Ball screws, lead screws, threaded spindles and piston rods leave heat treatment with a bow — and they carry the one feature a straightening press must never touch: a precision thread or hardened raceway running almost the whole length of the part. Press the wrong spot and a dent in a ball track or a bruised thread flank turns an expensive component into scrap before finish grinding even starts. This guide explains why slender screws bend, how to support and press them so neither the ram nor the supports ever load the thread, how to measure runout correctly on plain journals, and how an automatic closed-loop machine brings bend down to the ≤0.02 mm class while keeping a data record for every part — without a single press mark on the working surface.
By SHANGDA Engineering TeamReading time: 12 minutesFor: screw makers & machinery buyers
The one rule behind this article: on a screw, the thread is both the functional surface and a forbidden contact surface. Support and measure on plain journals, press only through smooth sections or soft wide contacts, keep threaded zones locked out of the press program — and let measured data, not the operator’s eye, decide every stroke.
The parts in this guide look like ordinary shafts and bend like ordinary shafts, but they cannot be straightened like ordinary shafts. The family includes ball screws — shafts with a helical, hardened and ground raceway through which a nut runs on recirculating balls; lead screws with a trapezoidal or similar sliding thread; other threaded spindles; and the closely related piston rod, which carries no thread but is plated and ground to a seal surface that is just as easy to ruin.
What unites them: a long, slender body; a heat-treated, high-hardness working surface running along most or all of the length; and a tolerance class measured in hundredths of a millimetre. On a plain transmission shaft, almost any smooth spot can serve as a support or a ram contact. On a screw, most of the surface is off limits. A dedicated screw rod straightening machine is designed around exactly that constraint, covering parts from Ø5 mm to 600 mm in diameter and 100 mm to 12 m long (longer screws handled in sections). Everything below explains how the process corrects the axis while the thread stays untouched.
The working surface of a screw is not a bearing journal that can carry a witness mark and be polished out later. It is the surface that transmits motion, and it is hardened precisely because it must resist wear — which also makes it unforgiving under a press.
Ball screw raceway. The helical track is surface-hardened and finish-ground; a row of balls rolls through it under preload on every revolution. A dent or flattened spot from a ram or a support acts like a speed bump: each ball strikes it as it passes, producing noise, torque ripple and fluctuating preload, and the repeated impact seeds fatigue damage. The track cannot be dressed by grinding afterward without destroying its geometry — a bruised ball screw is scrap.
Trapezoidal lead screw flanks. The nut rides the thread flanks along the whole stroke. A bruised or locally deformed flank makes the nut bind at that position, throws the load onto a handful of thread turns, and accelerates wear exactly where the mark sits. A mark that looks cosmetic changes the way the nut fits.
Piston rod seal surface. The plated or hardened outer diameter runs through a seal. Even a shallow score tears the seal lip and leaks; a dent concentrates wear at one point of the guide bushing.
Notice the common mechanism: the functional surface is hard. Under point loading it does not yield smoothly — it dents, bruises or cracks. The same force spread over a soft, wide contact is harmless; concentrated on a hardened ram corner it is permanent. Thread protection in screw straightening is therefore first of all a contact-placement problem, and only secondarily a force-size problem.
A bent screw is rarely anyone’s fault; four mechanisms stack up.
Heat-treatment distortion. Ball screw raceways are commonly hardened locally by induction — only a thin surface layer along the track transforms and grows in volume, leaving the shaft stressed and curved. Lead screws and threaded spindles may be quench-and-tempered or case-hardened; piston rods are hardened for wear resistance. Heating also relaxes the stresses locked in by turning and milling, and the part rebalances as it cools. The distortion physics is the same one that bows camshafts and crankshafts (see our heat-treated automotive shaft guide); screws simply add the thread-protection problem on top.
Extreme slenderness. Screws have a far higher length-to-diameter ratio than most transmission shafts, so their bending stiffness is low. A residual stress moment that a stiff crankshaft would ignore visibly bows a long screw.
