CNC machining of motor housings

• 1
  Key takeaways
• 2
  What a motor housing actually has to do
• 3
  Why the bore, not the outside, controls the result
• 4
  How 5-axis machining changes the setup count
• 5
  When milling is the wrong choice
• 6
  How to tell a good housing from a drawing that passed
• 7
  Step by step: from billet to a checked housing
• 8
  Surface finish targets for housing features

A motor housing is a locating device first and a cover second. The stator sits in the bore. The rotor spins inside the stator on a shaft supported by bearings at one or both ends. The gap between rotor and stator is often 0.3 to 0.8 mm, and it has to stay even all the way around. If the bore is 0.05 mm oval, that gap closes on one side and opens on the other. Efficiency drops, the motor runs hotter, and on a high-speed design the rotor may touch the stator.

The housing also carries a stack of tolerances that most drawings show only partly. Bearing bore diameter and roundness, coaxiality between the two bearing seats, perpendicularity of the mounting flange to the bore axis, and position of the bolt pattern. Each one matters, but coaxiality is usually the one that decides whether the motor is quiet or noisy.

Then there is everything that is not geometry. Heat leaves the stator through the housing wall into the frame or the air. Cable entries and connectors need a sealed, deburred path. Mounting feet have to survive vibration without cracking at a fillet. A housing that measures perfectly and rings like a bell at 6,000 rpm is still a failed part.

That is why this page treats CNC machining of motor housings as a geometry problem with thermal and dynamic consequences, not just a matter of hitting a tolerance callout.

• 1
  Bore roundnessTypically held to 0.01 mm or tighter on servo and traction housings.
• 2
  CoaxialityTwo bearing seats on a common axis; 0.02 mm total is a common drawing limit.
• 3
  Flange perpendicularityControls how squarely the motor bolts to a gearbox or pump.
• 4
  Wall stiffnessSets how much the bore distorts after the part leaves the chuck.

Boring a deep, accurate hole in a housing is harder than it looks because the tool is long and the wall is not stiff. A boring bar deflects under cutting force. Push it too hard and the hole comes out tapered, larger at the entry than at the bottom. Push it too gently and the tool rubs, which raises surface roughness and burns the edge. The sweet spot depends on material, bar overhang and coolant pressure.

Aluminium 6061 and 6082 cut cleanly at 2,000 to 4,000 rpm with carbide and plenty of coolant. The risk is built-up edge, which leaves a smear on the bore wall and ruins roundness. 7075 is stiffer and machines to a better finish but costs more and is harder to anodize evenly. Cast aluminium ADC12 machines well but can contain porosity that opens up when the bore breaks into a pore.

Steel housings change the picture. 1045 and 4140 need lower surface speed, more rigidity and often a roughening pass followed by a finishing pass with a different insert. Stainless 304 work-hardens if the tool dwells, so feed has to stay high enough to cut rather than rub. Titanium TC4 is worse still: it holds heat at the edge, and a bore that looks fine can measure 0.02 mm out of round because the part grew during cutting and shrank after cooling.

The practical rule is simple. Measure the bore after the part has cooled to room temperature and after all clamps are released. Anything measured hot, or measured while still bolted to the fixture, tells you very little about the part the customer receives.

A conventional route machines a housing in three or four operations: face and rough the bore, flip and finish the opposite face, then a separate operation for the bolt pattern and side ports. Every flip adds a re-clamping error. If each setup contributes 0.02 mm of positional error, the coaxiality stack can reach 0.04 mm before any cutting error is added.

A simultaneous 5-axis center with a trunnion table and a Ø400 mm rotary table can reach the bore, the end face, the mounting pattern, the cable port and the counterbores in one setup. The part stays clamped once. Positional error stops accumulating. On a housing with two bearing seats on opposite ends, this is often the difference between a 0.03 mm coaxiality result and a 0.06 mm one.

The trade-off is programming time and access. A deep bore inside a narrow housing may not clear the spindle nose at the angle you want. Long reach tooling adds deflection. On some geometries a 4-axis horizontal mill with a tombstone fixture is faster and cheaper than a 5-axis part, because the housing can be reached from two sides without repositioning the trunnion.

For prototype and low-volume work we typically run the first article on a 5-axis center, record the as-machined bore and coaxiality numbers, then decide whether production should stay 5-axis or move to a dedicated fixture on a 4-axis machine. The geometry decides, not the machine list.

• 1
  Single setupBest for housings with two coaxial bearing seats.
• 2
  4-axis with fixtureGood for open housings with side ports and one main bore.
• 3
  3-axis plus flipAcceptable when coaxiality callout is looser than 0.05 mm.

Milling from billet makes sense up to a few hundred or a few thousand units, especially when the design is still moving. Tooling cost is zero, lead time is days, and a revision costs a new program rather than a new die. Past that volume, the arithmetic flips. A die-cast or permanent-mould housing with a few critical faces machined will beat a fully milled part on unit cost.

Geometry sets another boundary. A housing with deep internal cooling channels, thin fins or complex internal ribs is often better cast or printed and then machined only where it seats. Trying to mill a 1.5 mm fin 40 mm deep invites chatter and tool breakage, and the finished wall will not be straight.

Size is the third limit. A 4,000 mm maximum processing size covers most industrial and traction motor housings, but a housing that is long and slender relative to its wall thickness will deflect under its own weight and under clamping. In those cases a fabricated and welded housing, machined at the bearing seats only, is more stable and cheaper.

The honest answer is that CNC machining of motor housings is a finishing and locating process. It is excellent at producing accurate bores, flat flanges and correct hole patterns. It is a poor way to produce large hollow shapes with deep internal detail, and no amount of machine time fixes that.

A housing can meet every dimension on the print and still fail in the application. The usual cause is that the print measured the part in a free state, while the application bolts it down and loads it. A flange that is flat within 0.02 mm free can be 0.05 mm out once four bolts pull it against a gearbox face.

So the inspection plan should mirror the assembly. Measure roundness and taper inside the bore at several depths. Check coaxiality between bearing seats by sweeping a mandrel or using a CMM with the part supported the way it will be mounted. Verify perpendicularity of the flange to the bore axis, not to a convenient face.

For low-volume and prototype work we run 100% inspection before shipment and issue reports on request. That covers raw material verification, in-process checks at the boring and pattern stages, and a final dimensional report. For a housing that will go into a safety-related assembly, ask for the bore and coaxiality numbers specifically. Those are the two that predict field behaviour.

GreatLight has run this route since 2011 across three wholly-owned plants and 7,600 m² of floor space, with 127 high-precision CNC machines including 16 simultaneous 5-axis centers. Housings are one of the part families where the 5-axis capacity earns its keep.