When you buy a ball screw, you're not buying a finished drive system. You're buying one component. It still needs:
End machining to fit the support bearings
A fixed-end support that can handle the axial load
A floating-end support that can handle the speed
A mounting surface that keeps everything aligned
A coupling that doesn't introduce runout
If any one of those is wrong, the screw will let you know. Usually at high speed, and usually in a way that sounds like the screw is the problem.
But a ball screw is a fairly simple mechanical element. If it's straight and the ends are machined correctly, it will run smoothly—provided the supports and mounting are up to the job.
The supports and mounting are where things go wrong.
I think the naming confuses people. "Fixed" and "floating" sound like two versions of the same thing. They're not. They do completely different jobs.
The fixed end carries the axial load. When the nut pushes the table, the force travels back through the screw and into the fixed-end bearings. Those bearings have to handle the full thrust of the axis, in both directions, plus whatever preload is designed into them.
The floating end does almost nothing for load. It just holds the free end of the screw in position so it doesn't sag or whip. It allows the screw to slide axially as it heats up, so thermal expansion doesn't bow the screw.
So when someone tells me their screw vibrates at high speed, I don't look at the fixed end first. I look at the floating end. That's the one that controls vibration, and it's the one people tend to under-spec.
A cheap floating-end bearing with too much internal clearance will let the screw move around. At low speed you won't notice. At high speed, it becomes a vibration source.
Every ball screw has a speed above which it becomes dynamically unstable. Below that speed, it runs smooth. Above it, it whips.
That speed depends on:
The screw diameter
The unsupported length between supports
The support arrangement
Whether the screw is pre-tensioned
How rigid the end supports are
A 16mm screw with 500mm between supports might be fine at 3000 RPM. The same screw with 1200mm between supports might start whipping at 1500 RPM.
This is why I always ask for the unsupported length, not the overall length. The overall length includes the end machining and whatever sticks out past the supports. The unsupported length is what actually matters for critical speed.
If your required RPM is more than about 80% of the calculated critical speed, you're in risky territory. You have a few options at that point:
Go to a larger diameter screw
Go to a larger lead so you need fewer RPM for the same linear speed
Shorten the unsupported length by moving the supports closer
Use a fixed-fixed arrangement with pre-tension
Reduce your target speed
Ignoring this calculation is how people end up with a screw that works fine on the bench and falls apart in the machine.
Here's something that catches people off guard. The end of a ball screw isn't a generic shape. It's machined to match a specific support bearing. That machining includes:
The journal diameter where the bearing sits
The shoulder the bearing seats against
The threaded section for the locknut
Sometimes a keyway or flat
The overall length of the machined section
If you order a screw with end machining for a BK12, and then try to mount it in a BK15 block, it won't fit. The journal will be the wrong size. The shoulder won't seat properly. The locknut won't thread on.
I've had customers order screws and support blocks from different suppliers, assuming the standard dimensions would match. Sometimes they do. Sometimes they don't. When they don't, the customer is stuck with a screw that can't be installed without modification.
The fix is simple: always confirm the end machining drawing against the support block drawing before the screw goes into production. If you're replacing an existing screw, measure the old one and send those dimensions. Don't assume the model number tells the whole story.
I've written about this before, but it's worth repeating because it comes up so often.
If the ball screw isn't parallel to the linear guides, the nut gets pushed sideways as it travels. That creates uneven resistance. The motor current climbs. The screw and nut wear faster. At high speed, it can also cause vibration.
A lot of people blame the screw when this happens. They order a new one, install it, and the problem comes back. Because the screw wasn't the problem. The alignment was.
Here's a simple way to check. Mount the screw and the rails, then run a dial indicator along the screw while moving the carriage from one end to the other. If the needle moves more than a few hundredths of a millimeter, the screw isn't parallel to the guides.
This matters even more when you're replacing a screw in an existing machine. The old screw was aligned to the old rails. A new screw, even if it's dimensionally identical, might not sit in the same position. Always re-check alignment after installation.
This is a detail that gets missed constantly.
The floating end is designed to let the screw slide axially as it heats up. That's how it prevents the screw from bowing when it expands.
But if the bearing is pressed too tightly into its housing, or if there's debris in the fit, it won't slide. Then the screw behaves like it's fixed on both ends. When it heats up, it bows. The preload changes. The friction increases. The vibration starts.
So when you install a floating end support, check that the bearing can actually slide. It should move smoothly but without excessive play. A little grease on the sliding surface helps. And if it feels tight, don't force it. Figure out why it's tight before you bolt everything down.
I've seen this happen on machines that were assembled in a clean room and then moved to a shop floor. The temperature difference was enough to make the screw expand, and because the floating end was stuck, the screw bowed. The customer spent two days troubleshooting the servo before he realized the mechanical problem.
A lot of people treat thermal expansion as something that only matters in high-precision labs. It doesn't. It matters on any machine that runs for hours at a time.
Steel expands by about 11 to 13 micrometers per meter per degree Celsius. On a 1000mm screw, a 5°C temperature rise adds roughly 0.06mm of length. That's enough to change the preload, move the table position, and cause the screw to bow if it's constrained at both ends.
On a precision CNC machine, that's a problem. On a general-purpose router, maybe it's tolerable. But on any machine where the position matters over time, thermal expansion has to be part of the design.
That's why the floating end exists. It lets the screw grow without fighting itself. And that's why you shouldn't just bolt the floating end down and forget about it.
If you're ordering a ball screw with end machining and support blocks, here's what I'd send to the supplier:
Screw diameter and lead
Overall length
Unsupported length between supports
Fixed end and floating end arrangement
Support block model or drawing
End machining drawing
Coupling details
Mounting orientation
Maximum operating speed
Acceleration
Load and cutting force
Duty cycle
That gives the supplier enough to check the critical speed, confirm the end machining, and make sure the support blocks are matched to the application.
If the supplier just takes your order and sends a price without asking about these things, you're buying a part, not a solution. And when something doesn't fit or doesn't run right, you'll be the one figuring it out.
Before you bolt everything down, run through this:
Confirm the end machining matches the support block
Check that the fixed-end bearings are correctly preloaded
Check that the floating-end bearing can slide axially
Verify the screw is parallel to the guide rails
Check the coupling for runout and alignment
Confirm the locknut is tight and the bearing is seated
Rotate the screw by hand and feel for tight spots
Run at low speed first and listen for unusual noise
If something feels wrong at low speed, it's not going to get better at high speed. Stop and figure it out before you run the machine.
A ball screw that vibrates at high speed is almost never a defective screw. It's usually a support problem, an alignment problem, or a critical speed problem.
The screw itself is a fairly simple part. It's straight, it has a lead, and it has ends machined to fit specific bearings. If those bearings are correct, if the alignment is good, and if the speed is within the screw's critical speed range, it will run smoothly.
But if any one of those things is wrong, the screw gets blamed. And ordering a new screw won't fix it.
So before you replace the screw, check the supports. Check the alignment. Check the critical speed. Check that the floating end actually floats.
Those are the things that determine whether the screw runs smoothly or shakes itself apart.
And if you're not sure, send the application details to your supplier. A few minutes of checking before the order is much cheaper than discovering the problem after the machine is assembled.