Double Grinding Head Vertical Axis Rotary Table: How It Actually Cuts

• 1
  What the double grinding head vertical axis rotary table actually does
• 2
  Where the process window opens and where it closes
• 3
  Fixturing, clamping and the errors they introduce
• 4
  How to prove the two faces are actually parallel
• 5
  Double head vertical machine against the alternatives

Two wheel spindles face each other on this machine. The part sits on a rotary table between them. On the vertical axis version, the wheel spindles stand above the table and the table rotates about a vertical centerline. Each pass brings the part through wheel one, then wheel two, without unclamping.

That arrangement matters because most flat parts need two parallel faces. On a single-spindle surface grinder you grind face A, flip the part, indicate it in again, and grind face B. Every flip adds stack-up error and consumes floor time. A double grinding head removes the flip entirely. Both faces are cut from the same table datum, so parallelism comes from the machine geometry rather than from your setup skill.

The rotary table is what makes the layout work on round and sector-shaped parts. The table indexes the workpiece past each wheel in sequence, and on many builds it also indexes to a second station for load and unload. That means the operator loads while the previous part is still being ground. Cycle time drops, but so does the floor space you need for a second machine.

Vertical axis machines differ from horizontal double-disc grinders in one important way. The wheels cut the top and bottom faces while the part is held flat on the table, so gravity and clamping both help locate the part. Thin parts stay flatter than they would on a through-feed horizontal machine, where the part is pushed between two wheels and has nothing under it.

The window is widest for parts with two flat, roughly parallel faces and a thickness you can measure easily. Washers, spacers, pump plates, valve bodies, brake shims and bearing races all fit. If the part has one flat face and one free-form face, the machine can still grind the flat one, but you lose the reason to buy the double head.

Thickness control is the first limit. The two wheels must leave a finished dimension, and that dimension depends on table height, wheel wear and thermal growth in the spindle. Machines that hold ±0.005 mm usually run a closed-loop in-process gauge or at least a post-process check with wheel compensation. Without feedback, expect drift across a long run.

Wheel choice sets the finish ceiling. A fine vitrified wheel at a light infeed will hold Ra 0.8–1.6 μm on hardened steel. Push the infeed for cycle time and the same wheel leaves Ra 1.6–3.2 μm with visible feed marks. You cannot grind soft aluminium on the same wheel that cuts hardened 4140; the wheel loads and the part smears.

Rotational symmetry helps, but it is not mandatory. The rotary table handles round parts best, then sector and kidney shapes. Long rectangular parts are possible if they fit inside the table swing, yet the wheel contact becomes intermittent and the table sees an unbalanced load. For long flats, a reciprocating table machine is usually the better call.

How you hold the part sets the practical accuracy more often than the machine spec does. Magnetic chucks are the default for ferrous parts. They are fast and they leave the top face clear, but they do not hold thin non-magnetic parts and they can warp a thin disc if the magnetic field pulls it down onto a surface that is not flat.

Vacuum chucks solve the non-magnetic problem and spread the clamping load. They need a clean, flat sealing face and a part with enough area to seal. Small parts with holes near the edge leak and lose grip. For those, a dedicated nest or a potting fixture with low-melt wax is a better answer, even though it adds a clean-up step.

Clamping pressure is the hidden variable. A part clamped hard will spring back after the grind and lose flatness. For thin discs, grind with the lightest hold that still resists the tangential wheel force. Measure after release, not while clamped. If you only inspect on the chuck, you will pass parts that fail on the customer's surface plate.

Dressing belongs in the same conversation. Both wheels need regular dressing to stay open and concentric. A wheel that glazes will burn the part and raise the finish reading. Track dress count against parts ground, and re-dress on a fixed interval rather than waiting for a burn mark to appear.

Parallelism is the number that decides whether the machine did its job. Measure thickness at several points around the part, then compute the spread. A micrometer on a round disc gives you four or eight readings; a flatness gauge on a granite plate gives you the whole face. Do both on the first article.

Flatness and parallelism are not the same thing and they fail for different reasons. A part can be perfectly flat and still be wedged, because one face is tilted relative to the other. Wedging usually traces back to a chuck that is not square to the spindle axis or to a part that was not seated before clamping. Check the chuck first.

Surface finish should be read in the same setup as the geometry. If you measure finish after the part sits overnight, you may be reading the effect of coolant residue rather than the grind. Wipe the face, then measure. Record the wheel, the dress count and the infeed so the next run starts from a known point.

For production, a sampling plan beats a single first-article check. Measure the first part, then every twentieth, and again after any wheel change. If thickness drifts in one direction, the machine needs compensation, not a new setup. Drift is normal; a step change is not.