Self-weight sag. A long, hot workpiece supported only at its ends sags under its own weight during quenching and tempering, and the sag can freeze in. Even finished screws stored horizontally with poorly spaced supports can take a slow set between centres.
Grinding stresses. Finish grinding leaves surface stresses of its own. When they relax — in storage, in assembly or in service — the part moves. That is why straightening is scheduled after the hardening and grinding stresses are already in the part, rather than before them.
The result is a smooth bow — or an S-curve with several high points on long screws — sitting underneath a perfect thread. The thread itself is fine; the axis beneath it is not.
Straightening is controlled three-point bending: two supports define a span and the ram pushes the high point between them, slightly past straight so that after springback the part lands on target. On a screw, every one of those three contacts has to sit where there is no thread.
Supports rest on plain journals — the end bearing seats, mounting shanks or smooth unthreaded sections — using dedicated V-blocks or support rollers faced with copper, bronze or polymer. Rollers never cradle thread crests; a support on the raceway would bruise it and, worse, its own error would be measured as phantom bend.
The ram presses only on smooth sections. The machine’s part program carries a map of the threaded zones. When the measured high point falls inside a thread, the control places the stroke at the nearest permitted smooth section and adjusts the support span so the three-point geometry still corrects that bend — the curvature comes out without the load path ever crossing the thread.
Contacts are wide and soft. Broad, profiled shoes spread the load over a large patch; nothing on a screw is pressed through a hardened steel corner.
Ball screws are straightened with the nut removed, or with spans arranged so the load path bypasses the nut and its return tubes; the raceway is never clamped or used as a reaction surface.
Piston rods rest on soft rollers positioned against non-seal sections wherever possible, so the plated surface that meets the seal is never put under load.
Supports on journals only · probes on plain diameters only · ram on smooth sections only, with threaded zones locked out in the program · every contact copper-, bronze- or polymer-faced and wide · force capped on every stroke. If a contact would land on a thread, move the span — never pad the thread itself.
A dial probe riding a thread crest reads the thread form itself — the rise and fall of the helix — and cannot separate that from bend. Runout on a screw is therefore measured on the plain diameters: end journals, bearing seats and smooth shanks, with the part rotated between centres or on support rollers placed where the service bearings will sit.
Multi-point mapping. A single probe cannot locate a bow. Short screws are measured at 2–3 points; long screws and rods at 5–8 probes along the axis, so the control builds a deflection curve and knows every high spot’s magnitude and angular direction — the ram always presses opposite the high point.
Measure twice per part. Incoming runout sets the correction plan; outgoing runout proves the result and is stored with the part.
Targets. Precision ball screws typically call for the ≤0.02 mm class; screws for automotive and motor-driven drives the 0.02–0.05 mm class; general rods and trapezoidal screws the 0.10–0.30 mm/m general-engineering range. Always state the value with its basis — which journals, which supports — as our TIR measurement guide recommends; the tolerance classes themselves are compared in our tolerance-101 primer.
Straightening does not alter lead or pitch: the correction is local plastic bending of the axis, and the thread geometry cut by the grinder is unchanged. But a curved axis destroys the thread’s function in service anyway:
The nut is forced to follow a curved path where the drawing promises a straight line, so preload fluctuates along the stroke and the balls share the load unevenly — a few carry everything while others go slack.
The symptoms are noise, torque ripple and heat, occasional binding near the supports, and positioning drift as the nut climbs and dips through a bowed section; the machine’s repeatability suffers even though the screw’s lead measurement reads perfect.
On piston rods the same bend wears seals and guide bushings unevenly and ends in leakage and scoring.
Straightness is what lets the lead accuracy the grinder achieved actually survive inside the machine — which is why screw drawings specify journal runout alongside lead accuracy, and why straightening is a value-adding process step rather than a rework corner.
