A three-link kinematic chain is an assembly of three rigid links connected by joints to transmit motion or force in a defined path. Unlike a four-bar linkage which forms a closed loop with a fixed ground, a three-link chain can be open (serial) like a robotic arm segment, or closed when the ground itself is counted as one of the three links. It exists to constrain motion to a useful trajectory with the fewest possible parts. You see it everywhere — from excavator boom-stick-bucket arms to SCARA robot wrists to the slider-crank in every internal combustion engine.
Three-link Kinematic Chain Interactive Calculator
Vary the number of planar links and lower-pair joints to see the Grübler mobility, constraints, and required independent inputs.
Equation Used
The calculator applies the planar Grubler mobility equation from the worked example. Here n is the total number of links including ground, j1 is the number of lower-pair joints, and M is the number of independent degrees of freedom.
- Planar mechanism with one link counted as ground.
- Links are rigid and joint clearances are ignored.
- All counted joints are lower-pair 1 DOF joints such as revolute or prismatic joints.
- Worked example uses no higher-pair joints.
The Three-link Kinematic Chain in Action
Three rigid links plus the joints between them give you the simplest non-trivial kinematic chain that still does useful work. Each link is a body that holds its shape under load — bending or twisting under operating force breaks the analysis, which is why excavator booms get welded box sections instead of round tube. The joints between the links are where motion happens. Most three-link chains use revolute joints (pin pivots) or prismatic joints (sliders), and each joint contributes 1 degree of freedom in planar motion or up to 3 in spatial motion.
Whether the chain is open or closed changes everything about how you analyse it. An open chain has one end fixed to ground and the other end free — think of a robot arm with shoulder, elbow, and wrist as three links. A closed chain loops back, so the slider-crank in a piston engine is technically a four-link chain when you count the engine block, but the kinematically active links are three: crank, connecting rod, and piston. Mobility — the number of independent inputs needed to fully define position — is calculated using the Grübler-Kutzbach criterion, M = 3(n−1) − 2j1 − j2 for planar linkages, where n is link count and j1, j2 are lower- and higher-pair joint counts.
Get the joint clearances wrong and the chain misbehaves in predictable ways. Pin slop above 0.05 mm in a precision serial manipulator stacks across all three joints — a 0.1 mm radial clearance at the shoulder, elbow, and wrist on a 600 mm arm produces roughly 0.6 mm of end-effector wander before you even apply load. Bushing wear shows up as backlash on direction reversal. Bent links, even slight ones, throw off the kinematic equations because every position calculation assumes link length is exact and constant. The fix is to measure each link centre-to-centre after fabrication and use the measured value, not the drawing value, in your inverse kinematics.
Key Components
- Ground link (Link 1): The reference frame everything else moves relative to. In a serial manipulator this is the base mount; in a slider-crank it's the engine block holding the crankshaft bearing and cylinder bore concentric to within 0.02 mm. Whatever you choose as ground, it must be rigid relative to the working forces — flex here propagates into every downstream measurement.
- Intermediate link (Link 2): The driven or driving link directly connected to ground. In an open chain this is typically the first powered axis (shoulder on a robot arm). Its length and the position of its joint axes set the workspace boundary, so a tolerance of ±0.1 mm on a 400 mm intermediate link is the practical limit for a 6-axis industrial robot like a Fanuc LR Mate.
- Output link (Link 3): The end-effector carrier or output member. Carries the load, the tool, or the slider. In a slider-crank engine this is the piston, constrained to linear motion by the cylinder bore. Mass here matters because it shows up as inertia in dynamic calculations — the lighter Link 3, the higher the achievable cycle rate.
- Lower-pair joints (j₁): Revolute (R) or prismatic (P) joints with surface contact. These constrain 5 of 6 spatial DOF or 2 of 3 planar DOF. Bearing fit class typically H7/g6 for a 20 mm pivot pin gives 0.020-0.041 mm running clearance — enough for free rotation, tight enough to keep positional error inside spec.
- Higher-pair joints (j₂): Cam-follower or gear-tooth contacts that constrain only 1 DOF in plane. Less common in a basic three-link chain, but they appear when one link rolls on another, like a cam-driven follower arm. Hertzian contact stress sets the load limit, and surface finish below 0.4 µm Ra is the practical cutoff for long life.
Industries That Rely on the Three-link Kinematic Chain
Three-link chains show up wherever you need controlled motion with minimum part count. The reason designers reach for them so often is simple — adding a fourth link doubles the joint count and the alignment headaches, while a two-link chain can't generate the path geometries most useful tasks need. Tolerance stack-up is the main thing that bites you in real builds. A robotic shoulder-elbow-wrist with 0.05° angular error at each joint produces roughly 0.5 mm tip error at 600 mm reach, which is fine for material handling but kills any precision assembly task. Get the link rigidity wrong and the planar linkage mobility analysis becomes meaningless because the bodies aren't rigid anymore. The most common failure mode in field service is bushing wear at the most-loaded joint, usually the one nearest ground — that joint sees the cumulative moment of every link beyond it.
