Gripper finger design that holds on.

The vendors' own ratings show how big the difference is. OnRobot rates the RG2 for 2 kg force-fit and 5 kg form-fit, the 2FG7 for 7 and 11 kg, and the 2FG14 for 14 and 20 kg. Same motor, same jaws. The only thing that changed is whether the fingers wrap the part.

So the first question in any finger design isn't "how hard can it squeeze?" It's "can the part sit in the fingers?" A shaft can drop into a V. A flange can rest on a ledge. A ring can be gripped from the inside so the fingers sit under its lip. A housing can have a pocket milled to its outline. Each of those turns a friction problem into a geometry problem, and geometry doesn't get oily on second shift.

Friction grips still have their place. Flat-sided parts with no feature to hook, parts you must not mark with a hard edge, and families where one finger has to fit many sizes all lean toward friction with a good pad. And some cells need the part to slip on purpose, like a peg that has to self-align as it goes into a chamfered hole.

If you're still picking the gripper itself, start with how to choose a robot gripper and the best robot grippers comparison. This page assumes the gripper is chosen and you're designing what bolts to its jaws.

The vendors agree on the physics and disagree on the bookkeeping, which trips people up. Schunk quotes gripping force as the arithmetic sum of both jaw forces. Its recommended workpiece weight assumes force-fit gripping with μ = 0.1 and a safety factor of 2 against slipping under gravity alone. You can check that against the PGN-plus-P 80, which is rated at 550 N. 550 × 0.1 ÷ (2 × 9.81) gives 2.8 kg, and Schunk lists 2.75 kg.

SMC works per finger instead. Its catalog sizing for two-finger grippers is F = m·g ÷ (2μ) × a, with a margin of a = 4 for the shocks of normal transport. That's where SMC's rule of thumb comes from: pick a per-finger force of 10 to 20 times the workpiece weight. Robotiq writes it as payload W = 2 × F × Cf ÷ Sf, and its example uses a tested friction coefficient of 0.3 for its silicone fingertips on lubricated steel with a safety factor of 2.4.

Three formulas, one idea. Convert everything to the same convention before you compare a Schunk number to an OnRobot number, or you'll be off by a factor of two and not know it.

Acceleration is the part people forget

A part sitting still only needs to beat gravity. A robot that starts a lift at 1 g of acceleration doubles the load for that moment, and a fast swing sideways adds a load the static rating never saw. In a 2021 how-to with Schunk, Assembly magazine notes robot accelerations typically run 0.25 to 3 g. If you don't know yours, ask the robot programmer, or assume 1 g and check later.

The angle is a trade. A sharper V, say 60°, doubles the effective normal force and centers firmly, but small diameter changes move the part a long way in and out of the V, so you burn stroke and the diameter range gets narrow. A 120° V adds only 15%, but takes a wide range of sizes with little stroke. Most shaft-handling fingers we draw land at 90°, which is a sensible middle.

Two practical notes. The Assembly how-to with Schunk points out that V-jaws need extra stroke to open around the part, so check that the open position clears your largest diameter plus the robot's approach error. And put a small flat or relief at the bottom of the V so it never bottoms on the part. Two clean line contacts are what make it center.

Centering in the other direction

A V-groove centers a shaft across its diameter, not along its length. If position along the axis matters, add a shoulder stop in the finger or a locating feature the part seats against. For short cylinders and rings, a three-jaw centric gripper centers in both directions on its own and is often simpler than clever two-jaw fingers.

V-grooves are also the cheapest form of self-alignment for machine tending, where the part has to land in a chuck or vise in a known spot. That job is covered end to end in machine tending cells.

Schunk's PGN-plus-P 80 shows how the numbers work. It's rated at 550 N closing force, 590 N with the grip point right at the jaw, and a maximum finger length of 125 mm with a maximum of 0.7 kg per finger. That length limit applies up to the 6 bar nominal pressure. If you run higher pressure for more force, Schunk says the finger length has to come down in proportion, because the jaw sees the same lever with more force behind it.

Smaller electric grippers have tighter limits. Zimmer's GEP2013 allows 100 mm and just 0.15 kg per finger. Robotiq's 2F-85 asks that custom fingertips stay within 100 mm of the fingertip base in height and width, and Robotiq's 2019 Hand-E manual caps the force at the fingertips at 100 N in any direction.

Robotiq also warns about something specific to its adaptive 2F grippers. Grip above the "equilibrium line" and the fingers can push the part out instead of holding it. If your fingertip design moves the grip point up, check it against that line in the manual.

Aluminum wins because it's a good compromise everywhere. It's light, so it stays under finger weight limits. It's stiff enough that the grip point doesn't move under load. And when the customer changes the part next year, a machinist can recut the pocket in an afternoon. The aluminum figures above are typical values from MatWeb; design to the certified minimums for the stock you buy.

Pads matter as much as the finger. A replaceable pad raises friction, protects a cosmetic surface and wears out instead of the finger. Bolt or dovetail it on. Bonded pads look tidy and then peel off at the worst time, and you can't swap one without pulling the whole finger.

For hot, abrasive or burred parts, put a hardened steel insert at the contact and keep the body aluminum. Castings straight off the shakeout line will chew through an aluminum V in a few weeks, and the first sign is usually a part that stops centering.

Robotiq's guidance for its custom fingertips lines up with all of this. Use light, rigid materials such as aluminum or polycarbonate. Add rubber pads or textured coatings for grip. Locate the fingertips with screws and dowel pins, add recesses to center the part, and avoid sharp edges and extensions that multiply torque.

Printing is the best thing to happen to finger design in years, mostly because it lets you be wrong cheaply. Print a pocket, try it on 20 parts, change it, print again by tomorrow. Robotiq recommends starting custom fingertips this way to check fit and motion before committing to metal, and we do the same.

The material numbers explain where printing stops. Markforged's own datasheet lists Onyx, a chopped-carbon nylon, at 40 MPa tensile yield and a 3.0 GPa flexural modulus. That's roughly a seventh of 6061-T6's yield strength and far less stiffness than aluminum's. Markforged also notes those are typical values from test plaques built to maximize results, not design limits. A printed finger under a steady clamp load will also creep over time, so a pocket that fit on Monday can be loose by the end of the month.

Continuous carbon fiber closes much of the stiffness gap where the fiber runs along the load. It doesn't fix wear at the contact. Our usual answer is a printed body with a bolted aluminum or steel contact insert, or a printed finger that stays printed for parts under a kilogram or so.

Schunk's BSWS system is a good example of what you're buying. It's a manual jaw quick-change between the gripper's base jaw and the finger, released with a hex key, and Schunk specifies a replacement accuracy of ±0.02 mm for a top jaw. The interface has its own load ratings, though. The size 80 adapter allows 63 N·m and 1,080 N, which may be lower than the gripper beneath it, so check both.

The real saving is rarely the minutes spent on screws. It's the re-teach. A finger bolted on with four screws and no dowels lands a few tenths off, someone touches up the pick and place points, and the first parts after changeover are the ones that crash. A repeatable interface removes that whole loop.

If fingers aren't enough, and the cell needs a vacuum cup one hour and a two-finger gripper the next, change the whole tool instead. That's what robot tool changers are for, and they come with their own trade-offs in weight and cost.