good as the loads you enter. If your machine has a high stall torque, an unbalanced load, or a drive that can reverse direction, put those numbers into the calculation instead of using only the nominal torque.

Spherical bearings: two names, different machines

Another phrase I send back at the review stage is spherical bearings. Which one? A spherical roller bearing and a spherical plain bearing both have spherical geometry, but they are different machines. A spherical roller bearing has rolling elements and handles high radial load plus moderate axial load, with self-aligning ability. A spherical plain bearing is a sliding bearing, often used for slow oscillation, heavy load, and misalignment. If you don't know which family the application needs, you don't have enough information to order.

The way I see it, the vendor who asks what the bearing actually does is more trustworthy than the vendor who immediately quotes a part number. That's a version of professional boundaries: focus on the load case, not the product category.

Ball screws: the specification has to be checkable

Ball screw specifications are where I get picky. A ball screw is not just a threaded shaft; it is a system of lead, nut, preload, accuracy class, and end machining. I use ISO 3408-1 accuracy classes because the phrase high precision is not an engineering specification. If the drawing doesn't state a class, a vendor can supply almost anything and claim compliance. That's not necessarily the vendor's fault; it's the spec's fault for leaving the door open.

In November 2024, I rejected an $18,000 ball screw because the purchase order did not state the accuracy class. The mechanical design passed every other check, but the absence of that one field made the part unverifiable. We fixed the spec, the vendor supplied the correct part, and the project moved forward. The extra cost was not the ball screw; it was the review cycle.

What happens when a linear actuator fails? Usually, the system tells you first.

If you're asking what happens when a linear actuator fails, don't picture only a dramatic crash. I didn't fully understand the value of failure-mode thinking until a quiet stall in Q3 2024 cost us a day of line downtime. The actuator didn't break; the system's assumptions did.

In that event, an actuator stalled under overload. The motor current climbed, the thermal switch tripped, and the load stopped mid-cycle. The holding brake held the load, so nobody was hurt. But because the actuator had not reached its limit switch, the controller didn't know where it was. The line stopped, the alarm log filled up, and maintenance had to mechanically return the actuator to home. The failure was not in the actuator alone; it was in the system design that didn't anticipate a stall before the limit switch.

If you ask me, the most important linear actuator specification is the behavior after a failure signal, not just the load rating. A failure can mean load drift, a jammed screw, a lost feedback signal, or an uncontrolled move if the brake cannot hold the torque. If a load must not drop, specify a holding brake, mechanical lock, or counterweight, or all three. If the actuator must stop precisely on feedback loss, verify that the brake is sized for the worst-case load, not the rated load.

The boundary of this advice

I have mixed feelings about one-stop shop blanket promises. On one hand, simplifying the supply base is helpful. On the other hand, no supplier is equally strong in every engineering discipline. A vendor who says this isn't our strength, here's who does it better earns my trust for everything else. I'd rather work with a specialist who knows limits than a generalist who overpromises.

This article is a starting point, not a design manual. If your application involves human safety, extreme temperatures, or regulatory approval, work with a licensed engineer. The boundary is real: I verify quality; I don't sign off on physics. And the most important quality check is the question before the part number: what is the machine expected to do when the component stops working?