LONGQIAO Technical Team | CNC Motion Components, Jinan Longqiao Electromechanical Equipment Co., Ltd. | Published: 26 August 2026
Cremalheira e pinhão earns its place when the axis is long enough that a ball screw would need a larger diameter, a support scheme, or both to stay below its critical speed. Below that point, a screw usually gives better stiffness and positioning for less money. The crossover is a calculation, not a fixed number.
A fuso de esferas carries the whole axis load through one rotating shaft. Stretch that shaft and three things move against you at once. Its critical speed falls, so rapid traverse gets capped by whip rather than by the motor. Its rotating inertia climbs, and since inertia rises steeply with root diameter, the usual fix for whip makes the acceleration problem worse. And its axial stiffness varies along the stroke, because the distance between the nut and the fixed bearing block keeps changing.
A rack does none of that. Add another metre and you bolt on another section. Rotating inertia stays exactly what it was, because the only things turning are the pinion, the reducer and the motor rotor. Axial stiffness is set by the pinion mount and the machine frame, not by how far the carriage has travelled from home.
That is the whole argument in one sentence: on a rack axis, length is a cost item; on a screw axis, length is a dynamics problem.
Where does the line sit? It depends on the screw diameter you are willing to carry, the traverse speed you need, whether you can rotate the nut instead of the screw, and whether the frame can support an anti-whip arrangement. Ask your screw supplier for the critical-speed and buckling figures for the exact diameter, lead, end-fixity and unsupported length you are considering. ISO 3408 defines the accuracy classes and the terminology those figures are quoted against, so the numbers you get from two suppliers are comparable. Run the check before you commit to a frame design, because changing your mind afterwards usually means changing the frame.
Most drive comparisons published from a finished-machine perspective stop at “screws are accurate, racks are long.” That is true and not very useful when you are choosing components. Here is the same decision broken into things you can actually specify or measure.
| Dimension | Cremalheira e pinhão | Parafuso de esferas |
| Adding travel | Bolt on another rack section; dynamics unchanged | Longer shaft lowers critical speed and raises inertia |
| Rotating inertia | Fixed: pinion, reducer, rotor only | Grows with length, and steeply with diameter |
| Axial stiffness along stroke | Roughly constant; set by pinion mount and frame | Varies with nut-to-bearing distance |
| Primary backlash source | Tooth flank clearance plus reducer backlash | Nut preload loss, bearing end play |
| Backlash after wear | Recoverable by re-setting mesh or preload | Requires nut replacement or re-preload |
| Lubrication demand | Continuous; tooth flanks are exposed | Sealed circuit into the nut |
| Chip and dust tolerance | Tolerant if lubricated and shielded | Needs bellows or wipers to survive |
| Backdriving | Backdrives readily; vertical use needs a brake | Depends on lead; low-lead screws resist more |
| Cost per added metre | Close to linear | Non-linear once diameter must increase |
| Positioning error correction | Pitch error compensation in the control, from a rotary encoder or a linear scale | Screw lead accuracy graded at manufacture |
| Routine maintenance | Check mesh depth and lubricant flow | Check preload and end bearing condition |
Backlash on a rack drive is a stack, and builders regularly fix the wrong layer of it.
The first layer is tooth flank clearance, which is set by centre distance. Push the pinion deeper into the rack and clearance shrinks. Push it too deep and you get tip interference, heat, noise and rapid flank wear, which gives you more backlash a few hundred hours later than you had on day one.
The second layer is the reducer. A planetary reducer sits between the motor and the pinion, and its own backlash appears at the pinion multiplied by nothing at all, but reflected to the load it is often the largest single contributor on the axis. A precision-ground rack behind a loose reducer is money spent in the wrong place.
The third layer is structural: the pinion mounting plate, the reducer flange bolts, the rack-to-frame joint. Under reversal these deflect elastically, and a dial indicator cannot tell that apart from gear backlash.
