Every camshaft and crankshaft comes off the heat-treatment line slightly bent. Carburizing, induction hardening, quenching and tempering do not treat a shaft evenly — surface and core heat at different rates, lobes and crank arms cool asymmetrically, and residual stresses from forging and machining rebalance as the metal hardens. The distortion is physics, not bad process control; the question is only whether it exceeds the drawing. On automotive parts it usually does, so straightening is a planned, measured process step, not a rework corner. This guide explains how the two parts distort differently, how they must be measured and pressed without bruising lobes or cracking fillets, and how CNC cells hold the 0.02–0.05 mm TIR class — with a data record for every part.
By SHANGDA Engineering TeamReading time: 13 minutesFor: automotive parts engineers & buyers
The one rule behind this article: you do not straighten a camshaft or crankshaft against a drawing — you straighten it against the way that specific part number bends after that specific heat treatment. Measure the journals, press exactly where the probes say, never press a hardened lobe or fillet, and re-measure after every stroke. Everything else is machinery.
Camshafts and crankshafts leave machining essentially straight; heat treatment changes that. Carburizing, induction hardening, quenching and tempering drive the part through steep temperature gradients and, for hardening processes, through a phase change in the surface layer. Three mechanisms work together:
Uneven heating and cooling. Thin sections — cam lobes, crank webs, flange edges — heat and cool faster than thick journals. Non-uniform thermal contraction pulls the axis out of line as the part quenches.
Phase transformation. When the surface transforms to hard martensite it occupies slightly more volume than the core beneath it. A case that grew on only part of the surface leaves the part stressed — and curved.
Residual stress release. Forging, turning and grinding leave locked-in stresses in the blank. Heating relaxes them; the part rebalances into a new, slightly bent shape as it cools.
The result is predictable in kind but not in magnitude: every heat lot distorts somewhat, and lots differ. Hardening also raises yield strength, so the metal springs back more after each stroke — the reason straightening hardened parts is harder than straightening soft blanks. Our springback compensation guide covers the physics; the practical consequence is that a fixed overstroke that worked on one lot can under- or over-correct the next.
That is why plants treat straightening as a standard post-heat-treatment step, with the part measured before and after. For the measure–press–recheck cycle see how a shaft straightening machine works; for the tolerance itself see the TIR measurement guide. This article is about what makes these two workpieces special.
Both parts rotate in bearings and are ultimately judged by journal runout — but their geometry produces different distortion patterns, measurement traps and correction strategies. A machine and process tuned for one will not necessarily handle the other well.
| Aspect | Camshaft | Crankshaft |
|---|---|---|
| Geometry | Long, slender shaft (sometimes tubular) carrying a row of eccentric cam lobes | Stiff, compact cluster of main journals, pin journals, crank webs and asymmetric counterweights |
| Typical distortion | Smooth overall bow along the axis, sometimes an S-curve; slight lobe-position shift | Small global bow; local angular misalignment between adjacent main journals, throw displacement, flange runout, occasional twist |
| Stiffness & force | Low stiffness — deflects easily; modest press force, many light strokes | High stiffness — higher local force needed, concentrated at the distorted throw |
| Measurement trap | A probe on a lobe reads the lobe's own eccentricity as “runout” — measure on bearing journals (or base circle with orientation known) | Each main journal must be probed separately; counterweight asymmetry and sag must not be read as bend; flange often specified separately |
| Where to press | Shaft body or bearing journals between supports; never on lobes | Near the bent web/throw, supports on adjacent main journals; never across finished pin journals or fillets |
| Main damage risk | Bruised lobes, local buckling of thin/tubular sections, reverse bend | Cracks at fillets and oil holes, bruised finished journals, locked-in stress |
| Automotive TIR target | 0.02–0.05 mm TIR class (selected precision features ≤0.02 mm) | |
The shared headline number is the 0.02–0.05 mm TIR automotive class — tighter than the 0.10–0.30 mm/m general engineering range, and unforgiving of “press it and hope” methods. The next two sections look at each part in turn.
A camshaft is long relative to its diameter, carries a row of eccentric lobes along its length, and on modern engines may be built up or tubular to save weight. That profile drives everything about straightening it:
The distortion is a bow. Heat-treatment stress release curves the slender axis smoothly — a simple arc or, on long shafts, an S-shape with several high points. The lobes themselves rarely bend; the shaft under them does.
