PEM inserts for sheet metal: hole size, push-out and coating

An M4 clinch insert in a 2 mm CRCA panel spun at 6 N·m on the assembly line, three weeks after the panel was coated. The insert had been set before pretreatment, the hole had been laser cut at 6.55 mm, and the paint had filled the knurl. Two errors, one symptom: the joint had no interference and no clean thread. PEM inserts tolerate neither fault, and both were built into the panel before it was formed.

Clinch inserts are a precision fit in a punched hole. They are not a fastener you drop into whatever hole the laser produces, and they are not a feature you add after the coating decision. Both of those assumptions are made on most drawings we see.

What hole size do PEM inserts need?

The hole size comes from the insert's part number, not from the screw that goes into it. An M4 knurled steel insert takes a 6.4 mm hole held at +0.08/0, because the shank has to pass through and the knurl has to bite into the sheet around it.

Three things go wrong when the hole is wrong:

Too large. The knurl cannot displace enough material into the undercut, so the insert sits loose. Push-out falls and torque resistance falls faster, because the insert rotates before the knurl can engage. A hole 0.15 mm over the top of tolerance is enough to lose most of the torque resistance.

Too small. The shank will not enter without deforming the sheet. The panel bulges around the insert, the insert stands proud of the face, and on thin sheet the edge of the hole can crack. Pressing harder does not fix it; it makes the bulge worse.

Out of round. A laser-cut hole in thick sheet can be 0.05–0.1 mm out of round because of the lead-in pierce and the kerf taper. The knurl then engages on one side only and the insert cocks slightly. That shows up as a thread that is not perpendicular to the face, which is worse than a weak insert because it misaligns the mating part.

Flushness. The insert sits in the sheet, not on it. On a visible face the head should finish within 0.1 mm of the surface, with the knurl buried. Too little setting force leaves the head proud; too much dishes the sheet around the head. Both faults telegraph through the powder coat as a ring.

The hole tolerance should be on the drawing as a diameter and a tolerance, not as a callout to a general tolerance block. General tolerances such as ISO 2768-1 cover dimensions without individual indications, and the medium class is far too wide for an insert hole.

Hole size, sheet range and push-out by thread

The values below are for standard knurled steel inserts in CRCA, installed with the correct setting force. Push-out is quoted at the middle of the sheet range and scales roughly with thickness, so use it as a comparison between sizes rather than a guaranteed figure for a specific part number.

The tolerance column is the same on every row because it comes from the insert, not from the sheet. What changes with thickness is how much material there is to displace.

Why push-out is lower than the shear calculation suggests

A simple model says the sheet has to shear around the insert on a cylinder of diameter d and height t, so the push-out force is:

F = π × d × t × τ

For an M4 insert in 2 mm CRCA with a shear strength of about 270 MPa, that gives π × 6.4 × 2 × 270 = 10.9 kN. The measured push-out is around 3 kN, roughly a quarter of that.

The difference is the joint geometry. The knurl does not shear the full cylinder; it displaces a ring of material into an undercut, and the failure mode is the material rolling out of the undercut rather than shearing cleanly. Push-out therefore depends on how much interference the knurl achieved, which depends on the hole being at the low end of tolerance and on the sheet being thick enough for the undercut to sit in full material.

The practical consequence: two panels with the same insert and the same nominal hole can differ by 40% in push-out if one hole is at 6.40 mm and the other at 6.48 mm. That is why the hole tolerance is +0.08/0 and not ±0.1.

The same model explains why an insert in aluminium behaves differently. Aluminium's shear strength is lower, so the shear area is not the limiting factor; the sheet yields around the knurl and the insert pulls out at a lower load from an identical hole. Use the insert variant made for the sheet rather than the steel part with a different plating.

Vibration finishes a joint that interference did not. An insert at the loose end of the hole tolerance will creep under a vibrating load even when its static push-out passes, and a thread-locking compound does nothing for push-out.

Setting order: before or after coating

This is the decision that fails most often, and it has only three workable answers.

Set after coating, hole masked. The insert goes into a clean hole of the correct size, the thread is untouched, and the coating is intact except where the head covers it. The masking plug costs a handling step per hole, and the plug has to survive the cure oven.

Set after coating, hole unmasked. The hole has closed by twice the film thickness, so the insert is tight. It can be pressed in, but the coating around the head cracks and the insert is effectively oversized for the hole. Not recommended unless the film is under 40 µm.

