How to Prevent Flash: Shut-Offs, Vents and Clamp Tonnage

At 2:40 a.m. in week 31 of a 16-cavity PBT connector program, a 0.03 mm resin feather appeared on the cavity 11 shut-off. The operator pulled 96 parts in 40 minutes, and the shift scrapped 412 flashing housings before stopping the press at 4:10 a.m.

Flash is resin that bleeds past steel meant to seal: parting lines, core shut-offs, insert seams and ejector pins. Melt at 30 to 80 MPa finds any gap wider than about 0.01 to 0.02 mm and fills it in a fraction of a second.

This guide goes deep on flash only, and it starts where flash starts, in the steel. Shut-off geometry, vents, clamp tonnage, plate deflection and component fit come first, process and resin second. A companion guide covers how to troubleshoot mold defects at overview depth for all nine common issues. Figures here are typical industry values with ranges unless marked as DieStrike capabilities.

The Snapshot

• Flash appears when shutoff clearance passes about 0.01 to 0.02 mm under 30 to 80 MPa of cavity pressure.
• Clamp sizing runs 3 to 5 ton per in2, roughly 40 to 70 MPa, over the projected area for most resins.
• Vent depth ranges 0.005 to 0.05 mm by material, with a 1 to 3 mm land and a relief cut 0.5 to 1.0 mm deep.
• A 35 to 60 mm backing plate over a 100 to 150 mm unsupported span can flex 0.01 to 0.03 mm under load.
• Hand trim of flashed parts adds 20 to 90 seconds and about $0.05 to $0.50 per part before automation.

What Flash Actually Costs a Program

A flashing mold is rarely a scrap problem first. It is a labor problem, a line-speed problem and then a quality problem. The feather itself weighs almost nothing, but every flashed part leaves the press unfinished.

Take a 16-cavity mold on a 30 second cycle, about 1,920 parts per hour and 15,360 per 8-hour shift. If 5% of parts flash, that is 768 pieces needing trim. Manual deburr runs 20 to 90 seconds per part, so one operator spends 4 to 19 hours per shift clearing the queue. At $25 per hour, direct labor lands near $100 to $475 per shift before rework is counted.

Flash also stops automation. Feeder bowls jam on loose films, vision stations reject parts, and robotic palletizers trap thin flash inside trays. Downstream cost multiplies fast: one flashed connector housing can stall a 1,200 parts-per-hour assembly line for 15 to 30 minutes while it clears.

Add the mold side of the ledger. Flash is the first sign that steel is moving or wearing, so the repair clock starts early. Deepened vents, nicked shut-offs or skipped preventive maintenance make the next flash cheaper to cause and costlier to catch. Spread the arithmetic: 1% of a 2,000,000-part run is 20,000 parts, and $0.15 trim cost makes that $3,000 in labor before rejects. Real flashing programs run well above 1%, which is why trim cost, line stops and repair time drive the business case for prevention.

Shut-Off Design: Area, Angle and Steel

A shut-off is any steel-to-steel surface that stops melt flow: the parting line, a core side wall, a slide face or an insert seam. The seal is metal contact, not a gasket, so it only works while the two faces press together hard enough to close gaps.

Flash begins when the gap at a shut-off opens past about 0.01 to 0.02 mm during injection. Melt viscosity sets which side of that window matters. A viscous PC part tolerates a 0.02 mm line, while a fluid LCP grade flashes at half that. The clearance budget is a design decision, and it starts with shut-off angle, land area and steel hardness.

Parallel shut-offs rely entirely on clamp force to stay closed, which makes them sensitive to tonnage and plate flex. Angled shut-offs add a self-centering wedge. Typical angles run 1 to 3 degrees, and the wedge converts cavity pressure into a closing component on the steel. Angles above 5 degrees need more steel and remove draft from the part. Angles below 1 degree risk galling as the faces wipe on every cycle.