Low stiffness cuts both ways. Little force is needed to correct the bow — and little force is needed to over-correct it into a reverse bend or to buckle a thin-walled tubular section locally. Camshaft straightening is a game of many small, measured strokes, not one confident hit.
The eccentricity trap. A probe touching a rotating cam lobe sees the lobe's lift profile — a large, normal eccentricity — and cannot distinguish it from bend. Measurement must reference the ground bearing journals (the surfaces that run in the cam bearings), or the lobe base circle with angular orientation tracked.
Lobes are forbidden press surfaces. Lobes are hardened and ground to a profile that controls valve timing; even a light bruise means scrap or regrind. Press loads go through the shaft body or journals, via soft or profiled contact shoes.
Supports on end bearing journals · probes at main journals (2–3 points on short shafts, 5–8 probes on long ones) · press on shaft body only, never on lobes · many light force-capped strokes · full re-measure after each stroke.
A crankshaft is almost the opposite problem. It is massively stiff in section — several main journals connected to pin journals through crank webs, with counterweights hung asymmetrically along the length. It barely bows as a whole; instead heat treatment moves it in local, angular ways:
Local misalignment between journals. A throw or web shifts slightly, tilting one main journal relative to the next. The error shows up as runout on specific journals rather than a smooth arc.
Flange and end effects. The flywheel flange and the snout often carry their own runout call-outs, because seals and balanced rotating assemblies mount there.
Asymmetry everywhere except the axis. Counterweights make the section uneven in every direction; the part must be supported and measured in a defined angular orientation, and press direction matters — pushing at the wrong clock angle wastes the stroke or introduces twist.
High, carefully placed force. Correcting a bent throw takes real force concentrated near the distorted web, with supports on adjacent main journals — like bending a stiff ladder at one rung. Finished pin journals, fillets and oil-hole regions are never used as press or support points without protection; they are the stress concentrators where cracks start.
Supports on main journals · each main journal probed separately plus flange runout · high-spot angle recorded · press near the bent web through protected contacts, never on fillets or oil holes · higher force per stroke, fewer correction points · every journal re-checked.
TIR (Total Indicator Reading) is the total swing of a probe against a rotating surface over one revolution — bending, out-of-roundness and eccentricity bundled into the number that predicts bearing behavior; the TIR measurement guide covers the method. The application-specific points are these:
Support on machined surfaces only. Camshafts rest on end bearing journals; crankshafts on main journals (or rollers positioned like the engine's bearings). Never support on lobes, counterweights or rough as-heat-treated surfaces — support error becomes a phantom bend.
Multi-point probing. One probe cannot locate a bow. Short shafts are mapped at 2–3 measuring points; long camshafts and crankshafts at 5–8 probes along the axis, each on its own journal, so the control builds a deflection curve.
Measure journals, not lobes. Probes ride camshaft bearing journals; on crankshafts every main journal is probed individually and the flange face checked separately. Lobe lift and counterweight swing must never enter the runout signal.
Record direction as well as magnitude. Each high point is logged with its angular position, because the ram must press opposite the high spot.
Measure twice per part. Incoming TIR sets the correction plan; outgoing TIR proves the result and becomes the part's record. Supports mimic service bearings so the reading means what the drawing means.
Targets: 0.02–0.05 mm TIR for automotive camshafts and crankshafts, selected precision features at ≤0.02 mm; the 0.10–0.30 mm/m range applies to ordinary bars and tubes, not engine parts. State value and method on the drawing — journals, supports and basis — or accept disputes later.
Straightening is controlled three-point bending: two supports define a span, the ram pushes the high point between them, deflected past straight so that after springback it lands on straight (the overstroke principle — see the springback guide and the machine working principle). On engine parts the strategy is part-specific:
Camshafts: ram on the measured high point along the shaft body. Low stiffness means a sequence of light, force-capped strokes walking down from the largest high point; springback on hardened material is large and lot-sensitive, so each overstroke is computed from measured response, not a fixed setting.
