Component · Actuator
Servo Gripper
A servo, a linkage and two fingers — the cheapest working end effector. What it can actually hold, why it crushes delicate things, and what to change first.
What it is
A servo gripper is the smallest complete end effector: one hobby servo, a linkage that turns the horn’s rotation into two jaws moving together, and two fingers with pads on them. Most are printed; the frame is a single part, the linkage is either a gear pair on the two finger pivots or a scissor, and the whole thing masses 30–60 g including the servo.
It exists because it is the cheapest object in robotics that turns a positioning system into a machine that does something. It also has a very specific set of limits, and every one of them traces back to a single fact: a hobby servo commands an angle and your object cares about a force.
The one number nobody publishes
Every listing quotes the servo’s stall torque. Almost none quote the lever arm — the distance from the servo horn’s centre to where the pad actually touches the object — and that is the half of the equation that decides what the gripper can do:
F_jaw = τ_stall / L| Lever arm | Stall force (SG90) | Continuous (~70%) | Jaw travel per servo step | Opening over a 120° sweep |
|---|---|---|---|---|
| 12 mm | 14.67 N | 10.27 N | 0.209 mm | 25 mm |
| 20 mm | 8.80 N | 6.16 N | 0.349 mm | 42 mm |
| 30 mm | 5.87 N | 4.11 N | 0.524 mm | 63 mm |
| 45 mm | 3.91 N | 2.74 N | 0.785 mm | 94 mm |
| 60 mm | 2.93 N | 2.05 N | 1.047 mm | 126 mm |
A short arm is better on force and on resolution. What it costs is range, and range is set by the widest thing you need to pick up. So the design order is: measure your largest object, pick the lever arm that opens far enough, and accept the force that falls out. If the force is not enough, that is when you go looking for a bigger servo or a different linkage — not before.
Measure L on the gripper you actually have. On an angular gripper it changes with the object width, which is why the same gripper feels strong on a 20 mm part and weak on a 50 mm one.
What it will and will not hold
With an SG90 on a 30 mm arm (4.11 N continuous), rubber-ish pads at μ 0.4, a safety factor of 2 and a brisk lift:
| Object | Holds? | Why |
|---|---|---|
| M6 bolt, 12 g | yes, easily | Light, high μ, indestructible |
| Wooden block, 40 g | yes | Comfortable margin |
| Chicken egg, 58 g | yes — but by stalling | Needs 1.9 N; the servo pins at 4.1 N, under the egg’s 30 N crush limit |
| Empty PET bottle, 20 g | yes, marginal | Smooth surface, low μ, and it dents |
| Paper cup, 8 g | no | Crushes at about 4 N, which is where the servo pins |
| Strawberry, 15 g | no | Crushes at about 3 N |
| 200 g part on a quick move | no | Needs 6.6 N; shorten the arm to 18 mm or slow the move |
Notice the pattern in the failures: the light delicate things fail and the heavy sturdy things do not. That is backwards from intuition and it is the core problem with this component.
Why it crushes delicate things
The gripper’s knob is a position. The object’s limits are forces. The exchange rate between them is the series stiffness of the finger, the pad, the linkage and the object, and with stiff printed parts that rate is brutal:
usable window (mm) = ( min(F_crush, F_servo) − F_slip ) / k| Finger | k | Window on a strawberry | Servo positions that work |
|---|---|---|---|
| Rigid printed | 25 N/mm | 0.108 mm | 0 |
| Foam pad | 4 N/mm | 0.677 mm | 1 |
| Spring in series | 0.8 N/mm | 3.386 mm | 7 |
| Printed flexure | 0.35 N/mm | 7.739 mm | 15 |
A rigid gripper never lands inside its own window. It overshoots and sits pinned at the servo’s force ceiling, which works only because that ceiling happens to be below the crush limit of most sturdy objects — the servo acting as a clutch you did not design. Fit a stronger servo and the clutch moves up, and now it breaks eggs too.
The gripper simulator lets you drag all of these against each other, and servo gripper force, stall and compliance works through the fix.
Building or buying
Print it. The three numbers that matter — lever arm, jaw opening, finger stiffness — are exactly the ones a bought kit fixes at values chosen for a product photo. A printed gripper lets you set the lever arm from your part and swap a flexure finger in ten minutes.
