Linear Guide Accuracy Classes and Preload: Matching Rails to Machine Needs
The rail spec on a drawing says “HIWIN HGH30” and carries a note, deep in the general notes, that reads “preload as required.” That note has shipped on fifty machines and caused forty of them to fall short on positioning or to develop a rumble after a year. The reason is not the rail brand. It is that accuracy class and preload are two independent axes of a decision, and “as required” leaves the decision to the assembly bench, where nobody has the time or the data to make it. This article walks through how to actually set those two values on purpose.
Accuracy Class Is a Geometric Tolerance, Not a Quality Badge
Accuracy class in a linear guide catalogue is a statement about how straight the running surface is and how consistent the rail height is from end to end. It is measured in micrometres over a given length, and every manufacturer publishes a table of these parallelism and height tolerances per class. A higher class is not “better quality,” it is “tighter geometry,” and it costs more for every metre of rail. Choosing C0 or SP when the application is a pallet transfer inside a weldment might buy you a guarantee you never use, while choosing a lower class on a precision inspection axis will cost you in rejection rates.
Read the catalogue numbers for what they actually mean. Class H (high) is a mainstream grade that holds a few hundredths of a millimetre over a metre: perfectly reasonable for most machine slides, pick-and-place axes and feed slides, and dramatically cheaper per metre than a precision grade. Class P and above tighten that to around ten micrometres of height variation and parallelism, which is what a grinding table or a CMM slide needs. The mistake is standardising everything at a precision class “to be safe,” which doubles rail cost and pretends that the mounting surface, machined in the same weldment at ordinary tolerances, can actually deliver the geometry the rail promises. A precision rail bolted to a poorly machined face is a precision rail behaving like a cheap one.
Preload Is About Behaviour Under Load and Reversal
Preload on a linear guide is set by the oversize of the balls relative to the racetrack, chosen during manufacture. Its job is the same as in a ball screw: eliminate internal clearance so that a reversal of load direction never crosses a gap. A guide without preload has a tiny dead band at each reversal, plus a tendency for the carriage to rock on the rails when the load centre is high. Preload also adds stiffness, because a preloaded carriage can be pushed a measurable distance before the internal ball contact starts to deflect.
The trade is heat and drag. Preloaded carriages run warmer, and every extra N of drag torque must be carried by the same servo that drives the axis. The catalogue offers preload codes from light to heavy, expressed either as a clearance/negative-clearance class or as a percentage of the dynamic load capacity. The engineering question is what the axis actually needs: a long, fast, lightly loaded axis wants light preload and low drag; a short, heavily loaded, high-stiffness axis that must hold position wants heavier preload; an axis that does both is exactly where the “matching to needs” part of this article’s title kicks in.
Matching Preload to Duty
A version of the same axis appears in three machines. In the first, a long stroke conveyor transfer, the carriage carries payload for most of the stroke and must simply move smoothly; heavy preload adds nothing but motor load and heat. Light preload, or even zero-clearance class, is correct. In the second, a grinding table where the wheel load pushes down through the carriage and the depth of cut must be repeatable, the stiffness benefit of heavier preload is worth the heat. In the third, a measurement stage where the only load is the probe and the requirement is smooth micro-motion, the axis wants light preload and the best accuracy class the budget allows, because geometric fidelity matters more than load transfer.
The deciding number is the ratio between the load the guide actually carries and its rated capacity. As a guide rule of thumb, keep the working load between about 10 and 30 percent of the dynamic load rating. Below that ratio, a heavy preload is wasted drag; above it, the guide is undersized regardless of preload and will shed life quickly. When this ratio lands in the sweet spot, preload class chosen to match the reversal and stiffness requirement, the axis behaves predictably for years.
Life Calculation With Real Inputs
Guide life uses the same power-law family as rolling element bearings, but the load, rather than being a clean axial number, is a combination of radial, reverse radial, lateral and moment loads acting on each carriage. The catalogue provides formulas for equivalent load, and the honest way to use them is to calculate the equivalent load at each phase of the duty cycle, weight them by time, and feed the mean into the life formula. This is the step most designs skip. A guide ordered on the peak clamping force alone is oversized for 95 percent of its life, while a guide sized on the average alone fails early if the cycle includes significant moment loading at the extremes of stroke.
For a concrete case, take a carriage carrying a vertical load of 3 kN plus a lateral pushing load of 800 N applied at a lever arm that creates a moment around the carriage centre. The moment component can multiply the equivalent load by a factor of two or three. Feed that through the life equation and a guide that looked adequate at first glance falls below the target shift count. Re-spacing the two carriages on the rail so that the moment is shared, or moving the load line closer to the rail centre, often restores the life without a bigger guide. Preload and spacing are the two hands of the same solution.
Stiffness and the Surrounding Structure
The guide does its part, but the stiffness that matters at the tool point is the series stiffness of rail, carriage, mounting bolts, base plate and frame. A rail bolted by M6 socket screws at the catalogue spacing into a 10 mm plate will flex more than the rail itself. The mounting surface should be machined to a flatness that at least matches the promise of the accuracy class, and the screw torque should follow the catalogue recommendation, applied in a sequence from the centre outward in two or three passes. Rails are happiest when their shimming and seat grinding are decided on the drawing, not discovered with a feeler gauge during assembly.
