The Goldberg 5R linkage is a spatial overconstrained mechanism with five revolute joints that folds and unfolds on a single degree of freedom. Michael Goldberg derived it in 1943 by combining two Bennett linkages and removing a shared link, extending Bennett's 1903 four-bar discovery into a wider family. It works by enforcing strict relations between link lengths, twist angles, and joint axes so the loop closes through 3D space rather than a plane. Designers use it for rigid foldable structures — deployable booms, origami-inspired panels, and compact stowing mechanisms — where a planar four-bar simply will not fold the way the geometry demands.
Goldberg 5R Linkage Interactive Calculator
Vary link lengths, twist angles, and tolerance to see Bennett-ratio closure error and whether the Goldberg 5R keeps its single moving DOF.
Equation Used
The Bennett compatibility check compares sin(alpha)/a for the two fused Bennett portions of the Goldberg 5R. The target twist is the alpha2 value that makes the ratios equal; error above the tolerance indicates likely binding instead of smooth single-DOF motion.
FIRGELLI Automations - Interactive Mechanism Calculators.
- Goldberg 5R compatibility is checked using the fused Bennett-linkage ratio.
- Twist angles are entered in degrees and converted to radians for calculation.
- A linkage is considered mobile when target twist error is within the specified angular tolerance.
- Equal-length, equal-twist default represents an exact compatible comparison case using the article scale and tolerance.
The Goldberg 5R Linkage in Action
Most engineers are taught Grübler's mobility formula early on, and that formula says a 5R loop in 3D space should be rigid — zero degrees of freedom. The Goldberg 5R defies that prediction. It moves with one DOF because the link lengths, twist angles, and joint offsets satisfy a special set of geometric constraints first worked out by Michael Goldberg in 1943. Break those constraints by even a fraction of a degree on any twist angle and the linkage locks solid. That sensitivity is the central design challenge.
The mechanism is built by taking two Bennett four-bar linkages that share a common link, then merging them so the shared link is removed and a five-link loop remains. Each Bennett four-bar already obeys the Bennett conditions — opposite links must have equal length and equal twist, and the ratio sin(α) / a must be the same for both pairs. When you fuse two compatible Bennett loops, the resulting 5R inherits these constraints in a combined form. The five joint axes do not lie in a plane and they do not meet at a point. They occupy general positions in 3D space, yet the loop still closes through every position of the input crank.
What happens if the tolerances are wrong? The single DOF disappears. A typical failure mode is a 5R that moves freely near its assembly position but binds hard at 30 to 50° of input rotation, because manufacturing error on the twist angles accumulates around the loop. We see builders machine the link bodies to ±0.05 mm and then ignore the angular twist between the joint axes, which is where the real precision needs to live. Twist angles must hold to roughly ±0.1° on a 200 mm-scale linkage if you want smooth full-range motion. Fail that and the linkage acts like a cheap puzzle — it works in one pose and jams in the next.
Key Components
- Five revolute (R) joints: Each joint is a single-axis hinge. Their axes occupy specific skew positions in 3D space — neither parallel, nor coplanar, nor concurrent. Bearing runout must stay under 0.02 mm on a 200 mm linkage to avoid binding.
- Link bodies: Each of the five rigid links carries two joint axes at a defined twist angle α and offset a. The twist angle is the critical spec — it must hold to ±0.1° because Goldberg's compatibility conditions are angular, not linear.
- Common-link interface (vestigial): Two parent Bennett loops shared a link before fusion. The geometry of that removed link still governs the relationship between the two halves of the 5R, which is why the link-length ratios on opposite sides are not independent.
- Input crank: Any one of the five links can serve as the driven member. In deployable hardware we usually drive the link with the largest twist offset because it gives the smoothest torque profile across the fold range.
- Locking or stop feature: Most practical Goldberg 5R hardware uses a hard stop or over-centre latch at full deployment. Without it, the single-DOF loop has no inherent rigidity in the deployed pose — it will fold back under load.
Industries That Rely on the Goldberg 5R Linkage
The Goldberg 5R is not a workhorse mechanism — you will not find it inside a tractor or a sewing machine. It earns its place where you need rigid folding through 3D space with one input, and a planar four-bar cannot get the job done. That usually means aerospace deployables, origami-inspired metamaterials, and academic research hardware where the spatial folding behaviour is the whole point. The constraints that make it hard to manufacture are the same constraints that make it valuable: when it works, you get coordinated 3D motion from a single actuator with no parasitic DOFs to control.
- Aerospace deployables: Used by research teams at Tianjin University and the University of Oxford as building blocks for rigid-foldable solar array substrates and deployable booms tested for CubeSat and small-sat platforms.
- Architectural structures: Chuck Hoberman's deployable structure work and follow-on academic research at Tsinghua University use Goldberg 5R and 6R cells as nodes in expandable canopies and shelter frames.
