Clinching for sheet metal: where it replaces a weld or a rivet

Clinching sheet metal joins two sheets with deformed metal and nothing else. A punch drives the stack into a die, the material thins under the punch face and flows sideways into the die undercut, and the button that forms is wider than the punch that made it. There is no heat, no filler, no fastener and no fume. That is why the joint turns up on enclosure doors, canopy panels and bracket stacks where a weld would distort 1 mm sheet and a rivet would leave a head standing proud of the face.

The decision is not cosmetic. A round clinch in 1.5 mm CRCA carries 3–4 kN in shear, while a 5 mm spot weld nugget in the same stack carries about 6 kN. Clinching therefore needs 1.5–2× the joint count to match a welded joint, and it needs both faces of the part open to a punch and a die. Those two facts decide most of the cases before strength is even calculated.

What follows is the limit set that makes a clinch joint work, the failures that make it fail quietly, the arithmetic for a real bracket, and the note that has to appear on the drawing.

Where clinching sheet metal replaces a spot weld

Replace a spot weld with a clinch when the stack is 1–3 mm total, both faces are flat and reachable, the load runs in shear, and the part is coated after joining. Keep the weld when the stack falls outside that band, when one side is closed by a formed return, or when the joint lives in vibration.

The access rule is often misunderstood. A spot weld needs an electrode on each side, but a pneumatically applied electrode can reach over a 20 mm flange and onto a curved face. A clinch tool cannot. The die has a cavity that must sit square and flat against the lower sheet, and the punch needs a rigid, perpendicular approach on the upper sheet. A die that lands on a bend, on a bead or on a 15° slope produces a button with an uneven undercut, and the joint that comes out of it is 40% weaker than the one on the test coupon.

The second misunderstanding is thickness. Clinching is a cold-flow process, so the material has to move. A 0.8 mm stack has very little material to push into the undercut, and a 4 mm stack needs press force that most handheld and bench tools cannot deliver. The practical window is 1–3 mm total, with 1.5 mm to 1.5 mm being the most forgiving combination on mild steel.

Clinch joint limits that decide the design

The values below are for a round clinch in CRCA with a matched punch and die. They move for aluminium, for high-strength low-alloy grades and for coated stock, and the die geometry has to be re-set for each.

What actually holds the button together

Three dimensions in the cross-section carry the load: the neck thickness, the undercut depth and the button diameter. The tool sets all three, and the material decides how much of each is available.

Neck thickness

The neck is the thinnest ring of material left between the punch face and the die cavity. It is where a clinch joint begins to fail in shear, because the punch has thinned the sheet by 40–50% at that point. A neck that is too thin gives a joint that snaps cleanly at 2 kN; a neck that is too thick means the material never flowed and the undercut is shallow. The tool designer trades one against the other, and the operator cannot fix the result by pressing harder.

Undercut and interlock

The undercut is the lip of material that has flowed out under the die shoulder and now sits behind it. Without an undercut there is no mechanical interlock, only friction, and friction in a vibrating panel is worth very little. Undercut depth on a well-set 1.5 mm joint is 0.15–0.25 mm. That is a small number, which is why die wear matters: 0.05 mm of wear on the die shoulder removes a quarter of the interlock.

Thickness ratio and material

A 3:1 ratio is the practical ceiling because the thin sheet has to bend around the thick one. At 4:1 the thin sheet necks before it reaches the undercut and the button splits on the die side. High-strength grades with 12–16% elongation behave like the thin side of a bad ratio even at equal thickness, and dual-phase steels often need a different tool or a different joint entirely.

Coated stock shifts the numbers. Zinc is soft and acts as a lubricant, so the material slides instead of locking; expect 10–20% lower shear strength on hot-dip galvanised sheet than on bare CRCA with the same tool. Paint is worse, because the film compresses and the joint loses preload in service.

Where the button lands

The die-side button is the constraint most drawings miss. It stands 0.3–0.6 mm proud, so a panel that clinches onto a frame cannot then sit flat on that frame. Either the joint moves to a flange that is not a seating face, or the mating part gets a relieving pocket, or the button is coined flat in a second station. On an enclosure door with a gasket, a proud button under the gasket line is a leak path that no amount of sealant fixes.