Crankshafts: supports on main journals adjacent to the distorted throw, ram loading the crank-web region through profiled soft-contact shoes. Force per stroke is higher, correction points fewer, and the asymmetric part's angular orientation is fixed before pressing.
Force and position monitored together. Position alone cannot tell yield from over-press on hardened steel; force alone cannot hold a dimension. CNC rams track both, and a force cap stops the stroke before the brittle case overloads.
Contact protection. Hardened rams never touch ground journals or lobes directly — bronze, polymer or profiled steel shoes distribute the load.
On soft bar stock, an over-enthusiastic stroke mostly means a reverse bend to chase. On a heat-treated camshaft or crankshaft it can end the part:
Hard surfaces crack. Bend too far and the tensile side opens micro-cracks — typically at cam-lobe edges, crank fillets and oil holes, the natural stress concentrators. They are often invisible to the eye and show up only at magnetic-particle inspection — or in service.
Bruises do not grind out. A ram mark on a ground journal or lobe may be deeper than the remaining finish-grind allowance; the part is scrap.
Locked-in stress moves later. A large over-bend leaves residual stress that can relax during final grinding — “the part moved after we finished it” — or slowly in service.
Thin sections buckle. Tubular or built-up camshaft sections can kink locally under excess force — a deformation no correction stroke reverses.
Hence under-correction with convergence: press a little, measure, press again. Extra measured strokes cost seconds; an over-pressed hardened shaft costs the part. Force capping and re-measurement after every stroke are not refinements on automotive work — they are the process.
The automatic cycle is the same closed loop for both parts: measure all points → compute each high spot's location and required overstroke → press with capped force → re-measure → repeat until all journals are inside target TIR. Springback means several strokes on hardened parts, and the control learns each part's response as it goes.
Camshafts: light corrections walk along the shaft, all points re-checked each cycle because a stroke at one high point redistributes curvature to its neighbors.
Crankshafts: journal-by-journal work — higher force at the bent throws, then a full re-measurement of every main journal and the flange.
Cycle time runs 20–90 seconds per part, depending on length, probes and correction strokes — typically 5–10× faster than manual press-and-dial-gauge work, with no operator-to-operator variation.
Recipes per part number store support positions, probe points, force limits and target TIR; a first-piece check per new heat lot absorbs springback scatter without rewriting the program.
Straightening does not happen in isolation — it sits between heat treatment and finishing, and its neighbors on the line are stress relief and flaw detection:
Where straightening sits. The typical order is hardening/quenching plus the main temper, then straightening in the hardened state, then an optional low-temperature stress-relief or stabilizing temper, then finish grinding. Stress relief relaxes the stresses the press introduced, so the shaft does not move during grinding or in service; because it can shift geometry slightly, some lines re-check runout afterward.
Crack inspection is the safety net. Hardened automotive shafts are routinely checked for surface cracks by magnetic-particle or equivalent non-destructive methods after straightening and finishing. Inspection does not make over-pressing safe — it catches parts where force control failed; a well-run cell aims to produce nothing for NDT to find.
Sequence and records belong together. Runout data, press-force data and inspection results stored against serial or batch number trace not just that a shaft passed, but how it was corrected — the record automotive customers expect.
A skilled operator with a hydraulic press and a dial indicator can straighten a camshaft or crankshaft — shops have done it for decades. At automotive volumes and tolerances, three problems make it the wrong tool:
Consistency depends on a person. The operator estimates overstroke from experience; heat lots, operators and shifts all vary. The 0.02–0.05 mm class leaves little room for judgment scatter, and the best operator's knowledge retires with them.
No data. A dial-gauge reading that existed only in someone's head is not a quality record. Automotive production expects measured before/after values with every delivery.
Throughput. Manual measure–rotate–mark–press–recheck runs to minutes per part; the CNC loop runs 20–90 seconds, and on an automatic line one operator tends 1–2 CNC machines or several linked machines — see the manual vs. automatic comparison.
CNC semi-automatic machines take loading and turning by hand, then measure, compute, press and re-check automatically with force-capped strokes and stored recipes. Fully automatic CNC machines and lines add robotic or gantry loading, automatic orientation, OK/not-OK sorting and a logged record per part — straightening with built-in 100% inspection, which is what traceability means in practice.
