How to Design Molds for Inserted Terminals: Lead Frames and Flash

A 24-cavity overmold tool for a 12-position connector ran clean at the trial and grew 0.03 mm flash fins at the terminal shoulders by shot 40,000. Coplanarity on the solder tails moved from 0.06 mm to 0.14 mm, and three lots were rejected at the customer's placement station. The nests had been cut at 0.1 mm clearance against a frame held to 0.05 mm, so the wear went into the steel, not the stamping. That sequence is a mold design result, not a process accident. This guide covers the inserted-terminal tool as one system: metal spec, nests, gating, heat-sink flow, frame lift, deflash and acceptance, each with the numbers that settle it.

Generic insert tooling rules live in the insert mold design guide; this article joins the stamped leadframe to the molding decision, in build order.

The Snapshot

• 0.05 to 0.1 mm is the nest window. Each terminal lands in a machined nest with that much clearance, and the nest floor must match terminal coplanarity of 0.05 to 0.10 mm or the frame floats during fill.
• Gate on plastic, never on plating. Gates of 0.5 to 1.5 mm land on a molded pad above 0.15 to 0.3 mm wall connectors; melt over tin plating above 300 °C at the gate lifts the finish.
• 40 to 80 MPa tries to lift a 0.2 to 0.8 mm frame. Projected areas need only 20 to 40 tons of clamp, so frame support comes from pillars under the nests, not from the press.
• A 0.02 mm mismatch prints a 0.03 mm fin. Insert-face errors become flash on the parting, and trim dies cut it on a 0.1 mm land with D2 edges at HRC 58 to 62.

The Metal Spec Decides the Mold

Before a cavity is cut, the frame print sets the tolerance budget. Inserted-terminal molds are designed around the metal, not around the plastic. A stamped copper-alloy leadframe for a connector arrives at 0.2 to 0.8 mm thickness, most often phosphor bronze, beryllium copper or a copper-iron alloy, and every molding feature inherits that range.

Three metal dimensions control the tool. First, coplanarity: terminals must sit within 0.05 to 0.10 mm on the seating plane, and the mold nests have to match that flatness or the frame rocks under injection pressure. Second, plating: tin or silver at 2 to 8 micrometers is a wear surface for the nest, not for the part, and a plated frame dragged across a burr picks up steel smeared off the nest wall. Third, springback: a frame that is not stress-relieved or annealed after stamping relaxes inside the cavity, and a terminal measured flat on the strip can lift 0.05 to 0.2 mm once it is cut free.

The stamping side of these numbers belongs to the connector terminal die guide, which covers punch clearance and coplanarity control at the strip. The molder's job starts with reading that print and asking one question: can my nest hold what your die releases? If the frame print allows 0.10 mm coplanarity and the connector requires 0.05 mm at the solder tail, the mismatch becomes a mold fix or a second trim operation. Decide that in the DFM meeting, before steel.

Nest Design and Insert Location

A nest is a machined pocket that locates one terminal and seals melt off under its shoulder. The working clearance is 0.05 to 0.1 mm per side in most tools. Too tight and insertion becomes a touch-fit operation that costs seconds per shot and damages plating; too loose and the frame shifts on the melt front, and the plastic wall behind the terminal goes out of spec.

Machine nest floors to ±0.005 mm against the seating plane, the standard mold tolerance for a precision tool shop, and hold ±0.002 mm on nest walls done in EDM. A 24-position frame with 12 nest cavities is a 24-feature location problem: put every nest datum off one side of the plate so a wear path does not stack error across the row.

Decide the loading method at the layout stage, because it dictates the nest. Manual loading tolerates a lead-in chamfer of 0.2 to 0.3 mm and dominates below 50,000 pieces a year. Film-feed or strip-fed loading, where the frame arrives on a carrier web indexed into the cavity, needs nests open on one face and a clamping plate that holds the web flat within 0.02 mm. Robot pick-and-place falls between the two and wants nests deep enough that a gripper overshoot cannot bend a terminal more than 0.1 mm.

Every nest needs venting at its floor. A trapped 0.1 mm air pocket under a seated terminal becomes a void or a short at the pad, so drop a 0.02 to 0.03 mm vent from the nest bottom to a parting-line vent land. It is the single most-skipped line in connector mold drawings, and one of the first things a heat check exposes.

