Vacuum Gripper: The Setup That Worked vs The One That Didn’t
A cardboard box handling cell. The robot picks boxes from a conveyor and places them on a pallet. The original gripper dropped a box about once every 20 cycles. The customer couldn’t tolerate it. Here’s the before setup, the after setup, and what changed.
Before: single flat cup, direct ejector, no reservoir
The hardware: One Ø50 mm flat suction cup. A vacuum ejector (Venturi) mounted directly on the cup. No reservoir. Supply pressure 5 bar. The robot picked the box, lifted 100 mm, moved 500 mm horizontally, placed the box.
What happened: About 5% of picks, the cup didn’t seal fully. The box was slightly deformed, or the top flap wasn’t flat. The vacuum dropped to 30% and the box slid off during the horizontal move. The ejector made noise and consumed air continuously.
Root causes:
- A single flat cup on a slightly uneven box has a bad seal. One cup = one seal point. If it doesn’t seal, nothing holds.
- No reservoir means the vacuum builds slowly. The ejector needs 0.5–1 second to pull vacuum. If the robot lifts before the vacuum is at -60 kPa, the box drops.
- Direct-mounted ejector means every cycle blows air. The ejector runs continuously while the gripper is active — 0.3 Nl per cycle, adds up on a 24/7 cell.
After: four bellows cups, reservoir, vacuum switch
The hardware: Four Ø40 mm bellows cups on a manifold. A 0.5 L reservoir between the ejector and the cups. A vacuum switch set at -50 kPa (threshold to confirm grip). The ejector runs only during the pick (PLC-controlled).
What changed:
Multiple cups = multiple seal points. Four cups on the box corners. If one cup doesn’t seal (slightly warped box), the other three hold. The bellows cups compress 5–10 mm, accommodating uneven surfaces. A flat cup sits on the highest point and misses the rest.
Reservoir = instant vacuum. The 0.5 L pre-charged reservoir delivers vacuum the moment the cups touch the box. No waiting for the ejector to build it. The robot lifts after the vacuum switch confirms -50 kPa — about 0.2 seconds, not 1 second.
Vacuum switch = interlock. The PLC doesn’t start the horizontal move until the switch confirms grip. If the box isn’t sealed, the robot waits 0.3 seconds and retries. No more moving with a weak grip.
Ejector off between picks. The ejector runs for 0.3 seconds during pick, then shuts off. The reservoir holds the vacuum through the move. Air consumption dropped by 70%.
The numbers, side by side
| Parameter | Before | After |
|---|---|---|
| Cup type | 1× Ø50 flat | 4× Ø40 bellows |
| Reservoir | None | 0.5 L |
| Grip confirmation | Timer only | Vacuum switch (-50 kPa) |
| Time to vacuum | ~1.0 s | ~0.2 s |
| Drop rate | 5% | <0.1% |
| Air per cycle | ~0.3 Nl | ~0.09 Nl |
The calculation that matters
The holding force of a vacuum cup is:
F = ΔP · A · μ
Where ΔP is the vacuum level (kPa), A is the effective cup area (mm²), and μ is the friction coefficient. For four Ø40 mm cups at -60 kPa: A = 4 × (π × 20²) = 5026 mm². F = 0.06 N/mm² × 5026 × 0.4 (μ on cardboard) = 120 N. The box weighs 2 kg = 20 N. We have 6× safety margin. The single cup was at 30 N holding force on a 20 N box — only 1.5× margin. One leak and it’s gone.
What I’d spec next time
For cardboard, boxes, or uneven surfaces: bellows cups, multiple points, reservoir, vacuum switch. For flat, rigid surfaces (glass sheets, machined plates): a single flat cup with no reservoir is fine — the seal is good and the vacuum builds instantly. The gripper that dropped boxes wasn’t undersized on force. It was undersized on seal redundancy and vacuum response.
The upgrade path: Start with one cup and a timer. If you have drops, add cups, then add a reservoir, then add a vacuum switch. Don’t skip the switch — it’s the cheapest insurance on the gripper.