Shut-off area matters twice. Wide lands spread wear and resist local crushing, but they also add sealing surface that must stay within the clearance budget across the whole face. A nicked corner or a low spot of 0.02 mm on a 40 mm land becomes a leak path the moment pressure spikes.

Steel grade decides how long the land holds its geometry. Unhardened P20 at 28 to 32 HRC bruises, wipes and peens shut-offs within tens of thousands of cycles. Production shut-off lands should run HRC 48 or harder. Hardened cavity steel is the default, with nitrided or coated inserts where abrasive fillers are present. Polish shut-off faces in the direction of draw, because cross-grain scratches of 0.005 to 0.01 mm become melt channels of their own.

Measure the design intent on the molded part. A consistent parting line witness of about 0.02 mm across the part usually means the shut-off is closing and sealing evenly. Witness that thickens on one corner, or disappears on another, is a steel geometry problem, not a process problem. Parting line placement, gate side versus far side, and stepped or tapered line design are covered in our guide to designing parting lines.

Vents and Relief: Depth, Land and Placement

Vents are deliberate gaps, and that is the paradox every molder lives with. A vent must let air out of the cavity, so it stays open. It must also sit below the flash threshold so melt cannot enter.

Depth runs from 0.005 mm for the most fluid resins up to about 0.05 mm for viscous ones. The land, the flat steel the melt must cross, runs 1 to 3 mm. The relief behind the land opens 0.5 to 1.0 mm deep to atmosphere. Cut relief every 25 to 50 mm along the parting line. Add it at every last-fill location: rib ends, boss tips, weld-line zones and core far sides.

← swipe to scroll →

Table: typical vent depth starting points by resin family; verify against the specific grade data sheet and trial shots.

Vents fail in one of two directions. Too shallow, and air traps cause burns, short shots or weak weld lines. Too deep by a few hundredths of a millimeter, and the vent becomes the first flash location. It is simply the widest gap the melt can reach.

Flash at a vent notch means the vent was deepened. Over-polishing, crash damage or a past repair usually ground it out, and cleaning with steel tools rounds the land further. Check vent depth at mold receipt and at every preventive maintenance stop. A 0.01 mm change at the land is invisible to the eye and fatal to the part. Full vent layout rules, including vented ejector pins and insert seams, sit in our guide to mold venting.

Clamp Tonnage: Projected Area and Pressure Math

Cavity pressure pushes the mold open along the parting line. Clamp force must beat that push across every square inch of projected area, or the parting line opens and flash follows. The sizing rule of thumb for most resins is 3 to 5 ton per in2, about 40 to 70 MPa over the projected area. High-flow and high-pressure materials sit on the upper end.

Projected area is the part silhouette seen looking down the tie bars, including runners and feed systems, not the total steel surface. A mold can carry a huge shut-off area in shear and need only modest tonnage. Only the projected component of melt pressure tries to separate the plates.

Run the math on real parts. A connector housing at 60 by 40 mm has a footprint of 2,400 mm2, about 3.7 in2. Sixteen cavities bring that to 59 in2. At 4 ton per in2, the clamp need is near 238 ton before feed system and safety margins. Adding 15% for runners and machine tolerance lands around 274 ton, so the mold books a 300 ton press.

← swipe to scroll →

Table: clamp sizing worked at 4 ton per in2, about 55 MPa, with 15% feed and safety margin.

Cavity pressure is not flat during the cycle. Average pressure sits in the 30 to 80 MPa band. At transfer, when injection switches to hold, pressure can spike 15 to 30% above average at the gate. That spike is when marginal shut-offs flash, which is why tonnage checks should use peak transfer conditions, not steady-state hold.

Two machine-side checks keep the math honest. First, confirm actual clamp force, because a hydraulic press at a 350 bar setting may deliver 10 to 15% less tonnage as oil warms. Second, watch platen parallelism and tie-bar stretch. Uneven clamp loads the short side of the mold and unloads the long side. The unloaded side flashes first.