- Construction equipment: Boom-stick-bucket excavator front end on a Caterpillar 320 — three hydraulically actuated links forming an open serial chain where each prismatic-equivalent cylinder commands one revolute joint.
- Industrial robotics: SCARA robot inner arm assemblies like the Epson G6 series, where shoulder, elbow, and Z-axis form a 3-link chain with two revolutes plus a prismatic for vertical motion.
- Internal combustion engines: Slider-crank mechanism in every reciprocating engine — Honda K20 and similar — with crankshaft, connecting rod, and piston as the three kinematically active links.
- Aerospace landing gear: Main landing gear retraction linkages on aircraft like the Cessna Caravan use a three-link folding mechanism to swing the gear into the wheel well within a fixed envelope.
- Heavy presses and stamping: Toggle-press drive on a Schuler MSE servo press where the three-link toggle multiplies torque near bottom dead centre to deliver tonnage with a moderate motor size.
- Agricultural machinery: Three-point hitch on a John Deere tractor — top link plus two lift links form a parallel three-link arrangement that lets implements float vertically while resisting fore-aft motion.
The Formula Behind the Three-link Kinematic Chain
The Grübler-Kutzbach criterion is the first thing you compute on any new linkage design — it tells you how many independent actuators you need to fully control the chain. At the low end of the typical range, M = 1 means a single motor or cylinder fully determines every link position (a slider-crank, a basic four-bar). M = 2 is the common sweet spot for three-link planar manipulators that need two independent motions, like a 2-DOF SCARA arm in plan view. At the high end, M = 3 in plane gives you full planar pose control (x, y, θ) and that's where parallel manipulators and 3-RRR robots live. If the formula returns M = 0 the chain is a structure, not a mechanism — useful for trusses, useless for moving things. M < 0 means it's overconstrained and only works because of fabrication slop or elastic deformation.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| M | Mobility — degrees of freedom of the chain, equal to the number of independent inputs needed to control it | dimensionless | dimensionless |
| n | Total number of links including ground | count | count |
| j1 | Number of lower-pair joints (revolute or prismatic) — each removes 2 planar DOF | count | count |
| j2 | Number of higher-pair joints (cam, gear contact) — each removes 1 planar DOF | count | count |
Worked Example: Three-link Kinematic Chain in a 2-DOF planar pick-and-place arm
A bench-top electronics assembly line in Penang is building a 2-DOF planar pick-and-place arm for placing 0603 SMD components onto pre-paste-printed PCBs. The arm uses a three-link kinematic chain — a fixed base, a 250 mm shoulder link, and a 200 mm forearm link — driven by two NEMA 17 stepper motors at the shoulder and elbow. The team needs to confirm the mobility before writing the inverse kinematics code, and they want to understand what happens if they later add a third revolute joint at the wrist for tool orientation.
Given
- n = 3 links (ground + shoulder + forearm)
- j1 = 2 revolute joints (shoulder pin, elbow pin)
- j2 = 0 higher-pair joints
Solution
Step 1 — at the nominal configuration with 3 links and 2 revolute joints, plug into Grübler-Kutzbach:
So the planar arm has 2 degrees of freedom — exactly what a 2-DOF pick-and-place needs. Two stepper motors, two independent inputs, every position in the workspace reachable. This is the sweet spot for x-y placement when tool orientation doesn't matter.
Step 2 — at the low end of the typical range, suppose the team locks the elbow joint with a clamp during a calibration routine:
Locking the elbow turns the arm into a single-DOF rotating beam — the end-effector traces a 450 mm radius circle around the shoulder. Useful for circular wipe tests, useless for x-y placement. This is why locking joints in software still requires a kinematic recompute, not just a motor disable.
Step 3 — at the high end, add a wrist revolute for tool orientation:
Now you have full planar pose control — x, y, and tool angle θ. Three motors, three independent inputs. This is what you upgrade to when component rotation matters, like placing polarised tantalum capacitors or chip-scale LEDs that have a defined orientation pad.
Result
Nominal mobility is M = 2, which matches the two stepper motors and confirms the inverse kinematics needs exactly two joint angles to define any reachable point. Locking the elbow drops mobility to 1 and reduces the workspace to a circle, while adding a wrist revolute pushes mobility to 3 and unlocks orientation control. The sweet spot for SMD placement without rotation is M = 2 — adding the third axis costs roughly 30% more in motor, encoder, and control hardware and is only worth it for orientation-sensitive parts. If your built arm doesn't behave as predicted, the most common causes are: (1) a hidden constraint in the cable management loop acting as a passive joint and reducing effective mobility, (2) a coupling slipping on the elbow motor shaft so commanded angle and actual angle diverge by 1-2°, or (3) frame flex at the shoulder mount under inertial load, which makes the ground link not actually rigid and breaks the n = 3 assumption.