- Robotics research: Reconfigurable modular robots built at IROS-affiliated labs use Goldberg 5R cells as joint modules where a single motor drives a coordinated multi-axis fold.
- Origami engineering: Brigham Young University's Compliant Mechanisms Research Group has published designs using Goldberg-type overconstrained loops to convert flat-folded sheets into rigid 3D shells.
- Medical devices: Deployable stents and surgical retractors prototyped under origami-inspired medical research programs use Goldberg 5R kinematics to expand a compact stowed shape into a load-bearing 3D structure.
- Educational kinematics: Mechanism teaching kits from groups like the Cornell KMODDL collection demonstrate Goldberg 5R alongside Bennett and Myard linkages as the canonical spatial overconstrained set.
The Formula Behind the Goldberg 5R Linkage
The Goldberg compatibility conditions are what tell you whether a chosen set of link lengths and twist angles will actually move. They are not a performance equation — they are a yes/no gate. At the low end of the design space, with small twist angles around 20°, the 5R folds into a nearly flat package but the deployed stiffness is poor and the input torque needed near the singularity climbs sharply. Around 60° twist (the typical sweet spot for deployable hardware) you get a clean fold ratio of roughly 4:1 stowed-to-deployed and predictable torque. Push twist angles above 80° and the linkage starts to occupy too much volume in the stowed state — you lose the compactness that justified using a 5R in the first place.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| αi | Twist angle between successive joint axes on link i | degrees or radians | degrees |
| ai | Common-perpendicular distance (link length) between successive joint axes on link i | mm | in |
| θi | Joint variable (input/output rotation at joint i) | degrees or radians | degrees |
| di | Joint offset along axis i (zero for a pure Bennett-derived Goldberg 5R) | mm | in |
Worked Example: Goldberg 5R Linkage in a deployable HF antenna mast on a marine research buoy
Your team is building a deployable HF antenna mast for a Wave Glider-class autonomous marine research buoy. Stowed length must fit inside a 250 mm tubular bay, deployed length needs to reach 900 mm above the waterline, and the whole fold must run on a single 12 V gearmotor pulling no more than 2 A. You select a Goldberg 5R cell with three repeat units stacked end-to-end. Each cell uses link length a = 120 mm, and you need to verify the twist-angle relationships and predict deployed length across the practical twist range of 30°, 60°, and 75°.
Given
- a1 = a2 = 120 mm
- α1 (parent Bennett loop A) = 60 degrees
- α2 (parent Bennett loop B) = 60 degrees
- Number of stacked 5R cells = 3 —
- Target deployed length = 900 mm
Solution
Step 1 — verify the Bennett compatibility ratio that the Goldberg 5R inherits from its parent loops at the nominal 60° twist:
Step 2 — compute the deployed length per cell at nominal 60° twist. For a Goldberg 5R built from equal-length Bennett parents, deployed cell length Lcell ≈ 2 × a × sin(α):
Three stacked cells give 3 × 207.8 = 623 mm deployed. That is short of the 900 mm target — you would need to bump link length a to 175 mm or add a fourth cell. Useful number to know before you cut metal.
Step 3 — at the low end of the practical twist range, α = 30°, the cell collapses toward a flatter fold:
Three cells give only 360 mm deployed. The mast looks like it barely extended — visually it is just over a third of target reach, and the deployed stiffness against wave-driven side load is poor because the joint axes sit at shallow angles to the load direction.
Step 4 — at the high end, α = 75°:
Three cells give 695 mm deployed — closer to target, but the stowed envelope grows because the fold no longer collapses neatly into the 250 mm bay. Above roughly 80° twist you also lose the clean 1-DOF behaviour because manufacturing tolerance on α dominates the kinematic budget.
Result
At nominal 60° twist, three stacked Goldberg 5R cells deploy to 623 mm — short of the 900 mm target, telling you immediately to either lengthen the links to 175 mm or add a fourth cell. Across the operating range you see 360 mm at 30° twist (visibly under-deployed, poor stiffness), 623 mm at 60° (the geometric sweet spot for fold ratio and stowed packing), and 695 mm at 75° (better reach but the stowed shape no longer fits the 250 mm bay cleanly). If your built prototype measures, say, 580 mm instead of the predicted 623 mm, the most common causes are: (1) twist-angle machining error on one or more links exceeding ±0.3°, which shortens the effective fold by binding the loop early, (2) joint-axis offset di not held to true zero — even 0.5 mm of unintended axial offset turns a Goldberg 5R into a near-singular linkage that cannot reach full extension, or (3) bearing pre-load too high, applying enough friction at the input crank that the gearmotor stalls before completing the deployment stroke.
