A high-speed ball screw that whips is usually caught in the field, not on the drawing. What procurement and engineering should confirm before the deposit, and where Pinsi fits into the selection step.

A customer sent me this last month:

"I need an SFU1605 for a 100 kg table. Will it work?"

I didn't answer right away. Not because I didn't know, but because the question as written can't be answered honestly. A 100 kg table on a horizontal axis with proper linear guides is a very different problem from a 100 kg table on a vertical Z-axis. Same screw. Same weight. Completely different result.

And that's the thing about SFU1605. It's one of the most common ball screws in small CNC and automation, and it's also one of the most misunderstood. People see the model number, look up a load rating, and assume that number is the answer. It isn't.

So let me explain how I actually think about it—not from a catalog, but from machines I've built, fixed, and in a couple of cases, watched fail.

A practical guide to preload, lost motion, and when replacing a ball screw actually makes sense
When an operator reports about 0.05 mm of backlash on the X-axis, the first reaction is often: “Should we replace the ball screw?”

That is usually the wrong first move.

Lost motion at the tool or table does not automatically mean the ball nut is worn out. The error may be coming from the coupling, bearing support, nut housing, linear guide system, or even the machine frame. If the real source is somewhere else, installing a higher-grade screw or a new screw assembly may cost a lot of money and change very little.

Preload is important, but it is only one part of the picture.

A ball nut with the right preload can have very small internal clearance and good axial stiffness. But if preload is increased too much, the machine pays for it through higher friction, more heat, greater motor torque demand, and faster wear.

That is why preload should be treated as a design decision, not as a knob t

A customer sent me a video last month. The screw was spinning on a test stand, and you could hear it—a rising whine that got worse as the RPM climbed. By the time the motor hit around 1800 RPM, the whole assembly was shaking hard enough that he didn't want to run it anymore.

He asked me if the screw was defective.

I asked him three questions instead.

What's the unsupported length? What support blocks are on each end? And how fast are you trying to run it?

His answers were: about 1100mm between supports, BK12 on the fixed end and BF12 on the floating end, and he wanted 2500 RPM so he could hit 25 m/min with a 10mm lead.

That right there told me the screw was probably fine. The problem was that the support arrangement and the operating speed didn't match.

This is one of the most common issues I run into when people order ball screws for CNC machines. The diameter is right. The lead is right. But the screw still doesn't run well, and nobody can figure out why.

It's almost never the

A customer once sent me a message:

“I need a 1610 ball screw for my CNC router. 1000mm travel. Quote please.”

I asked what motor he was using. He said a NEMA 23 stepper, around 3 Nm, with a 36V supply. Then I asked how fast he wanted the machine to move.

He said, “As fast as possible.”

That’s when I knew the 10mm lead was probably a mistake for his setup. Not because 10mm is bad, but because his motor wouldn’t have enough torque to actually use the speed he was imagining. He had picked the larger lead because it looked faster on paper.

That conversation happens a lot. People see “10mm lead = double the speed” and stop the calculation there. But lead changes more than speed. It changes thrust, acceleration, motor RPM, and sometimes the whole feel of the machine.

So if you’re trying to choose between 5mm and 10mm, let me walk through how I actually think about it.

A guy emailed me last week asking for a price on a C5 ground ball screw for his CNC router. He was building a machine to cut wood and plastic, with a target tolerance of maybe plus or minus 0.1 mm. Nothing crazy.

I asked him why he specifically wanted C5.

He said, “Because C5 is better, right?”

That’s a very common way to think. And it’s not completely wrong. C5 does have tighter lead accuracy than C7. But on a wood router with a flexible frame, a handheld router spindle, and a workpiece that moves with humidity, that extra screw accuracy is probably going to disappear into the noise.

I’m not saying C5 is useless. I’m saying it’s only worth paying for when the rest of the machine can actually use it.

So if you’re trying to decide between C5 and C7 for a ball screw, here’s how I think about it—not from a catalog spec sheet, but from the machines I’ve worked on and the mistakes I’ve seen.