When to Use a Three-link Kinematic Chain and When Not To
Choosing a three-link chain over a four-bar or a parallel mechanism comes down to what trajectory you need, how much workspace you can give it, and how much error stack-up you can tolerate. Three-link serial chains give you the largest workspace per unit footprint, but errors at each joint compound at the tip. Four-bar linkages give you a constrained, repeatable path with only one input, but a fixed shape — you can't reprogram a four-bar to draw a different curve. Parallel three-link mechanisms (delta, 3-RRR) give you stiffness and speed at the cost of a smaller workspace.
| Property | Three-link serial chain | Four-bar linkage | Parallel 3-RRR mechanism |
|---|---|---|---|
| Mobility (planar) | 1-3 DOF, scales with joint count | 1 DOF, fixed | 3 DOF, fully controlled pose |
| Workspace per footprint | Large — reach equals sum of link lengths | Small — coupler curve only | Medium — bounded by link triangulation |
| Tip positional accuracy | ±0.1-0.5 mm typical, errors stack across joints | ±0.02 mm, single-input repeatability | ±0.05 mm, errors average instead of stacking |
| Maximum end-effector speed | 1-2 m/s, limited by link inertia | Path-locked, up to 5 m/s on small four-bars | 5-10 m/s, e.g. ABB FlexPicker delta |
| Load capacity | Moderate — cantilever bending dominates | High — closed-loop force path | High — distributed across three legs |
| Cost per axis | Lowest — one motor per joint | Lowest if 1 DOF is enough | Highest — synchronized multi-axis control required |
| Application fit | Pick-and-place, robot arms, excavators | Engines, suspension, fixed-path mechanisms | High-speed packaging, flight simulators |
Frequently Asked Questions About Three-link Kinematic Chain
Encoder agreement only proves the motor shafts moved as commanded — it tells you nothing about what happened downstream. The most common cause is timing-belt or harmonic-drive compliance between the motor and the link. A standard GT2 belt on a 60 mm pulley has roughly 0.2 mm of effective backlash plus elastic stretch under load, which shows up as 0.1-0.3° of unaccounted joint motion.
Second cause is link flex. A 200 mm aluminium forearm with a 5 mm wall section deflects about 0.4 mm at the tip under a 5 N inertial load during a fast move. Encoders on the motor side never see this. The diagnostic is to mount a dial indicator at the end-effector and command a hard stop — any reading after the motors stop is mechanical compliance, not commanded motion.
Closed chains buy you stiffness and force capacity at the cost of workspace. If your application needs to push hard along a defined path — like a toggle press at bottom dead centre, or an engine slider-crank converting combustion pressure to torque — a closed three-link arrangement gives you a force path that loops back to ground without cantilever bending.
Open chains win when you need a large reachable workspace and reprogrammability. The rule of thumb: if the trajectory is fixed for the life of the machine, close the loop. If the trajectory needs to change in software, leave it open and pay for the stiffness with bigger links.
Grübler-Kutzbach is a counting formula — it doesn't know about geometry. It assumes joints are in general position. When joint axes are parallel, coincident, or share special geometric relationships, the formula undercounts mobility. This is called an overconstrained mechanism, and classic examples include the Bennett linkage and the Sarrus mechanism.
The fix is to use the actual rank of the constraint Jacobian instead of the Grübler count. In practice, if you've built something that moves and the formula says it shouldn't, check whether all your revolute axes are exactly parallel — that's the most common case where geometric coincidence adds a hidden DOF the formula misses.
It depends on reach and on whether errors at each joint are correlated or random. Worst case, errors add directly. A 600 mm reach arm with 0.05 mm radial clearance at each of three joints sees roughly 0.05 mm × 3 = 0.15 mm of tip play in the worst direction, plus another 0.1-0.2 mm from link flex.
Practical limits: for general material handling, 0.5 mm tip play is fine. For PCB assembly with 0603 components on 0.5 mm pitch, you need under 0.1 mm, which means H7/g6 fits on every pin and preloaded crossed-roller bearings at the most-loaded joint. Anything looser and you'll see component placement misregistration on the first reel change.
Three-axis arms have a singularity locus — the elbow-extended configuration where the wrist, elbow, and shoulder become collinear. Near that singularity, small joint motions produce large end-effector motions, and the inverse kinematics becomes ill-conditioned. Most controllers handle this by reducing speed near singularities, which makes the arm feel sluggish exactly where you wanted speed.
The other issue is that the third joint adds another error source. A wrist motor with 0.1° backlash adds roughly 0.1° × tool length of tip error directly. If your tool is 50 mm long that's only 0.09 mm extra, but if it's a 200 mm pipette tip in a lab-automation arm, you've just added 0.35 mm to your error budget. Always include the tool length in the error analysis, not just the link lengths.
Yes, but the count depends on whether you're doing kinematic or dynamic analysis. For trajectory planning you treat each cylinder as a prismatic input commanding the angle of the revolute joint it spans, so the chain is three links plus three inputs and the cylinders disappear into the joint definitions.
For force or stress analysis you have to count the cylinders as real links. A boom-stick-bucket with three cylinders is actually a multi-loop chain with about 7 links and matching joint counts, and Grübler then gives M = 3 — three independent hydraulic flows commanding three independent positions. Both views are correct; pick the one that matches the question you're trying to answer.
References & Further Reading
- Wikipedia contributors. Kinematic chain. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