The common misconception: helical teeth do not remove backlash. A helical rack raises contact ratio, spreads load over more teeth and runs quieter than a straight-tooth rack at the same module, which is why most CNC gantries use one. Clearance between flanks is still whatever centre distance and tooth thickness make it. If a supplier’s helical rack is described as low backlash, ask which of the three layers above that description refers to.
The edge case worth knowing: on very long gantries, ambient temperature change moves the rack joints. A rack that meshes correctly in a 20 °C workshop in the morning can bind at a joint after a summer afternoon in an uninsulated shop. Set mesh at working temperature, not at first light, and leave the joint scheme to the rack supplier rather than improvising it on site.
Four arrangements are in common use, and they are not interchangeable.
Pick option 4 first and prove it inadequate before paying for option 2. Plenty of gantry routers are shipped with dual-pinion hardware that the cutting process never needed.
This is where a component decision becomes a system decision, and it is the step most often done backwards, by choosing a pinion and then hunting for a reducer that fits it.
LONGQIAO supplies planetary gear reducers alongside rack and pinion sets, which is the practical reason to quote both from one place: the interface question above gets answered once, by one party, before anything ships.
On a vertical axis, the drive backdrives under gravity as soon as torque drops, so the axis needs a holding brake and a control that engages it correctly. That is a solvable problem and an added failure mode.
On short, high-precision axes, the rack contributes error a screw does not have and gains nothing, since a screw of modest diameter has no whip problem over a short stroke.
On axes that must hold position rigidly against a sustained side load, a preloaded ball nut is generally the stiffer answer for the money.
And on machines where cleanliness of the mechanism is part of the specification, an open tooth mesh carrying an oil film is a chip magnet by design.
LONGQIAO does not publish dimensional tables for rack and pinion sets, and quotations are prepared against your axis. To get a usable quote in one round rather than four, send:
Since 2017 the company has supplied full-axis packages rather than single parts, so a rack and pinion enquiry can be quoted together with the guias lineares, the reducer, the drag chain and the lubrication pump that go with it. Where linear guides form part of the same axis, ISO 12090 gives the shared terminology for rolling linear bearing dimensions, which makes cross-supplier substitution simpler later.
A: There is no universal number, and any article that gives you one has skipped the calculation. Take the traverse speed and stroke you need, ask your screw supplier for the critical speed of the candidate diameter and lead at that unsupported length and end fixity, and see whether it clears with margin. If it only clears by growing the screw diameter to the point where inertia hurts your acceleration, you have found your crossover.
A: Helical carries more teeth in contact and runs quieter, which matters on a long gantry moving fast. It also produces an axial thrust component that the pinion bearing and the reducer output bearing have to absorb, so it is not free. On slow, light axes a straight-tooth set is simpler and cheaper.
A: Mechanically yes, if there is frame surface to mount the rack and space behind for the reducer and motor. The part people underestimate is the control side: the travel per revolution changes, the tuning changes, and any pitch error compensation table from the screw is now meaningless and must be rebuilt.
A: No. It removes one layer of the stack. Tooth flank clearance and structural deflection at the pinion mount remain, and on a poorly mounted carrier the structural part alone can exceed everything else on the axis.
A: More often than a screw, and the interval depends on duty and environment rather than a calendar. Set an interval, then verify it by inspecting the flanks: a thin continuous film is correct, dry patches mean the interval is too long, and lubricant thrown across the machine means too much per cycle. Adjust from the evidence.
A: A linear motor removes the mechanical stack entirely and gives higher acceleration, at a considerably higher cost per metre, with a demand for a linear encoder over the full travel and a cooling and contamination plan. Rack and pinion remains the practical choice for the large majority of gantry routers, plasma tables and laser machines.
A: That is set by the rack quality grade, the joint quality if it is a jointed run, the reducer backlash, the frame, and whether the control has pitch error compensation and a feedback source at the load. Quote all of those together. An accuracy figure quoted for the rack alone tells you very little about the finished axis.
A: The motor encoder cannot see anything that happens after the pinion. If your process tolerates the mechanical stack, the encoder plus a compensation table is enough. If it does not, a scale at the load closes the loop around the mesh and the structure, and the axis behaves differently enough that the tuning has to be redone.
