Choosing a Cobot Gripper: Stroke, Force, Cycle Life

Size the gripper on holding force under acceleration, not on part weight. A 2 kg part in a friction grip with a coefficient of 0.2, accelerated at 5 m/s², needs roughly 170 N of clamping force once a safety factor of 2 is applied, which is well above what many compact electric grippers deliver. Get that number first, then check stroke, mass and cycle life.

The force calculation, done properly

For a friction grip the required clamping force per jaw pair is the part weight plus inertial load, divided by the friction coefficient, multiplied by a safety factor. In symbols: F = m (g + a) / (mu x n) x S, where n is the number of gripping surfaces and S the safety factor.

Worked example. A 2 kg aluminium part, steel jaws with smooth contact so mu is about 0.2, two opposing surfaces, robot acceleration 5 m/s², safety factor 2.

• Weight plus inertia: 2 x (9.81 + 5) = 29.6 N
• Divided by mu x n: 29.6 / (0.2 x 2) = 74 N
• Times safety factor 2: 148 N, rounded up to 170 N for margin on jaw wear

Change one variable and the answer moves sharply. Knurled or urethane jaw faces lift mu to 0.4 or 0.6 and halve the requirement. A form-fit grip, where the jaws capture a feature rather than squeezing a face, removes friction from the equation entirely and is almost always the better engineering answer when part geometry allows it.

Choosing the type before the model

Two observations that decide more projects than the force column. Vacuum fails on porous, textured or perforated surfaces and on parts with release agent still on them, which is why sand-cast and moulded parts so often end up in mechanical jaws. Pneumatic grippers are cheap and strong but need an air supply routed through the dress pack, and that hose is a common source of cable failures on a collaborative arm that rotates its wrist freely.

Stroke and the part family

Stroke is total jaw travel, and it must cover the largest part plus clearance for approach and release, plus any variation across the part family. A common mistake is sizing stroke to the nominal part and discovering that the tolerance band or a second variant needs 4 mm more travel than the gripper has.

Practical allowances: 3 mm to 5 mm clearance per side for approach into a nest, 2 mm for part tolerance, and enough extra to release without dragging on a fixture. A part family spanning 40 mm to 78 mm therefore needs roughly 50 mm of usable stroke, not 38 mm.

The collaborative constraints on top

A gripper on a collaborative arm inherits the contact limits, and it is usually the part of the assembly that fails them. Three rules follow directly.

• Jaw closing is a quasi-static hazard. A finger caught between closing jaws cannot retreat, so the quasi-static limits apply. That is 140 N and 260 N/cm² at the palm, and the closing force of a 400 N gripper is far past it.
• Geometry beats speed. Pressure limits bind before force limits at any small contact area. Rounding jaw edges and increasing contact area is cheaper than derating the arm.
• Loss of power must not drop the part. Mechanically latching or spring-closed grippers hold on power failure. A gripper that opens when the supply drops turns an emergency stop into a falling-object event.

Many collaborative gripper models publish a certified maximum closing force in the region of 120 N to 150 N precisely to stay under the hand limit. If your force calculation demands 300 N, the honest conclusion is that the task is not a contact-permitted task and the cell needs a guard plate over the gripping zone.

Cycle life and what actually wears

Manufacturer cycle life figures assume rated load and clean conditions. What shortens them in practice is side load from misaligned nests, grit in the jaw guides, and jaws that overhang further than the design intends, which multiplies the moment on the guide rails. A gripper rated for 10 million cycles at rated conditions can reach end of life in under 2 million with a long jaw and a misaligned fixture.

At a 12 second cycle on two shifts, roughly 240,000 grips a month accumulate, so a 10 million cycle rating is around 3.5 years. Plan the rebuild rather than discovering it.

Frequently asked questions

Sources