Rack Machining Gear Grinding: How Tooth Accuracy Is Restored After Heat Treatment

• 1
  Why a Rack Changes Shape After Heat Treatment
• 2
  What Happens Inside a Rack Machining Gear Grinding Pass
• 3
  Grades, Measurements, and What the Numbers Mean
• 4
  When Rack Machining Gear Grinding Is the Wrong Choice
• 5
  Materials and Setups That Work in Practice
• 6
  Grinding Versus Other Rack Finishing Routes

A rack starts as a bar of steel that is milled or hobbed to a straight tooth profile. The cutting operation leaves a tooth surface that looks correct on a bench inspection. Then the part goes into a furnace for case hardening or through hardening, and the geometry moves.

Three things drive that movement. The first is thermal: heating to 820–930 °C and quenching creates a temperature gradient from surface to core, so the surface contracts and the core resists. The second is phase change: austenite turns into martensite, and martensite takes up more volume, which grows the case faster than the core. The third is stress release, where residual stresses from the earlier machining pass redistribute during tempering.

On a long rack these effects stack up along the length. Pitch error accumulates, and the tooth flank that was straight may now show a lead deviation of 0.02–0.08 mm over a 1,000 mm bar. Helix angle can drift by a few arc minutes. Nobody can predict the exact number before the first heat-treat lot, because it depends on section thickness, quench rate, and how the part was supported in the furnace.

That is the whole reason rack machining gear grinding exists. You cannot hob a hardened rack at 58–62 HRC with a normal cutter. The grinding pass is what brings the tooth back to the drawing after the distortion has already happened.

• 1
  Thermal gradientSurface cools faster than core, so the two shrink at different rates.
• 2
  Martensite growthCase volume expands during the quench, bending a straight bar.
• 3
  Stress releaseTempering lets locked-in machining stress move the part again.

Most racks are ground on a continuous generating or profile grinding machine with a dressable aluminum oxide or CBN wheel. For a rack, the worktable travels along the bar while the wheel rotates and the tooth is generated by the coordinated motion of wheel axis and table axis. A single flank is ground, then the part indexes to the next tooth space.

The wheel is dressed before each batch and re-dressed at intervals set by the material and stock removal. On a typical case-hardened 20MnCr5 rack, depth of cut per pass runs 0.02–0.05 mm, with total stock removal per flank of 0.10–0.25 mm depending on how much distortion the heat treat produced. Feed rates are low, often 5–20 mm/min along the tooth, because the goal is size and form, not metal removal speed.

Cooling matters more than most people expect. Flood coolant at 20–60 L/min keeps the grinding zone below the tempering threshold of the case. If the surface reaches 250–300 °C locally, you get grinding burn, which shows up as a soft blue patch and a tempered martensite layer that will spall later. A burn check with nital etch after grinding catches this before the part ships.

For long racks, the setup is the hard part. A 2,000 mm or 4,000 mm bar needs support along its full length and a table that does not sag or bow. Any error in the machine guideway shows up directly as lead error in the finished rack, so the grinding machine is calibrated against a laser interferometer or a granite straightedge before long work starts.

• 1
  Wheel choiceCBN for high volume, aluminum oxide for flexible small batches.
• 2
  Stock removal0.10–0.25 mm per flank, taken in 0.02–0.05 mm passes.
• 3
  Burn controlFlood coolant and low feed keep the case below its tempering range.

Rack accuracy is usually quoted against a national gear standard. After grinding, a rack can typically reach German standard grade 5 or 6, which corresponds to a total profile deviation of roughly 0.022 mm on a mid-size tooth. Without grinding, a hobbed and heat-treated rack often lands around grade 7 or 8, and the pitch error grows with length.

The parameters that matter are profile deviation, lead deviation, pitch deviation, and cumulative pitch over the full rack. Cumulative pitch is the one that hurts rack and pinion systems, because it adds to the pinion error and shows up as position error at the far end of travel. On a 1,000 mm rack, a 0.02 mm cumulative pitch error translates to a few tens of micrometers of position error depending on the pinion diameter.

Tooth surface finish also moves. Ground flanks typically finish at Ra 0.2–0.8 μm, versus Ra 0.8–1.6 μm for a hobbed flank. That lower roughness reduces friction in the mesh, lowers running temperature, and matters most in high-cycle positioning axes such as a gantry or a machine tool slide.

We measure racks on a gear measuring center for profile and lead, and on a granite bed with a height gauge or laser for cumulative pitch and straightness. A full inspection report is available on request, and every rack is checked 100% before shipment.

• 1
  Profile deviationShape of the flank compared to the theoretical involute.
• 2
  Cumulative pitchThe one that drives position error over a long travel.
• 3
  Surface finishRa 0.2–0.8 μm after grinding cuts mesh friction.

Grinding is not free. It adds a machine setup, a dressing cycle, coolant and wheel cost, and inspection time. If the rack is soft, unhardened, and running in a low-precision application such as a manual gate or a simple adjustment slide, a milled or hobbed rack at grade 8 does the job and costs less.

Small module racks are also hard to justify. Below module 1, the tooth space is narrow, the wheel is thin, and the risk of burning or chipping the tip rises. In that range, a ground rack is possible but the cost per tooth climbs quickly, and a precision-milled rack is often the better trade.

Very long racks present a different limit. Grinding a 4,000 mm bar needs a machine with that stroke and a support scheme that holds the bar straight through the whole pass. Not every shop can do it. Above that length, the usual answer is to grind shorter segments and join them with a dowel and bolt pattern, then re-check cumulative pitch across the joint.

The other case to avoid is grinding a rack with heavy distortion that should have been straightened first. If the bar is bowed by 0.5 mm before grinding, the wheel has to remove uneven stock, and the case depth ends up different on the two flanks. Straighten and stress-relieve, then grind.

• 1
  Soft, low-precision racksA milled rack is enough; grinding adds cost without benefit.
• 2
  Small modulesBelow module 1, burn and tip-chip risk rises fast.
• 3
  Badly bowed barsStraighten and stress-relieve first, otherwise case depth varies.

Case-hardening steels such as 20MnCr5 and 17-4PH (SUS630) grind cleanly once the case is uniform. Through-hardened 4140 and 4340 at 28–32 HRC are also common for racks that need toughness more than wear resistance. Stainless 420 and 440C appear in food and medical equipment, where corrosion resistance matters and the rack may be ground after hardening to hold a tight lead.

Tool steel racks are ground after hardening for stamping and forming dies, where the rack acts as a positioning element. In those cases the grinding pass is often the final operation, and the tooth is ground to size with no further machining.

On the machine side, a rack is usually held on a magnetic chuck or in a fixture with clamps along the length. Clamping force has to be even. A single over-tightened clamp bends the bar, and the ground tooth will spring back when the clamp is released. We check straightness after unclamping, not during.

For racks that pair with a pinion, we prefer to grind the rack and the pinion in the same setup family so the pressure angle and helix match. Mixing a ground rack with an unground pinion is possible, but the contact pattern will sit toward one end of the tooth unless the pair is matched by measurement.

• 1
  Case-hardened steel20MnCr5, 17-4PH: uniform case grinds predictably.
• 2
  Through-hardened steel4140, 4340 at 28–32 HRC for toughness.
• 3
  ClampingEven force along the bar, or the rack springs back after release.