Two rails on the same axis must be aligned to each other within a tolerance in the same league as their class, and the usual practice is to reference one rail as the master and let the second float with a small lateral clearance at some of its bolt holes. Building both rails to dead-parallel perfection is expensive and fragile; a master-slave arrangement gives the same positioning accuracy for less cost and far less assembly pain.
A Selection Checklist Worth Keeping
Work the axis duty cycle into an equivalent load per phase and a mean life. Decide the accuracy class from the geometric tolerance the axis must hold, not from habit. Choose preload from reversal behaviour and stiffness need, then verify the drag budget against the servo. Check moment loading at both ends of stroke, not just at mid-stroke, and re-space carriages to shorten the lever arm. Specify mounting surface flatness and fastener torque on the drawing. Mark one rail as the master datum and float the other. Finally, verify the working load lands in the 10 to 30 percent window of the rated capacity. A guide, like a screw, is a machine component with a duty cycle, not a consumable bolt. When the numbers are done correctly once, the “preload as required” note can finally be struck from the drawing.
How Preload Shows Up on the Bench
Disassembly to check preload is rarely needed if the axis is measured at installation and then again at the first service interval. Two bench tests give fast answers. The first is the drag test: push the carriage by hand along a fresh, wiped rail and feel for a dead band at each reversal point. A carriage with internal clearance clicks or shifts visibly at reversal; a preloaded carriage moves smoothly and instantly changes direction. The second is the rigidity check: mount a dial indicator on the machine frame against the carriage face, apply a modest sideways push with a known force, and read the deflection. Compare the measured deflection to the stiffness the supplier’s data sheet predicted for that preload class. A deflection two or three times the prediction suggests the actual preload is lighter than the specification, and the axis will show the same lost motion at the tool point.
Temperature is a slower but equally honest witness. Run the axis continuously for an hour at operating speed, then measure the rail temperature near the carriage. A healthy preload for a mid-size axis should keep the guide within a few degrees of ambient. If the rail is hot enough to feel distinctly warm through a glove, the preload is heavier than the duty needs, or the mounting has enough friction to burn energy the guide should not absorb. Either way the fix is a conscious choice, not a shrug.
Why Carriage Count and Spacing Beat a Bigger Rail
When an axis flexes at the midpoint, the natural response is to order a heavier rail. Often the cheaper answer is to change the number or spacing of the carriages. A single carriage under a load creates a large moment arm from the rail centre to the load line, and that moment dominates the equivalent load. Two carriages on the same rail share the moment and cut the per-carriage moment load roughly in half. Four carriages on a long slide reduce it further, and because guide life scales as the cube of load, halving the equivalent load can multiply life by a factor near eight.
The practical rule is to lay out the carriage positions to bracket the heaviest load excursions, especially overhang at the ends of stroke. A table with a heavy workpiece that hangs past the end of the travel deserves carriages spaced to keep the load between them for as much of the stroke as possible. This is a geometry exercise done in the CAD model in twenty minutes, and it routinely doubles the calculated life without spending a yuan on a bigger rail.
Accuracy Class Versus Repeatability: Know the Difference
A common confusion in guide specifications is using accuracy class as a proxy for machine repeatability. The guide’s accuracy class describes static geometry: how straight and how constant in height the running surface is. Machine repeatability additionally depends on the ballscrew, the servo’s position loop, thermal drift, and the structural loop between guide and tool. A C0 rail on a machine with a drifting screw and a soft frame will not deliver C0 parts; a Class H rail on a rigid, thermally managed machine can deliver excellent repeatability. The guide is one term in the error budget, and it should be selected within a budget, not in isolation.
Write the error budget before ordering rails: pick a target positioning accuracy for the axis, assign a slice of it to guide geometry, a slice to screw pitch error, a slice to servo following error, and a slice to thermal growth. Only then divide the geometry slice by the stroke length and read the accuracy class off the catalogue. This forces the team to acknowledge that a precision rail cannot compensate for a loose screw or a soft bed.
Lubrication and Serviceability as a Selection Input
Preload and accuracy class are decided at purchase, but the machine is serviced on its knees. A guide that is awkward to grease will be under-lubricated and will fail early regardless of class. Choose fittings that are reachable from outside the machine, prefer automatic or central lubrication where the cycle count is high, and specify the grease type and interval on the maintenance sheet, not in the agents’ collective memory. Rails running dry make the preload feel erroneous, produce a squeal at reversal, and are the most common cause of an otherwise correct axis falling out of tolerance within months.
Corrosion protection matters too. Welding splatter, coolant mist and grinding dust attack the raceway and the carriage seals. If the machine environment is dirty, specify shielded carriages with wipers and keep the guide out of the direct splash line in the layout. A guide that survives ten years in a clean, lubricated environment is the same guide that fails in eighteen months in a dirty one.
The Note That Should Replace “Preload as Required”
Instead of leaving the decision to the bench, the drawing should carry a functionally meaningful note. Something like: “Accuracy class H; medium preload; master rail datum marked; second rail floating at last two bolt holes; working load limit 3.2 kN per carriage; grease NLGI 2 lithium every 2,000 cycles via port P1.” That one line turns a grey decision into a black box with a specification, gives the assembly shop a target to verify, and gives maintenance a service interval that keeps the axis honest. The difference between a guide chosen this way and a guide ordered “as required” is exactly the difference between a machine that drifts after a year and a machine that holds its envelope for the life of the unit.