Sizing a bracket joint: the arithmetic

Take a 1.5 mm CRCA bracket, 120 mm × 80 mm, clinched to a 1.5 mm panel with six joints at 30 mm pitch. Each joint carries 3.5 kN in shear, so the group carries 21 kN before any safety factor. A four-weld version of the same bracket, with a 5 mm nugget at 6 kN, carries 24 kN. The clinched version is not weaker, but it needed two more joints, and the joint line is longer for it.

Now apply a real load. A canopy bracket carrying a 12 kg accessory through a 5 g shock sees 590 N. Even with a safety factor of 3 the group is at 1.8 kN against 21 kN, so strength does not govern. What governs is the panel: the six buttons sit in a line 30 mm apart, and the panel between them is unsupported for 30 mm. At 1.5 mm that is stiff enough. At 0.8 mm it is not, and the joint line will show as a ripple on the painted face.

That is the useful conclusion from the sum. On light gauge, the joint count is set by panel stiffness and by the spacing that stops the sheets from breathing, not by the shear capacity of the button. Twenty-five to thirty millimetres is the working pitch for that reason, not because the button needs the company.

What Ganesh runs, and what we do not

Clinching is not a process on our floor, and we will not quote it as one. What we do is build the parts that a clinched assembly needs: laser-cut blanks on the 12 kW fibre laser with the 40 ft × 16 ft bed, formed parts on the 200 T, 5-axis brake with the 4,450 mm bed, and the checking fixture that proves the joint positions before the parts ship.

Where a drawing asks for clinching, the joint normally converts to one of the routes we run: spot welding at the 30-plus manual stations, MIG on the three CLOOS robotic cells, or a rivet where the joint has to come apart. The conversion is not free. A weld leaves a heat mark on a painted panel, so the coating sequence moves after joining, and a rivet leaves a head that has to be accounted for in the stack.

Our own tool room builds the checking fixtures, which matters more here than it sounds. A clinch pattern is a hole pattern with no hole, so the fixture has to locate on the button centres, and a fixture built from the drawing rather than from the tool is the usual reason a first article is rejected. The welded assemblies route covers what happens after the joint is made, and the custom parts range covers the brackets and panels that carry these joints.

What to put on the drawing

1. Joint symbol and process. Name the process, the joint diameter and the standard you are working to. A generic "join here" note will be interpreted as a weld.
2. Joint positions as a datum dimension, not as a chain from the part edge, so the pattern is inspectable after forming.
3. Both faces. State which face takes the punch and which takes the die, and state the maximum button height on the die side.
4. Stack and thickness ratio, with the material grade and the coating weight for each layer.
5. Edge distance and pitch, or a note that the default of 1.5 die diameters and 25 mm applies.
6. Flatness of any mating face that the button must not disturb, with the relieving pocket if one is needed.
7. Joint test: a shear or peel value on the first article, and the frequency of the pull test in production.

The hole-to-bend distance guide explains why a joint near a bend is a bad joint: the bend pulls material away from the button and the undercut closes up. If the only place a joint can go is within 10 mm of a bend line, the design is asking for a weld instead.

When clinching is the wrong joint

Clinching loses to a spot weld in three situations. The first is fatigue. The button is a notch on both faces, and a bracket that sees 10⁶ cycles at 2 kN will crack at the neck long before a welded equivalent does. The second is a closed section. If one face of the joint is inside a box that is already formed, neither the die nor the punch can reach it, and the joint has to be made before the box closes. The third is a joint that must be fluid-tight, because a clinch is a controlled leak.

It also loses on repair. There is no way to release a clinch except by drilling, and re-clinching the same spot does not restore the strength because the material is already cold worked. Where a panel has to come off in service, use a rivet or a screw and design the head clearance for it.

Aluminium is a separate case. A clinch in 5052-H32 works, but the interlock is shallower, the button is more prone to cracking, and the joint needs a different die. Aluminium to steel is worse: the button creates a crevice between two dissimilar metals, and the corrosion that follows is at the joint, which is the one place you cannot inspect. Use an isolating washer and a rivet, or keep the materials apart.

Send the drawing with the stack, the joint positions and the load case through the RFQ route and the joint decision gets made before the tool is cut. The design and CAD hub covers the rest of the drawing rules that keep a fabricated part inspectable, and the panel welding guide covers the joint that usually replaces this one.