Set before coating. The insert is in place and the panel goes through seven-tank pretreatment and powder. Chemistry stays in the knurl annulus and bleeds out during cure, leaving a ring of contamination. The thread fills with powder; on an M4 × 0.7 thread a 70 µm film is enough that a screw will not turn by hand. Chasing every thread with a tap restores it and strips the plating from the flanks in the process.

There is a fourth option that is sometimes the right one: use a blind insert with a sealed head, set it before coating, and plug only the thread. It costs more per insert and removes the knurl bleed path.

Verify the joint after cure as well as before. A push-out check on a bare panel proves the hole and the setting force; it does not prove that the coating stayed out of the knurl. Repeat the check on a coated part from the same batch, and inspect the annulus under a magnifier for a paint bridge between the insert and the sheet.

A worked example: 200 inserts on one panel

A rack panel carries 200 M4 inserts. Setting time is about 10 seconds per insert with a press and a locating fixture, so the panel absorbs 2,000 seconds, or 33 minutes, of press time before it is a panel at all.

At that count the design is worth revisiting. Three alternatives:

Extruded and tapped holes. A drawn collar formed in the press tool adds a station but removes 200 pieces of hardware and 33 minutes of press time. The thread is in the sheet, so it is only as strong as the sheet thickness allows, and it cannot be replaced if it strips.

Fewer, larger fasteners. Redesigning the mating part to use 40 M6 inserts instead of 200 M4 inserts cuts setting time to about 7 minutes. It usually requires a change to the mating bracket, which is why it is rarely considered.

Weld studs or weld nuts. No hole tolerance to hold and no insert to press, but the weld damages the coating on the far side, spatter has to be removed, and the thread still needs protection through the paint line.

For a one-off panel, 200 inserts is a day of work and nobody argues. For a product at 5,000 units a year, the same 200 inserts is 2,750 hours of press time, and the extruded hole is the only design that survives the arithmetic.

What Ganesh runs

Insert holes are punched on a 250 T press where the part is a repeat, which holds the +0.08/0 tolerance far more consistently than a laser-cut hole in 3 mm sheet. On low-volume and prototype parts the hole is cut on the 12 kW fibre laser with its 40 ft × 16 ft bed, and the diameter is checked on the first article before the batch runs.

Press tools, progressive dies, welding fixtures and checking fixtures are built in the in-house tool room, so an insert hole that needs a pilot punch or a stepped punch can be made rather than negotiated. Tooling is quoted separately from the piece price.

Coating is seven-tank pretreatment, an 18 ft × 14 ft electrostatic powder booth and a curing oven, with DFT measurement and cross-hatch adhesion testing. Masking is a line instruction, so it has to arrive with the job card, not only on the customer drawing. The coating capability page describes the line, and the pretreatment process article explains where the chemistry sits in a knurled annulus.

Typical sheet metal tolerance is ±0.5 mm, tighter on critical features by agreement. An insert hole is a critical feature, and it should be called out as one. PPAP and first article inspection are available, and a push-out check on the first article is the only way to know that the hole, the sheet and the setting force agree.

What to put on the drawing

• Insert part number, material and finish, including the plating specification.
• Hole diameter with its own tolerance, not a reference to the general tolerance block.
• Sheet thickness range the insert is rated for, and the actual thickness of the part.
• Install side, and the direction the insert is pressed from.
• Whether the insert is fitted before or after coating, and which mask is used.
• Minimum push-out and torque resistance if the joint is load-bearing, with the test method.
• Thread locking or sealing requirement, if any.
• A note on what happens if the insert is damaged after coating, because removal and replacement is a repair procedure, not a rework step.

When this route is wrong

Clinch inserts need access to both faces. A closed box section with inserts inside cannot be set after the box is welded, because the press has nothing to react against. Those parts need the insert set before welding, with the weld heat kept away from the knurl, or a different fastener entirely.

Below 0.8 mm sheet the standard insert has too little material to displace, and the panel distorts around every hole. Aluminium needs its own insert variant, not a steel insert in a softer sheet. And a thread that has to carry a structural load belongs in a weld nut, a tapped plate or a through-bolt, not in 1 mm sheet with a knurl in it.

We do not run in-die tapping, so a tapped hole is always a separate operation. If the design needs a thread in the sheet rather than in an insert, the rivnut comparison sets out the alternative and where it fails.

Send the panel drawing with the insert part number, the hole callout and the coating sequence to request a quote. We will confirm the hole against the sheet range before the punch is ground. More fastener design notes sit in the design and CAD hub, and the stamped part inspection guide covers the first article checks that catch a hole at the top of tolerance.