An automatic screw straightening cell runs the same disciplined sequence for every part:
Measure. Probes map runout at 2–3 points on short screws or 5–8 on long ones, on plain diameters only.
Compute. The control fits the deflection curve, locates each high point, chooses a press point from the permitted smooth zones, and calculates the overstroke — how far past straight to push. Hardened raceway steel springs back hard, and springback shifts between heat lots, so the overstroke is derived from the part’s measured response instead of a fixed setting; the springback compensation guide covers the mechanics.
Press. A force-capped stroke through a soft, wide shoe — the cap is what protects the hardened surface from overload.
Re-measure and repeat until every section is inside target TIR, with the machine refining its estimate of the part’s response as it converges. The full cycle — the measure–press–recheck principle — runs 20–90 seconds per screw, typically 5–10 times faster than manual press-and-dial work.
Each part number gets a stored recipe — support positions, probe points, the thread-zone map, force limits and target TIR — and every part leaves with a logged before/after record: 100% inspection and 100% logged data, by default.
A careful hand with a hydraulic press and a dial gauge can salvage the occasional bent screw. At production volumes the method breaks down exactly where screws are most vulnerable:
The convenient flat spot is a thread. When the high point sits on a threaded zone, the easy move — press right there — is the stroke that dents a raceway or bruises a flank. V-blocks placed wherever the part balances often end up cradling thread crests.
Over-press cracks hardened steel. Hard surfaces tolerate very little plastic strain; one stroke too far opens micro-cracks in the raceway that are invisible to the eye and surface only at crack inspection — or in service. A hand-operated press has no force cap.
Reverse-bend chasing. Without a re-measurement after every stroke, over-correction is noticed late and bent back, working the part back and forth.
No records. Gauge readings that existed only in someone’s head are not a quality record; a screw buyer asking for runout data gets nothing.
Long screws defeat the method. A screw approaching 12 m with an S-curve needs a whole sequence of carefully placed corrections; doing that reliably, part after part, is beyond what even a good operator can sustain.
The closed loop is identical for the whole family; what changes is which surface is protected and where contacts are allowed.
| Part | Surface that must never be loaded | Supports and ram contacts | Watch-out |
|---|---|---|---|
| Ball screw | Hardened, ground helical raceway; nut return tubes | End journals and smooth shanks; nut removed or bypassed; broad soft shoes | Highest hardness — gentlest force-capped strokes; a dented track is scrap |
| Trapezoidal lead screw | Thread flanks the nut rides on | Journals and unthreaded sections; bronze- or polymer-faced contacts | A flank bruise changes nut fit even when the part looks fine |
| Threaded spindle | Precision thread form along the body | Plain diameters only; thread zones locked out of the program | Often very long and slender — expect 5–8 probes and several correction spans |
| Piston rod | Plated / hardened seal surface | Soft rollers on non-seal sections | Sag-prone; a scored seal surface becomes a leak |
Two application-specific configurations cover the family: the ball screw and shaft straightening machine for screws and precision shafts, and the piston rod straightening machine for long plated rods.
For a repair workshop handling one bent screw a week, a careful operator with gauges is adequate. For a screw manufacturer or a production line, the arithmetic points toward automation:
Speed. 20–90 seconds per part versus minutes of rotate–mark–press–recheck by hand — typically 5–10 times the throughput, with one operator tending 1–2 CNC machines or several linked automatic units.
Consistency. The same recipe runs on every screw, shift after shift; no operator-to-operator scatter, and no tacit knowledge that retires when a person leaves.
Evidence. Every part measured before and after, with the record attached to the batch — what buyers of precision screws increasingly expect.
The investment case follows from those three: with steady screw or rod volume, automatic straightening pays back often within 1–2 years with steady volume, through labour saved, rework avoided and parts that ship with data instead of hope.
Whether a screw straightens cleanly to target — without a mark on the thread — is predictable from information you already hold. Send these with your inquiry:
Drawings — overall length, journal diameters and positions, threaded zones versus smooth sections, nut details and weight. The machine range covers Ø5 mm to 600 mm and 100 mm to 12 m.