I keep seeing the same question on forums and in emails:

“Should I use SFU1605 or SFU2005 for my CNC?”

Usually the person asking already has a machine design in mind. They’ve read that the 20 mm screw is “stronger” or “more rigid,” so they’re leaning toward the bigger one just to be safe.

That instinct isn’t wrong. But it also isn’t the whole story.

SFU1605 and SFU2005 are both 5 mm lead ball screws. The only major difference is the screw diameter—16 mm versus 20 mm. That 4 mm changes how the screw behaves mechanically, but it doesn’t automatically make the bigger screw the right choice.

I’ve worked on small CNC routers, compact linear stages, and a few heavier gantry machines. In almost every case, the final decision came down to what the axis actually had to do, not just which screw looked more substantial on paper.

So instead of giving you a catalog-style comparison, let me walk through how I actually think about this choice.

What to Check Before Replacing an Original Linear Guide Block

When a CNC machine has a worn linear guide, the obvious solution is to order the same part again.

Sometimes that's the best choice.

But not always.

We've had replacement projects where the original guide was available only with a long lead time, or the cost didn't make sense for an older machine. In those cases, a compatible linear guide slider can be a practical alternative.

The problem starts when “compatible” is treated as meaning “same size.”

It isn't.

A guide block can fit the rail and still have different preload, running resistance, accuracy, or load behavior. In a precision machine, those differences can show up after installation rather than when the part is sitting on the workbench.

So if you're sourcing a THK or HIWIN compatible linear guide slider, here's what I would check before placing the order.

What I Check First When a Linear Guide Starts Causing Trouble – Based on Real Shop Floor Calls

I’ll be honest: when a service call comes in about a CNC axis “losing accuracy,” my first reaction isn’t to grab a new linear guide block.

It’s to ask a few boring questions.

Because in my experience, the guide gets blamed for things it didn’t do, and replaced for things it could have kept doing for another year. Meanwhile, the real culprit – misalignment, a dirty wiper, or even a loose bolt – sits there unnoticed.

So if you’re staring at a machine that used to hold ±5 microns and now wanders around like it’s had too much coffee, let’s walk through how I actually diagnose this. No catalogs, no generic checklists – just what I’ve learned from chasing axis problems in real workshops.

I got a call a while back from a machine builder who was convinced his linear guides had the wrong preload. The carriages had felt smooth when they arrived, but after he bolted the rails to the base, one of them felt noticeably tighter.

He wanted to return the guides.

That wasn't the first time I'd seen this. And in most cases, the preload wasn't the real problem.

But preload does matter. It just matters differently from how a lot of people think.

A Practical Look at Load, Preload, Installation, Rigidity, and Long-Term Accuracy A linear guide block rarely gets much attention when a new CNC machine is being designed. The ball screw gets discussed. The servo motor gets sized. The controller and encoder are carefully specified. The guide block? It is often selected by rail width and load rating, then forgotten. That usually works—until it doesn't. A guide that feels perfectly smooth during commissioning can develop higher running resistance, vibration, or positioning errors after thousands of operating hours. In some cases, the guide itself isn't even defective. The problem started with the original selection or installation. After working with linear motion systems, I've found that most guide problems come down to a handful of decisions made before the machine ever runs. Here are seven of the most common ones.

Introduction: Don't Trust That Factory Test Report Too Much

You finish commissioning. The positioning accuracy is dead-on, repeatability looks gorgeous. You figure the job's done.

Six months later, yield starts tanking. You dig in and find the position has wandered by a handful of microns. The maintenance guys tell you the guide sounds rough, and the ballscrew feels a little loose. You stare at the shiny factory test report and wonder what the hell happened.

I've seen this movie way too many times. In automation, initial accuracy is just for the spec sheet. Long-term accuracy is what actually keeps the line running. That dozen or so test cycles the stage did back at the factory tell you nothing about how it'll behave after millions of starts and stops, under real load, with temperature swings. The skill that really matters is making precision stay, not just making it show up. Here's what I've picked up along the way—some of it the painful way.