Gate on the Plastic Pad, Not the Plating

Melt should enter the cavity against plastic-designated real estate: a pad, a rib foot or a boss that the part drawing can spare, never directly over tin or silver plating. A gate landing on plating concentrates jetting force and heat on a 2 to 8 micrometer finish layer, and anything above 300 °C at the gate front on a tin-plated frame is a plating-lift candidate.

Typical numbers: gates of 0.5 to 1.5 mm on connector housings with 0.15 to 0.3 mm nominal wall. Edge gates and tab gates dominate multi-cavity connector tools because they shear clean on a trim line. For thin-wall positions, a film gate spread over 3 to 6 mm lowers the local shear and keeps the frame from being used as a runner.

Two philosophies then split the design. A tab-seal design runs the melt against a steel land that pinches the insert face and blocks flow under the shoulder; it produces a cleaner seat but demands the shut-off numbers in the flash table below. A flash-cut design lets a thin collar of plastic form around the terminal and removes it downstream on a trim die; it forgives nest wear but adds an operation and a scrap stream. On solder-seat surfaces, choose tab-seal and pay the tolerance. On hidden faces, flash-cut is cheaper per part.

Whatever the gate type, keep gate trim inside the part plan. A vestige standing 0.2 mm high on a mating face blocks the seal; a vestige on the outside rail trims with the flash bar. Gate location is the first cosmetic decision of an overmold, and the cheapest one to change on paper.

Melt Flow Around Metal: The Heat Sink Effect

Copper-alloy frames are not passive occupants of the cavity. A 0.3 mm beryllium-copper or phosphor-bronze strip conducts heat away from the melt several times faster than the surrounding PBT or PA66 shell can, and the flow front freezes against the frame rail first. On a 12-position frame, the insert itself acts as part of the runner system, so any fill simulation run without the metal modeled is the wrong simulation. Mesh the frame into the analysis or accept that the last-fill prediction is fiction.

The practical counter-moves are local. Raise mold temperature 10 to 20 °C around the insert zone, with a dedicated circuit or a beryllium-copper insert section that holds the extra heat. Move the gate so the front meets the rail at an angle rather than head-on. Open a wall transition from 0.15 mm to 0.25 mm in the 2 to 3 mm before the frame end, so the freeze does not seal the path before pack pressure arrives.

Cavity-to-cavity balance follows the same logic. In a 24-cavity connector tool, the cavities whose frames sit closest to the runner freeze last, and the imbalance shows as a 1 to 3 percent weight spread before any steel has worn. Trim the drops to the hot side by 0.05 to 0.1 mm rather than chasing the problem with hold pressure, and verify with shot-weight data across 30 consecutive shots at the trial: a spread above 0.5 percent on a 0.5 g shot is a flow problem, not a machine problem. The layout rule for circuits near insert zones is simple: the insert steals heat locally, so the circuit has to give it back locally, with a baffle or bubbler within 8 to 15 mm of the rail.

Cavity Pressure, Frame Lift and Back-Pressure Design

The clamp math on an overmold is deceptive. Projected area on a connector body is small, so 20 to 40 tons of clamp looks generous against a 125 to 250 ton press. But cavity pressure of 40 to 80 MPa acts on the full face of a 0.2 to 0.8 mm frame, and the frame has almost no stiffness of its own. It lifts, it shifts, and the plastic wall on the far side thins until it flashes or shorts.

The answer is back-pressure: support under the nest that the cavity pressure pushes the frame into, instead of support above it fighting the press. Put steel pillars or land bridges directly under each nest floor within 2 to 4 mm of the terminal shoulder, and size the clamp-plate contact so the closing half touches the frame area before melt arrives. On strip-fed tools a spring or elastomer pad behind the carrier web holds the strip flat within 0.02 mm through the fill.

Deflection check: a 0.25 mm phosphor-bronze terminal spanning 6 mm between supports under 60 MPa cavity pressure bows roughly 0.1 to 0.2 mm mid-span. Halve the unsupported span and the bow drops by about a factor of four, which is why nest support pitch, not clamp tonnage, is the lever. Mold-opening controllers on the press catch a lifted frame as a pressure trace anomaly in the first 100 shots, so make the trace check a written deliverable in the T1 plan.

