Sink marks are rarely a moulding problem and almost never fixed by more pressure. Where they really come from, how to diagnose them in an hour, and the design, tool and process fixes that work, in that order.
It is 4:40 on a Friday afternoon. The first-off parts of a new appliance housing are on the quality bench, and under the inspection lamp there is a faint shadow on the show face, right behind every screw boss. The setter has already pushed hold pressure up twice. The sink got shallower, the parts started flashing at the parting line, and now the customer's quality engineer is asking whether the tool is fit for purpose.
We have seen versions of that afternoon many times. And in most of them, the answer was decided months earlier, on a CAD screen, long before anyone pressed the cycle-start button.
What a sink mark actually is
A sink mark is the visible footprint of volumetric shrinkage. Every thermoplastic shrinks as it cools: amorphous materials such as ABS, PC and PC/ABS by roughly 0.4–0.7%, semi-crystalline materials such as PP, PA and POM by 1.5–2.5% or more. Wherever the section is thicker, the core stays molten longer than the skin. When it finally shrinks, it pulls the already-frozen skin inwards, and on a glossy surface a depression of just 0.02–0.03 mm is enough for a customer to see.
That is why sink always appears opposite a thick feature: a rib, a boss, a gusset or an intersection of walls. The thick spot is the cause; the surface is just where the evidence shows up.
The ten-minute diagnosis
Before changing anything, answer four questions. They separate design problems from tool and process problems faster than any amount of trial and error.
- Where is it? Directly opposite a rib, boss or junction points to design. Far from the gate on otherwise uniform walls points to packing.
- Does it change with hold time? Run three cycles at your current hold time, then three with it doubled. If the sink barely moves, the gate has frozen before the section can be packed. No amount of pressure will fix it.
- Is the cushion stable? A cushion that bottoms out or wanders by more than a few millimetres means the screw cannot hold pressure. Check the check ring before you blame the part.
- Is it worse on one cavity? On a multi-cavity tool, sink on one cavity only usually means unbalanced filling or a cooling circuit that is blocked or wrongly connected.
Why the fix usually starts in the design
The rule of thumb that prevents most sink is simple: ribs and boss walls should be about 50–60% of the nominal wall thickness for amorphous materials, and nearer 40–50% for high-gloss or semi-crystalline parts. A 3.0 mm wall in ABS wants ribs around 1.5–1.8 mm thick at the base. We explained the reasoning in why uniform wall thickness matters: every local thickening is a reservoir of shrinkage waiting to show up on the surface.
Bosses are the usual offenders. A solid 8 mm boss on a 2.5 mm wall is a sink mark by design. Core it out so the boss wall matches the rib rule, and tie it to the nearest wall with a gusset rather than letting it stand as a lump of material. Where you genuinely need mass, for example for a thread-forming screw, move the boss off the show face or hide it behind a feature line.
The expensive lesson is steel safety. Making a rib thinner after the tool is cut means adding steel back to the cavity: welding, re-machining and re-polishing, often 2–4 weeks and a significant invoice. Making a rib thicker only removes steel and is cheap. That asymmetry is exactly why sink belongs in a Design for Manufacture review, not in a mould trial.
Tool fixes, when the design is fixed
When the part is already released and the tool already exists, you still have options, roughly in order of cost:
- Gate position and size. Sink far from the gate is often a gate that freezes too early. Opening the gate or moving it nearer the thick section lets you pack it for longer.
- Cooling behind the thick feature. A dedicated baffle or bubbler behind a boss core shortens the time it stays hot. Conformal cooling inserts go further, at a price.
- Texture. A VDI 27–30 texture hides sink that a polished surface shows. It is a cosmetic decision, so agree it with the customer rather than slipping it in.
- Gas assist or microcellular foaming. Both pressurise the core of a thick section from the inside. They are powerful, but they add process complexity and need proper validation.
Process fixes that do not create new problems
Once design and tool are right, the process window does the rest. The common mistake is treating hold pressure as the only knob.
- Run a gate freeze study. Weigh parts at increasing hold times. When the weight stops rising, the gate has frozen. Set the hold time a little beyond that point and stop guessing.
- Pack, do not cram. Hold pressure is typically 50–80% of the peak injection pressure. Pushing it higher flashes the parting line and builds moulded-in stress long before it removes deep sink.
- Mould temperature. A hotter cavity surface keeps the skin soft for longer and can let it follow the shrinking core without a visible dent. Raising the mould temperature on the show side by 10–15°C can help, but it costs cycle time.
- Melt temperature. Counter-intuitively, a slightly lower melt temperature can reduce sink because there is less thermal shrinkage to manage. Change it in small steps.
Make one change at a time, and record the result on parts measured the same way each time: a profilometer, or a fixed lamp angle with limit samples agreed with the customer. Otherwise you will be arguing about opinions instead of millimetres.
What good looks like
On a well-designed part in a well-built tool, sink is designed out, verified in mould flow and trials alongside the other common moulding defects, and held in control by a documented process window, not by an experienced setter's intuition on the night shift. The parts on the bench at 4:40 on a Friday look the same as the ones approved at PPAP.
The questions to ask before your next tool is cut
- Have the ribs and bosses been checked against the 50–60% rule for this material and surface finish?
- Has a mould flow analysis predicted volumetric shrinkage, and where it will show?
- Is the gate located to pack the thickest sections, not just to fill the part?
- Is the cosmetic standard for the show face agreed in writing, with limit samples?
If any answer is "not yet", that is the cheapest moment to fix it you will ever have.
