CNC machining vacuum cup: how suction holds a part

• 1
  Key takeaways
• 2
  What a CNC machining vacuum cup actually does
• 3
  Sizing the hold before you cut
• 4
  Which surfaces seal, and which leak
• 5
  Building a vacuum setup that repeats
• 6
  Where a vacuum cup setup stops working
• 7
  Step by step: from bare table to first cut
• 8
  Vacuum cup or mechanical clamp

A vacuum cup is a pad with a closed groove, a sealing cord and a port. When the part lands on the cord, the groove volume becomes a closed cavity. A pump pulls air out of that cavity, and the pressure difference between the shop air above the part and the near-vacuum below it presses the workpiece onto the pad. Nothing sticks to the part. The only thing holding it is atmosphere, spread evenly across the sealed area.

That even spread is the point. A vise or a toe clamp loads a few square millimeters, which dents a 1.5 mm aluminum cover or cracks a ground ceramic face. Suction loads the whole footprint. On a 200 × 300 mm pad at 70 kPa differential, the theoretical hold is roughly 4,200 N. Even after seal losses and safety factors, that is enough for a light finishing pass across the same plate.

The pad does not have to be a flat plate. A CNC machining vacuum cup can be a small round cup on a robot gripper, a grooved plate on a mill table, or a shaped nest that matches a curved casting. The physics does not change. Only the sealed area, the seal geometry and the support under the part change.

One limit shows up immediately. Suction is a normal force, so it resists lift-off and tilting, not sliding. Feed forces still push the part sideways. A vacuum setup works when the cut is light and the part has enough friction and edge support to resist that push.

• 1
  Normal force onlyThe cup pulls the part down; it does not pin it sideways.
• 2
  Closed cavity requiredAny open gap between cord and part drains the pressure difference.

Start with the sealed area, not the pad outline. A 200 × 300 mm pad with a 4 mm cord at the outer edge has a little under 58,000 mm² of working area. At 70 kPa that is about 4,060 N of theoretical hold. Take 50% as a working number for a machined surface with a good cord. That leaves roughly 2,000 N for the cut.

Then compare that to the cutting force. A 12 mm end mill taking a 1 mm radial pass in 6061 aluminum at typical parameters pulls a few hundred newtons. A 50 mm face mill at a heavy depth of cut can pull well over 1,000 N. The same pad is comfortable for one and marginal for the other.

Vacuum level matters less than people expect. Going from 60 kPa to 90 kPa adds 50% to the hold, but it also increases leak-driven pump load. If the surface leaks, more vacuum does not fix the seal, it just makes the pump run harder. Fix the seal first, then raise the level.

Area is the lever you control. Adding a second zone, widening the pad, or moving the cord outward to the part edge usually gains more hold than any pump upgrade. For small parts, a Ø50 mm cup gives about 1,370 N at 70 kPa, which is fine for drilling and light milling but not for roughing a steel block.

• 1
  Use working areaMeasure inside the seal cord, subtract ports and support pins.
• 2
  Keep a 2:1 marginHold force should be at least twice the peak cutting force.
• 3
  Add zones for small partsSeparate circuits stop one leak from killing the whole pad.

A vacuum cup needs a continuous contact line. Machined faces, lapped faces, ground plates and most cast skins seal well. Sandblasted, plasma-cut or heavily textured surfaces leak at the peaks and valleys. As a rule, a face that reads Ra 3.2 μm or better seals with a soft cord; rougher than that, expect the pump to cycle constantly.

Porous materials are the second problem. Die-cast aluminum, sintered bronze and some composites pass air through the body, not around the seal. No cord can close a leak that runs through the part. Coatings or an impregnation step sometimes help, but most shops move these jobs to mechanical clamping.

Part stiffness matters too. A 0.8 mm stainless shim will bow into the grooves under suction and lose contact in the middle. Support pins or a fine grid under the pad keep the part flat. Without them the part deflects, the seal opens, and the cut leaves chatter marks that look like a spindle problem.

Temperature shifts the picture as well. Aluminum grows about 23 μm per meter per °C. A 200 mm plate that warms 10 °C during roughing moves roughly 46 μm across its length. On a ±0.005 mm feature that is enough to matter, so keep coolant steady and check the part after it settles.

• 1
  Ra 3.2 μm is the practical line
• 2
  Porous body, no seal

Pick the cord first. A 3 mm cord in a 2.5 mm groove gives a light squeeze and suits finished faces. A 5 mm cord in a 4 mm groove tolerates rougher stock and higher clamping loads. Harder cord (70 Shore A) resists creep and holds dimensions longer; softer cord (40 Shore A) conforms to slight curvature but flattens over thousands of cycles.

Then place the support. Pins, a lapped grid or a machined nest carry the part so the seal only has to close the gap, not hold the part flat. Keep support height within 0.02 mm of the pad face across the whole area, or the high spots take all the load and the low spots leak.

Route the vacuum in zones. Two or three independent circuits let you hold a long part while cutting near one end. A single circuit on a 500 mm plate means one small leak at the far edge drops the whole setup. Zones also let you release one area to reposition without losing the rest.

Check the setup before the first heavy pass. Set the part, pull vacuum, wait 30 seconds and watch the gauge. A drop of more than 5 kPa in that window means a leak worth finding. Then take a light pass and look at the chip and the gauge together. If the gauge moves with each tooth, the part is lifting and the cut is too aggressive for the hold.

• 1
  Match groove to cordGroove depth around 70% of cord diameter gives a controlled squeeze.
• 2
  Support before sealPins or a nest carry the load; the cord only closes the gap.
• 3
  Zone long partsIndependent circuits limit the damage from one leak.

The first hard limit is cut direction. A vacuum cup resists lift and tilt, but a slotting cutter that pushes the part along the table relies on friction alone. Friction coefficient between a rubber cord and machined aluminum is roughly 0.5, so the side resistance is about half the hold force. Once the feed push approaches that number, the part creeps.

The second limit is part size. Below about 50 × 50 mm the sealed area is too small to carry a useful load, and the cup takes more time to set than a vise. Above roughly 600 mm, seal consistency across the whole pad becomes the problem, and a single low spot leaks enough to matter.

The third limit is cycle time on mixed work. A vacuum setup rewards repetition. If the shop runs five different parts a day, the fixture build time eats the gain. If it runs one part for a week, the same fixture pays back within the first shift.

Finally, thin flexible parts need a plan for release. Breaking a vacuum seal on a 0.5 mm membrane can pull the part out of flat. Vent the cavity slowly, or add a small positive-pressure pulse, so the part comes off without a snap.

• 1
  Side push is the weak axisFriction is about half the hold force on machined aluminum.
• 2
  Small parts lose the mathUnder about 50 × 50 mm, sealed area is too small to help.