More tonnage is not always better. Over-clamping bends plates, closes vent gaps the melt needs and loads support pillars unevenly. The right number is the smallest clamp that keeps the parting line sealed at peak pressure, which is exactly what the projected area math finds.

Plate Deflection and Component Fit

Steel flexes, and at flash tolerances even small flex matters. Cavity pressure pushes against the cavity floor, the floor pushes the backing plate, and any unsupported span between support pillars bows. A plate that bows 0.02 mm under load has opened a shut-off by the full flash threshold, no matter how perfect the polish.

Plate thickness in production molds typically runs 25 to 60 mm, and support pillar spacing decides how much of that thickness is useful. Run the beam math on a working example. A 45 mm backing plate spans 140 mm between pillars. Loaded by 50 MPa over a 120 mm strip, it deflects about 0.02 mm at mid-span. That is not a theoretical failure; it is the flash threshold reached by ordinary geometry.

Deflection scales steeply with the variables a designer controls. Flex drops with the cube of plate thickness. A 45 mm plate is about 2.1 times stiffer than a 35 mm plate over the same span. Flex grows with the fourth power of the unsupported span, so pillar spacing of 100 to 150 mm beats adding plate steel. A few extra pillars under the cavity zone cost little and remove most of the bending risk.

Component fit is the second half of the deflection story. Slides, cores, inserts and lock blocks each carry a fit clearance that behaves like a shut-off under pressure. Slide-to-heel block fits are set in the 0.005 to 0.02 mm range when new. As gibs and guide surfaces wear past 0.03 to 0.05 mm, the slide shifts and flash appears on the moving side of the part. Side lock systems exist to take that load, and our guide to selecting side locks covers the preload math behind them.

Check the whole stack, not just the cavity. Ejector plates must sit flat and preloaded. Cooling plates must not bow under water pressure. The cavity plate to backing plate fit must be true metal contact, not a shimmed gap. Ground plate flatness typically holds 0.005 to 0.02 mm across the face.

Every 0.01 mm of stack-up error lands directly on the parting line budget. Plate grades, thickness and pillar layout are where this geometry is bought in the mold base.

Process Side: Melt, Pressure and Viscosity

Process cannot create flash in a mold whose steel is truly sealed, but it decides the fate of every marginal shut-off. When clearance already sits near 0.01 to 0.02 mm, process variables push the gap across the threshold for a few hundred milliseconds every cycle.

The transfer spike is the main process lever. Late switchover overpacks the cavity. Pressure spikes 15 to 30% above the set hold value exactly when the mold is fullest. Move the switchover point earlier by 2 to 5 mm of stroke. Shifting 3 to 8% of fill volume out of the pressure phase often clears a marginal flash without touching the tool.

Hold pressure and melt temperature act on viscosity directly. Cutting hold pressure 15 to 25% lowers the force feeding the gap. Dropping melt temperature 10 to 15 C raises viscosity and slows bleed in fluid resins. Both moves trade against sink and shrinkage, so they only work inside the part's quality window.

Injection speed profile matters at the end of fill, not the start. High speed through the last 5 to 10% of stroke generates the pressure spike. A deceleration ramp before transfer is standard practice on flash-prone tools. Fill time changes of 0.5 to 2 seconds can be tested on the press without a single mold modification.

Machine condition is part of the process. A clamp that reads 300 ton but delivers 260 unloads one corner of the mold. Toggle backlash and uneven tie-bar stretch do the same. Verify clamp force at the start of every run and after press maintenance. Process work cannot fix a press that will not hold tonnage.

Materials That Flash: Viscosity and Processing Windows

Every resin flashes; they only differ in how easily and how thin. The driver is viscosity at melt temperature. The practical proxy is how the grade flows under high shear in the mold, not the datasheet melt flow rate at low shear.