Goldberg 5R Linkage vs Alternatives
The Goldberg 5R sits in a narrow band of the design space. It earns its keep against a planar four-bar when you need true 3D folding, and against a Bennett 4R when you need an extra link to route geometry around an obstacle or to add a stacking interface. Compare it on the dimensions practitioners actually search on — fold ratio, manufacturing tolerance, cost, single-DOF guarantee, and application fit.
| Property | Goldberg 5R linkage | Bennett 4R linkage | Planar 4-bar linkage |
|---|---|---|---|
| Degrees of freedom | 1 (overconstrained) | 1 (overconstrained) | 1 |
| Folding behaviour | 3D spatial fold, two-stage compactness | 3D spatial fold, simple symmetric | Planar only |
| Twist-angle tolerance required | ±0.1° at 200 mm scale | ±0.15° at 200 mm scale | Not applicable (planar) |
| Typical fold ratio (deployed:stowed) | 3:1 to 5:1 | 2:1 to 3:1 | 1.5:1 to 2:1 |
| Manufacturing cost (relative) | High — precision twist fixtures needed | Medium-high | Low |
| Best application fit | Stacked deployables, 3D origami cells | Single-stage deployable hinges | General planar mechanisms |
| Risk of binding under tolerance error | High — locks if α off by 0.3°+ | Moderate | Low |
Frequently Asked Questions About Goldberg 5R Linkage
That is the classic signature of accumulated twist-angle error around the loop. The 5R is overconstrained — it only works because Goldberg's compatibility conditions hold exactly. Near the assembly pose the linkage has enough joint compliance to absorb small geometric errors, but as the input rotates, those errors propagate around the closed loop and at some angle the geometric mismatch exceeds what the bearings can flex through. The loop locks.
Diagnose it by measuring the twist angle on each link with a rotary stage and a dial indicator. You are looking for any link where α deviates from spec by more than 0.3°. The cheap fix is shimming the joint blocks; the right fix is re-machining the offending link on a 4-axis mill with the twist set in fixturing rather than relying on post-machining alignment.
Two Bennett 4Rs in series give you two degrees of freedom — you need two actuators or a synchronization mechanism between them. A Goldberg 5R fuses the two Bennett loops into a single closed kinematic chain with one DOF, so one motor coordinates the whole fold. That is the entire design rationale.
Pick the Goldberg 5R when you need single-actuator coordination and the geometry of the parent Bennett loops is compatible. Pick stacked Bennetts when you need independent control of two fold stages, or when the Bennett-compatibility condition cannot be satisfied for the link lengths your packaging demands.
The Goldberg 5R has no inherent stiffness in the deployed pose. It is a 1-DOF mechanism, and that single DOF runs right through the deployed configuration. Under any load that drives motion along the fold direction, the linkage simply folds backward unless you have added a hard stop, an over-centre latch, or a tensioned cable.
The fix is structural, not kinematic. Add a deployment latch that engages at full extension, or design the parent Bennett geometry so the fully-deployed pose sits a few degrees past a singular configuration where the input torque required to reverse-fold spikes naturally. Both approaches are common in deployable space hardware.
For a classroom demonstrator at 100-150 mm scale, FDM-printed links work — but only if you print the joint bores oversize and press in metal bushings, and you accept that twist-angle tolerance from a typical hobby printer is around ±0.5°, which means the linkage will bind partway through its range.
For functional hardware, the joint axes must be defined by metal features. SLS nylon or printed plastic with bonded steel pin sleeves can hit ±0.2° if you design the print orientation carefully. Anything load-bearing or aerospace-grade goes to machined aluminium or stainless with ground dowel pins for the joint axes — that is the only reliable route to ±0.1° twist tolerance at 200 mm scale.
You are approaching a kinematic singularity. The Goldberg 5R, like every overconstrained spatial linkage, has input angles where the instantaneous mechanical advantage between the input crank and the output motion drops toward zero. Near that pose, a small motion of the output requires a very large motion of the input — equivalently, a small load on the output demands a large input torque.
This is often a feature, not a bug. Designers deliberately place the singularity just past the deployed pose so the structure self-locks under reverse load. If the singularity is in the wrong place for your application, you have to redesign the parent Bennett geometry — twist angles α1 and α2 control where the singularity sits along the input range.
Geometrically there is no cycle-life limit — the kinematics do not care how many times you fold it. The limit is in the joints. Because the linkage is overconstrained, any wear in the bearings introduces effective tolerance on the joint axis position, which behaves exactly like a manufacturing twist-angle error. After enough cycles, a worn linkage will start to bind partway through its range even though it deployed cleanly when new.
For research robotics applications cycling thousands of times, use crossed-roller bearings or preloaded angular contact bearings with sub-5 µm runout. For one-shot or low-cycle aerospace deployments, simpler bushings are adequate because the total motion budget is small.
References & Further Reading
- Wikipedia contributors. Overconstrained mechanism. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