Wire-Formed Pins and Thin Terminals

Some inserts are not stamped strips but formed pins: 0.3 to 0.6 mm wire or stamp-and-form contacts standing proud in the cavity. They behave like ejector sleeves in reverse. The radial gap between pin and cavity wall has to stop flash without scoring the plating, and on PA66 or PBT that gap sits at about 0.005 mm at the seal line, tighter than most shops hand-fit on the first pass.

Two failure modes own this geometry. First, pin deflection: a 0.4 mm pin with 8 mm of free length under a melt front that lands on its tip drifts 0.05 to 0.15 mm and opens a wall-thin path. Keep free length under 3 times the pin diameter, or support it with a guide bush at the far end. Second, heat: a thin pin conducts the freeze line along its length, so the plastic around the pin root packs late and sinks. Local mold temperature plus a rib foot around the root fixes both at once.

Ejection of a part with molded-in pins needs the pin seats clear of any witness. If ejector contact is unavoidable, place it on the plastic body and size the pin per the automotive connector ejector guide, which covers the mark-free layouts. On solder-tail planes, a 0.1 mm ejector ring is a solderability reject: flux pools in the ring and the tail fails the dip test at 260 °C.

Flash at the Insert Face: Causes and Numbers

Flash on an insert tool has a short causal list, and each cause has a measurement. A worn nest at 0.1 mm clearance stops sealing melt under the terminal shoulder and prints a fin along every terminal in one cavity, not all cavities. A shut-off mismatch of 0.02 mm at the insert face prints a 0.03 mm flash fin on the parting line of the whole tool. Over-clamping that peens a seal land widens clearance the same way over time, which is why a tool can run 50,000 shots clean and then flash everywhere.

← swipe to scroll →

Table note: flash thickness is read with a caliper or optical comparator on a trimmed witness plate; clearances are measured on the steel, not inferred from the part. Confirm every number against your own shut-off print before corrective grinding.

The generic flash playbook is in how to prevent flash; the insert-tool difference is that the parting line is broken by dozens of metal islands, so the shut-off map has 10 to 40 more seal points than a plain tool of the same size. Count them in the DFM review, because every one is a place the tolerance stack-up lands.

Deflash and Cut-Off Tooling

Flash-cut designs pay for their forgiveness downstream. The inter-terminal flash bar, a plastic web between positions where the frame rails were molded over, needs a second operation: a trim die that cuts the bar and the gate tabs in one hit. The working numbers are a 0.1 mm trim land, D2 or SKD11 edges hardened to HRC 58 to 62, and 300,000 to 800,000 hits between regrinds depending on whether the bar is unfilled PBT or glass-loaded.

The cheaper answer is designing the flash to break away. Leave a 0.05 to 0.1 mm notch at each rail crossing in the cavity, so the bar snaps on the deflash station or in the orienting chute without a blade. Break-away designs need a witness test on 500 parts before release: the break must leave a stub under 0.15 mm and never crack into the housing wall.

Where the parting fin lands on the solder seat itself, no trim die fully solves it, because the cut face becomes the seating face. That is a tab-seal mold problem to fix at the steel, per the flash table above: seal in the mold, trim only the outside rails, and keep every functional surface out of the deflash path. The industry-side rules for connector cavities are in the automotive industry page.

Inspection, Pull Tests and Solderability

A molded-in terminal cannot be measured with calipers once the plastic is over it, so acceptance uses three indirect methods. Cross-section one shot per validation lot and measure the molded-in length against print, target typically 1.5 to 3 times the terminal thickness for a mechanical lock. X-ray a sampled frame position to confirm the terminal sits centered in its polymer sheath with at least 0.15 mm of cover on every face; a 0.05 mm wall over a corner is a crack path in thermal cycling.

Pull-off force is the headline number. Automotive connector specs commonly call for 20 to 60 N per terminal, measured at a 90 degree peel or straight pull per the drawing, on 10 samples with no plastic at the failure surface at the minimum. A 30 N median that scatters from 15 to 60 N is a fill or shrink signal, not a strength problem. Retention geometry details, from the simple emboss to the through-hole key, are covered in the insert mold design guide linked at the top of this article; here the discipline is the sample plan, not the shape.

