CSWA study materials
CSWA study resources

Four-Bar Linkage That Jumps Instead of Rotating

CSWA · Put your knowledge to work in context.

All case studies

Four-Bar Linkage That Jumps Instead of Rotating

A test fixture linkage is modeled as an assembly: a fixed ground plate, a 40 mm crank, a 120 mm coupler and a 60 mm rocker, joined by four 6 mm dowel pins in reamed holes. Each joint was mated with Concentric (pin OD to hole ID) plus Coincident (link face to link face). To keep the links flat, the modeler also added a Parallel mate between the crank's front face and the rocker's front face, plus a Distance 0 mate between the coupler and the ground plate. SolidWorks flags two mates in yellow as over-defined. Dragging the crank makes the linkage snap between two positions instead of rotating smoothly, and Physical Dynamics stutters and stops. The modeler's proposed fix is to set the four Concentric mates to 'Free'.

1. Show why the assembly is already over-constrained by the four revolute joints alone, before the Parallel and Distance mates were added.

Read the worked answer

Count the degrees of freedom and the constraints.

  • Free bodies in the loop: 3 moving links (crank, coupler, rocker) × 6 DOF = 18 DOF.
  • Each joint is Concentric + Coincident: concentric removes 4 DOF, coincident removes 1 → 5 DOF removed per joint, leaving the single rotation a pin joint should leave.
  • 4 joints × 5 = 20 constraints against 18 DOF → 2 redundant constraints, before any extra mates.

That redundancy is inherent to a closed loop of revolute joints, not a modelling mistake. A planar four-bar is a 1-DOF mechanism, and the solver is being asked to satisfy 20 equations in 18 unknowns. SolidWorks tolerates a small amount of this and still solves, which is exactly why it feels fine until something tips it over.

The Parallel and Distance 0 mates then add two or three more constraints on top of an already-redundant system. That is what turns a tolerated redundancy into yellow flags and a linkage that cannot find a consistent position - hence the snapping between two solutions.

2. Which mates would you remove, and how do you keep the links coplanar without them?

Read the worked answer

Remove the Parallel and the Distance 0 first. They are the added, non-functional mates and they duplicate what the joints already enforce.

Coplanarity comes free from the existing chain. Each Coincident mate pulls a link face flat against its neighbour's face; pin the chain to the world once and every link inherits the plane:

  • Mate the ground plate's back face coincident with the assembly's Front Plane (one mate), and let the four joints carry that orientation down the chain.
  • If you want an explicit plane check without the redundancy, add a single Parallel on one link - not all four, and not two links against each other.

Then verify the mechanism, not just the mate tree. Drag the crank: it should rotate smoothly through a full revolution or stop at a genuine toggle position, not snap. If flags remain after removing the added mates, use MateXpert to name the specific conflicting entity - it will point at the exact mate that duplicates another, which is faster and safer than guessing.

3. Why is setting the Concentric mates to 'Free' the wrong fix, and what does it actually do?

Read the worked answer

'Free' does not delete a constraint. It tells the solver to stop solving that mate, so the redundancy is hidden rather than resolved.

What that costs you:

  • The joint is no longer a revolute joint. A pin that should be held on its axis is free to slide along it and to tilt, so the linkage has DOF it should not have - the parts will separate or flop out of plane under drag or Physical Dynamics.
  • The assembly looks cleaner in the tree while the model becomes less correct, which is the worst trade in a fixture that is being used to check for interference and stroke.
  • Any downstream drawing, mass property or interference check now reports on a mechanism that does not match the hardware.

The right sequence: delete redundant mates, re-drag to confirm one clean DOF, and only then consider relaxing a mate if MateXpert proves a specific unavoidable redundancy in the closed loop - and if you do, relax the one with the least design meaning and document why. 'Loosen everything until the yellow goes away' converts a solvable over-definition into an unsolvable under-definition.