Whether a camshaft or crankshaft straightens cleanly to target is predictable from information you already hold. These six items let a builder assess straightenability, propose a configuration and run a trial on your samples:
Drawings — overall length, journal diameters and positions, lobes/throws, flange details, weight. Machines cover Ø5 mm to 600 mm and 100 mm to 12 m (longer parts in sections); a model typically covers one diameter band, e.g. Ø20–120 mm.
Heat-treatment route — carburizing, induction hardening, or quench-and-temper; which surfaces are hard; whether parts arrive straight from the line or after tempering.
Annual volume and batch pattern — sets the automation level (standalone CNC vs. automatic line).
Target TIR — value plus basis: which journals, what supports, which gauge; automotive shafts typically call for 0.02–0.05 mm (see the TIR guide).
Incoming distortion — measured TIR before straightening if you have it, plus lot-to-lot variation; this sizes force and stroke.
Line conditions — floor space, 380–480 V three-phase power, and any data-export or automation interface needs.
Sample parts settle what spreadsheets cannot: springback is demonstrated, not promised, and a trial run produces before/after TIR data you can verify. Custom machines typically deliver in 60–120 days; the buyer's guide walks through the full specification sheet.
Almost always in practice. Hardening processes distort every part to some degree through uneven cooling, phase transformation and residual-stress release; whether straightening is required depends on whether measured runout exceeds the drawing. At the 0.02–0.05 mm TIR class, as-heat-treated parts are usually outside tolerance, so straightening is planned as a standard step rather than treated as rework. Typically the 0.02–0.05 mm TIR class, with selected precision features at ≤0.02 mm. State the number with its measurement method: which journals are measured, how the part is supported, and whether flange runout is separate. A bare value such as “0.05 mm” without that basis causes acceptance disputes — our TIR measurement guide shows how to write it unambiguously. Yes, if over-pressed: micro-cracks start at stress concentrators such as cam-lobe edges, crank fillets and oil holes, and bruises on ground lobes or journals may not clean up in finish grinding. Controlled straightening prevents this by pressing only at measured high points on permitted surfaces, capping force, using soft contact shoes and re-measuring after every light stroke so the part converges on tolerance. Magnetic-particle or equivalent crack inspection after straightening remains the safety net. It is possible at low volumes with a skilled operator, but it is judgment-based: overstroke is estimated by eye, results vary between operators and heat lots, and no stored measurement records are produced. CNC and automatic machines measure, press with force caps, re-check and log every part in a 20–90 second cycle, typically 5–10× faster than manual work. For series production, manual presses are the exception. The typical order is hardening/quenching with the main temper, then straightening in the hardened state, then an optional low-temperature stress-relief or stabilizing temper, then finish grinding — with crack inspection (magnetic-particle or equivalent) after straightening/finishing. Stress relief relaxes press-introduced stresses so the shaft stays straight through grinding; some plants re-check runout afterward. Confirm the sequence with your machine supplier against your heat-treatment route. Drawings (dimensions, journals, lobes/throws, weight), the heat-treatment route (carburizing, induction hardening or quench-and-temper), annual volume and batch pattern, target TIR with its basis, and measured incoming runout if available — plus sample parts for a trial run. Machines cover Ø5–600 mm diameter and 100 mm to 12 m length, run on 380–480 V three-phase power, and custom units typically deliver in 60–120 days.Do camshafts and crankshafts always need straightening after heat treatment?
What TIR should I specify for a camshaft or crankshaft?
Can straightening crack a hardened camshaft or crankshaft?
Can a manual hydraulic press straighten automotive camshafts and crankshafts?
Does straightening come before or after stress relief and crack inspection?
What do I need to send for a straightenability assessment?
Since 2008, SHANGDA has built automatic straightening machines for shafts from Ø5 mm to 600 mm and 100 mm to 12 m — including CNC cells and fully automatic lines for heat-treated automotive parts, all CE certified. Send your drawings, heat-treatment route, annual volume and target TIR; our engineers will assess straightenability, recommend the machine and probe configuration, and prove the result with a trial run on your sample camshafts or crankshafts — with before/after TIR data for every part.
Send Your Drawings for a Free AssessmentSee Automatic CNC Straightening Machines