Solderability closes the loop. Dip the tails at 260 °C for 5 to 10 seconds per the board process and reject any lot showing dewetting, and check the intermetallic on a cross-section: a tin-plated frame over copper that never reached the reflow interface properly shows up as a thin or absent IMC layer after one dip cycle. Two mold-side failures cause this. Silicone mold-release spray migrates onto plating and survives baking, so keep release agents off any tool that touches a solder tail; and an oil film from stamping that was never stripped burns into the plastic during overmold. If a lot fails dip-test adhesion, audit the last chemical touch before the cavity, which is almost always the release can, not the steel.

The Connector Overmold RFQ Checklist

Five documents turn an inserted-terminal quote from a guess into a number. Send them together and the DFM comes back in 24 hours; miss one and the tool gets built on assumptions.

• Frame print with coplanarity. Thickness 0.2 to 0.8 mm, seating-plane flatness callout (0.05 mm or 0.10 mm), springback spec, and whether the frame arrives annealed or work-hardened.
• Plating spec. Tin, silver or gold-flash thickness in micrometers, bath type, and the reflow or dip temperature the finish must survive.
• Molded-in length and retention method. Emboss, cut-out or through-hole, with the minimum polymer cover in millimeters on mating and seating faces.
• Trim rule. Tab-seal versus flash-cut, the allowed flash stub in millimeters, and which surfaces may never see a cut face.
• Pull test and sampling. Required force per terminal in newtons, the peel geometry, sample count per lot, and the reflow profile behind the solderability gate.

Write the lead-time expectation next to those five items. Insert and overmold tools at a precision shop run 2 to 5 weeks against a standard component supply of 3 to 7 days, because the nest work is EDM and jig-grinding at ±0.002 mm, not profile milling. A buyer who sends the checklist gets that schedule quoted honestly; a buyer who sends only a 3D step gets a number built on the vendor's guesses about coplanarity and trim.

Inserted Terminal Mold FAQ

Q1. What clearance should a terminal nest have?

0.05 to 0.1 mm per side for a stamped copper-alloy frame, with the nest floor held to ±0.005 mm against the seating plane. Tighter jams plating during loading; looser lets the frame shift under a 40 to 80 MPa melt front and flash the shoulder.

Q2. Why can melt not gate onto the plating?

Tin or silver at 2 to 8 micrometers is not a melt-resistant surface. Jetting and friction heat above roughly 300 °C at the gate lift or discolor the finish, and a gate vestige on a solder tail rejects at the dip test. Gate on a plastic pad 0.5 to 1.5 mm instead.

Q3. How do I stop a thin frame from lifting in the cavity?

Back-pressure support, not clamp tonnage. Put pillars under each nest within 2 to 4 mm of the terminal shoulder, keep unsupported spans short, and hold the strip flat within 0.02 mm with a clamp plate or spring pad on fed tools.

Q4. Do I need a trim die or can flash break away?

If the fin lands on a hidden rail, notch it at 0.05 to 0.1 mm in the steel and break it free. If it lands on a solder seat or mating face, design a tab-seal mold instead and let a D2 trim die at HRC 58 to 62 cut the outside bars at a 0.1 mm land.

Q5. What pull-off force is normal for automotive terminals?

20 to 60 N per terminal is the common band, tested per the drawing on 10 samples per lot. Scatter wider than 2 to 1 is a fill or shrinkage signal, so chase gate and pack settings before questioning the resin.

The Bottom Line

Inserted-terminal molds live by five numbers: coplanarity matched at 0.05 to 0.10 mm, nest clearance at 0.05 to 0.1 mm, gates of 0.5 to 1.5 mm landing on plastic rather than plating, support pillars within 4 mm of every terminal shoulder, and a 0.1 mm trim land under D2 edges at HRC 58 to 62. The metal spec is the mold spec: read the frame print first, cut steel second. Every mismatch the stamping releases, the overmold pays for in flash, lift or solderability rejects, and all three are cheaper to prevent on paper than to grind out of a hardened cavity at shot 40,000. The cavity and seal rules that sit downstream of this article are in the automotive connector mold guide listed in related reading; the strip-feed side is in the terminal die guide linked above.

DieStrike builds insert and overmold tooling to IATF 16949 with ±0.005 mm mold accuracy, ±0.002 mm on EDM and jig-grinding, HRC 62 max heat treatment, and 120+ machines behind a 2 to 5 week standard lead time. Tier-1 programs with TE, Amphenol, Luxshare and Dongshan have trained us on one rule: the connector tolerances that survive PPAP are the ones the frame print and the cavity print agreed on before the first electrode was cut.