What to get right:
| Decision | Do this |
|---|---|
| Lever arm | Set by the widest object, then live with the force |
| Linkage | A gear pair on the two pivots keeps the jaws symmetric and self-centring. A single-finger design pushes the part sideways as it closes |
| Finger material | PETG or nylon for flexures. PLA creeps under sustained load and takes a set within weeks |
| Pads | TPU or stick-on EVA foam. Roughly doubles μ over bare PLA, and adds compliance for free |
| Compliance | An extension spring between horn and finger is the highest-value 40p in the build |
| Hard stop | So a runaway command cannot coil-bind the spring and put full stall torque through the linkage |
| Feedback | A lever microswitch on one finger, about 50p. This is what turns a demo into a machine |
| Wiring | Servo power on its own rail, common ground, a few hundred µF near the servo |
What it does not do
Worth being explicit, because these are the reasons to reach for something else:
- No force control. You command position and infer force. If your task needs a specified force, you want a load cell or a current-sensing driver, not this.
- No grasp detection on its own. A plain hobby servo tells you nothing. Add a switch.
- Poor on round, tapered or irregular objects. That is force closure, not a force problem, and squeezing harder makes it worse.
- Cannot hold indefinitely. Stall current, heat, and a dead servo. Design the duty cycle.
- Two contacts let the part pivot. Long objects rotate out of the jaws mid-move. Three contacts or a wrapping finger fixes it.
Where it fits
It is the end effector for the 4-DOF arm kit and for the pick-and-place robot. It is driven exactly like any other servo — one PWM channel, or one of sixteen on a PCA9685 once the arm has more joints than the board has timers.
If you are picking flat, smooth objects from above, stop and price a suction cup and a small pump first. It will be cheaper, faster and gentler than anything you can print.
Explore the graph
Used in these builds
Projects, learning paths, and simulators that include the Servo Gripper.
Questions
Servo Gripper FAQ
What is a servo gripper?
A two-finger end effector driven by a single hobby servo, usually through a gear pair or a scissor linkage so both jaws move together. It is the cheapest thing that will pick an object up, it bolts onto the wrist of any hobby arm, and it is what almost every first pick-and-place build uses. The whole assembly is typically a printed frame plus one SG90, and costs a few pounds.
How much can a servo gripper lift?
Jaw force is stall torque divided by the lever arm, so an SG90 at 0.176 N·m with a 30 mm arm gives 5.87 N at stall and about 4.11 N continuously. What that lifts depends on friction: with the standard sizing formula F = m(g + a)·SF/(μ·n), 4.11 N of jaw force at μ 0.4 with a safety factor of 2 and a brisk lift holds roughly 100–120 g. Halve the lever arm and you double the force — and halve how wide the jaws open.
Why does my servo gripper crush things?
Because you command a position and the object cares about a force, and stiff printed fingers convert between them far too abruptly. With rigid fingers the usable window between dropping a strawberry and destroying it is about 0.1 mm of jaw travel, while one servo step moves the jaws around 0.5 mm. There is no command that lands in the window. Put a foam pad, a spring or a printed flexure in series and the window opens in proportion to 1/k.
Do I need feedback on a servo gripper?
Not to grip, but yes to know whether the grip worked. A plain hobby servo is open-loop to you — it holds an angle and tells you nothing. The cheapest useful answer is a lever microswitch on the inside of one finger, about 50p, giving a digital is-there-something-there signal. Current sensing also works and additionally detects a crush, at the cost of a threshold you have to calibrate.
Will holding an object damage the servo?
Yes, if you hold for long. A gripper closed on an object is a stalled servo drawing around 0.5 A with no airflow, and an SG90 gets hot in tens of seconds and dies in minutes of that. Open the jaws as soon as the part is placed, back the command off a step or two after grasping, power the servo from its own supply, and for long holds use an over-centre linkage or a mechanical latch that draws nothing.
Should I buy a kit or print one?
Print one. The geometry that matters — the lever arm, the jaw opening, the finger stiffness — is exactly what a bought kit fixes for you at values chosen for the photo rather than for your part. A printed gripper lets you set the lever arm from the widest object you need and swap a flexure finger in ten minutes. Buy a kit only if you want a metal frame for something heavy.
Further reading