LCP is the extreme case. Near-zero viscosity at 330 to 360 C lets it penetrate gaps that stop other resins, so its vent depths run 0.005 to 0.01 mm. Its flash is a brittle hairline film, nearly invisible until parts jam an assembly line. PA66 and POM follow close behind with thin films, while PBT flashes readily near the gate on long runs. Viscous amorphous resins such as PC and ABS flash rarely, because thick melts need wider gaps and higher pressure to bleed.

High melt flow rate grades widen the gap between good and bad. A PP grade at 30 to 40 g/10 min flows more like water than the same resin at 4 to 8 g/10 min. A connector-grade PBT or PA66 chosen for thin walls is exactly the low-viscosity material that flashes first. Material selection for flash-prone parts should weigh flow against the part geometry, never pick the highest flow grade by default.

Processing windows change the arithmetic, too. Crystalline resins such as PBT, PA66 and POM flash more as melt temperature climbs toward the top of their 10 to 20 C band. Viscosity falls with every degree. Moisture is a second variable: wet PBT or PA66 hydrolyzes into lower molecular weight chains that flow thinner and flash thinner. Dryers holding dew point near minus 40 C keep the melt viscosity where the mold was designed for it.

Fillers and reinforcements cut flash proneness. Glass-filled grades at 30 to 50% loading flash less than unfilled versions of the same resin. Fibers raise effective viscosity, but they abrade vents and shut-offs faster. That trades a flash problem for a wear problem. Aramid or PTFE-modified grades sit between the two extremes.

Material guides cover the most common flash-prone connector resins. The PBT and LCP guides are written around processing windows and tool steel choices.

Detecting Flash: Where It Appears First

Flash announces itself in predictable places before it becomes a reject wave. The first locations are over-deep vents and shut-off corners with nicks or low spots. Insert seams and the parting line opposite the gate follow, where clamp load is lowest and plate flex is highest.

Catch it at the right magnification. A 0.05 mm flash on PP is visible to the naked eye. An LCP hairline at 0.01 to 0.02 mm needs a 5 to 10x loupe or a telecentric vision system. Sample discipline matters more than the tool.

Pull 5 parts after the mold reaches thermal equilibrium at 30 to 50 cycles. Inspect the parting line perimeter and shut-off witnesses. Repeat the check at the start of every shift.

Weight and geometry catch what eyes miss. Flash adds 0.01 to 0.5% to part weight, depending on thickness and perimeter. A weight check on a 10 part sample flags a forming feather before it becomes a reject stream. Dimensional checks and a 0.02 mm witness measurement quantify the trend. Flash that grows 0.005 mm per month is a wear curve, not an event.

Read flash as a diagnostic, not a nuisance. A mold that ran clean at 300 ton and flashes at the same settings after 200,000 cycles is telling you steel has moved. Shut-off lands wear past HRC 48 zones, vent lands peen open, or guide components pass 0.03 to 0.05 mm of clearance. Flash on one cavity of a multi-cavity tool points at a cavity-specific fault: an insert seat or a local vent. Flash on every cavity points at clamp, plate or process causes.

Fix the steel before tuning around it. Grinding and re-fitting a shut-off restores the original clearance budget. So do re-cutting vent depth to the material table and replacing worn gibs or side locks. Our mold repair and maintenance service does this work to documented geometry, not by feel. Process settings that hide a 0.02 mm steel fault today will fail again the day the resin grade changes.

The Flash-Prevention Checklist

Prevention is a list of decisions made at design, verified at sampling and re-checked at maintenance. Run this list on every flash-prone program:

• Confirm shut-off angles of 1 to 3 degrees on cores and slides, with hardened lands at HRC 48 or harder.
• Verify the clearance budget: no shut-off gap above 0.01 to 0.02 mm at the drawing stage and on first articles.
• Set vent depth from the material table, 0.005 to 0.05 mm. Keep lands at 1 to 3 mm, with relief 0.5 to 1.0 mm deep every 25 to 50 mm.
• Size clamp tonnage at 3 to 5 ton per in2, about 40 to 70 MPa, over projected area including runners. Add 10 to 15% margin.
• Layout support pillars so unsupported spans stay under 100 to 150 mm under the cavity zone, with plates of 35 to 60 mm.
• Specify side locks on slides and large shut-offs so wear does not migrate into the parting line fit.
• Record vent depth, shut-off hardness and pillar layout in the mold book so maintenance can verify them, not guess them.
• Sample 5 parts at 30 to 50 cycles at every run start and measure witness, weight and vent-line condition.
• Clean vents with brass or copper only, and re-check depth after any crash, polish or repair event.
• At preventive maintenance, inspect shut-off lands for nicks, peening and hardness drop before flash appears on parts.

The checklist works because every item is a number that can be measured. A mold with documented vent depths, verified shut-off clearance and a clamp math sheet starts below the flash threshold. It stays there until something in the list changes.

FAQ: Preventing Flash

Q1. How much clearance causes flash on a shut-off?

Flash appears when the gap at a shut-off opens past roughly 0.01 to 0.02 mm under cavity pressure. Viscous resins tolerate the upper end. Fluid grades such as LCP flash below it. The clearance budget is set by the resin, not by a universal number.

Q2. What is the right vent depth to avoid flash?

Depth follows the material table. Use 0.005 to 0.01 mm for LCP, and 0.01 to 0.03 mm for POM, PA66 and PBT. Viscous PC and forgiving PP and PE grades run up to 0.05 mm. The land must stay 1 to 3 mm long, and the relief behind it must open 0.5 to 1.0 mm deep to atmosphere.

Q3. How much clamp tonnage do I need to stop flashing?

Size at 3 to 5 ton per in2, about 40 to 70 MPa, over the projected area including runners. A 16-cavity connector tool at 59 in2 needs roughly 238 ton at 4 ton per in2, so it books a 300 ton press. High-flow resins and thin walls justify the upper end of the range.

Q4. My mold flashed clean for a year and now flashes at the same settings. What changed?

Assume wear before process drift. Shut-off lands soften or nick, and vent lands peen open past the material depth. Guides or side locks wearing past 0.03 to 0.05 mm let steel shift under pressure. Measure vent depth and shut-off witnesses, then repair the steel before changing the process.

Q5. Can process settings alone fix a flashing mold?

Only if the steel is marginal, not failed. Earlier switchover, lower hold pressure and a deceleration ramp reduce the transfer spike that triggers flash. Cut hold pressure 15 to 25% and ramp down in the last 5 to 10% of fill. If clearance already exceeds 0.02 mm at the shut-off, process tuning delays the inevitable; the steel needs repair.

The Bottom Line

Flash is a clearance problem with a price tag. The gap that opens past 0.01 to 0.02 mm at a shut-off is designed into the steel. Four numbers set it: shut-off angle of 1 to 3 degrees, and vent depth of 0.005 to 0.05 mm. Clamp tonnage runs 3 to 5 ton per in2, with plates stiff enough not to bow.

Get those numbers right, and the process has almost no way to flash the part. Get one wrong, and no amount of process tuning keeps flash away for long.

The cost model decides how much prevention is worth. Trim labor at $0.05 to $0.50 per part compounds fast. Line stops measured in shifts and repair visits measured in weeks compound faster. Hardened shut-off lands, documented vent depths and a real clamp math sheet pay for themselves on the first long run.

DieStrike builds molds under IATF 16949 across 120+ machines. Geometry holds ±0.005 mm standard and ±0.002 mm precision. Molds ship in 2 to 5 weeks, and standard components in 3 to 7 days. Shut-off clearance, vent depth and plate stack arrive machined to these numbers, not hand-fitted on the floor.

The same discipline serves connector OEMs including TE Connectivity, Amphenol and Luxshare. Hairline LCP flash on a 0.5 mm pitch housing is a field failure, not a trim issue. Give the steel the clearance budget it needs, and flash stops being a defect to fight and becomes a number to verify.