Material and heat-treatment condition — alloy steel or bearing steel; surface-hardened raceway versus quench-and-temper; which surfaces are hard, plated or uncoated.
Thread specification — ball screw or trapezoidal, size and lead, and whether the nut can be removed for straightening.
Target TIR with its basis — which journals, which supports; precision ball screws usually land in the ≤0.02 mm class.
Annual volume and batch pattern — decides between a standalone CNC machine and an automatic line.
Line conditions — floor space, 380–480 V three-phase power, and any data-export or automation interface needs.
Sample screws settle what drawings cannot: springback is demonstrated on your own metal, and a trial run comes back with before/after TIR data you can verify. Custom screw rod straightening machines typically deliver in 60–120 days.
Not when the process is built for screws. Supports and ram contact only plain journals and smooth sections through copper-, bronze- or polymer-faced shoes; probes measure on plain diameters; threaded zones are locked out of the part program; and every stroke is force-capped. Damage happens when a general-purpose press loads the thread or raceway directly — a support V-block cradling crests, or a ram pressing a high spot that sits on a threaded zone. On a purpose-built screw straightening machine the load path never crosses the thread. Yes. Hardened raceway steel springs back strongly after each stroke and springback varies between heat lots, so the overstroke must be calculated from the part’s measured response rather than set by feel. The safe method is a sequence of light, force-capped strokes with a full re-measurement after each one; pushing a hardened screw too far in a single stroke can open micro-cracks in the raceway. The closed-loop measure–press–recheck cycle exists precisely to converge on target without reaching that point. Never on the thread. A probe riding thread crests reads the rise and fall of the helix itself, not bend. Runout is measured on the plain diameters — end journals, bearing seats and smooth shanks — while the part rotates between centres or on support rollers placed where the service bearings sit. Short screws are mapped at 2–3 points and long screws at 5–8 probes, giving the control a full deflection curve of the axis. No. Lead and pitch are set by thread grinding, and straightening corrects the axis by local bending without stretching the screw or altering the thread geometry. What ruins lead accuracy in service is leaving the bend in: a curved axis forces the nut through a curved path, causing preload fluctuation, uneven ball loading, noise, binding and positioning drift. Straightening protects the accuracy the grinder achieved; it does not disturb it. It is possible for occasional repair work with a very careful, experienced operator, but it is the wrong tool for series production: the nearest press or support point often lands on a thread, hardened steel can crack under an uncapped stroke, no measurement records are produced, and a long S-curved screw needs more carefully placed corrections than hand methods sustain. Automatic cells finish each screw in 20–90 seconds — typically 5–10 times faster — with every part measured and logged. Precision ball screws generally call for the ≤0.02 mm class; screws used in automotive and motor-driven drives typically meet 0.02–0.05 mm; general rods and trapezoidal screws are judged against the 0.10–0.30 mm/m general-engineering range. Whatever the number, write it with its measurement basis — which journals are probed and how the part is supported — so acceptance cannot be disputed.Does straightening damage the ball screw raceway or the thread?
Can a ball screw be straightened after hardening?
Where is TIR measured on a screw — on the thread?
Does straightening change lead accuracy or pitch?
Can a manual hydraulic press straighten long screws?
What TIR should a straightened screw hold?
SHANGDA has built automatic straightening machines since 2008, all CE certified, for screws, rods, shafts, bars and tubes from Ø5 mm to 600 mm and 100 mm to 12 m — including dedicated cells for ball screws, lead screws and piston rods that measure and press only on plain journals and keep the thread untouched. Send your drawings, heat-treatment condition, thread specification and target TIR; our engineers will assess straightenability, propose the support, contact and probe configuration, and prove the result on your sample screws — with before/after runout data for every part.
Send Your Screw Drawings for a Free AssessmentSee Screw Rod Straightening Machines