Most troubleshooting on a molding floor still happens the same way it did thirty years ago: a defect shows up, somebody who “knows the machine” reaches for the same three knobs they always reach for, and the part either clears up or it doesn’t. When it clears up, nobody writes down why. When it comes back next shift, the cycle starts over.
That approach treats every defect as a machine-setting problem. It isn’t. A defect is a symptom, and the same symptom can come from the material, the mold, the process, or what happens to the part after it ejects. Reaching for pack pressure when the real problem is a wet resin lot doesn’t just fail to fix the part — it buries the cause under a new variable.
The method below is the one the historical WJT Associates training built its troubleshooting module around: sort the defect into a family first, then chase the root cause inside that family. This article doesn’t reproduce that proprietary curriculum. It applies the same logic — diagnosis before adjustment — in original form, and adds the discipline that makes it work on a real floor.
Why “turn a knob and see” fails
The trial-and-error habit survives because it occasionally works, and when it does the technician gets credit for it. But it has three structural problems that compound over a production run:
- It changes more than one variable. A processor under pressure rarely changes one setting, waits for the process to stabilize, and reads the result. They change three, the part looks better, and now nobody knows which change mattered — or what the other two are quietly doing to dimensions.
- It hides the real cause. If short shots are coming from a worn non-return valve and you fix them by cranking shot size and pack, the valve keeps getting worse. The day the extra shot size can’t cover for it, you get a process that swings between short and flashed on the same setup.
- It doesn’t transfer. A fix that lives in one technician’s head is worthless on the next shift and worthless when the job moves to another press. Knowledge that isn’t written down isn’t knowledge the plant owns.
Root-cause troubleshooting is slower for the first ten minutes and faster for the next two hours. That trade is almost always worth making.
The five defect families
Sorting the defect into a family is the step that does the most work, because it tells you where to look before you touch anything. Nearly every common molding defect falls into one of five families, and each family points to a different set of causes.
| Defect family | Typical defects | Where the cause usually lives |
|---|---|---|
| Material-related | Splay, brittleness, black specks, color streaks, inconsistent shrink | Drying, contamination, regrind ratio, lot-to-lot viscosity, purge discipline |
| Filling | Short shots, weld/knit lines, jetting, flow marks, gas burns | Fill speed and pressure, venting, melt temperature, gate condition, check ring |
| Pack and hold | Sink, voids, dimensional drift, flash, sticking in cavity | Pack pressure and time, hold profile, gate seal timing, cushion stability |
| Cooling | Warp, ejection marks, stress cracking, long cycle, dimensional variation cavity to cavity | Cooling time, water flow and temperature, water-line blockage, uneven tool temperature |
| Post-mold | Warp that appears later, assembly fit problems, dimensional change after hours/days | Part handling, fixturing, annealing, moisture pickup, regrind in a later lot |
A defect can have causes in more than one family — sink is the classic example, since it can be a pack problem or a cooling problem. That’s exactly why the family step matters: it stops you from treating sink as a pack problem by reflex when the water lines are half-blocked and the part is ejecting too hot to hold its shape.
Separate the four sources before you adjust
Inside whatever family the defect falls into, every cause traces back to one of four sources. Naming the source is what keeps you from changing the wrong thing.
- Material — drying, contamination, regrind level, a resin lot at a different melt index than the one the process was built on. A material cause will often appear right after a lot change, a color change, or a startup, and no machine setting created it.
- Mold — wear at the parting line or shut-offs, plugged vents, a worn gate, blocked or scaled water lines, a sticking ejector. A mold cause tends to show up in the same location shot after shot and gets worse with run time, not with setting changes.
- Machine/process — a setting that drifted, a check ring that no longer seals, a transfer position that moved, a heater band reading set point but not delivering melt temperature. Process causes track with what was changed and when.
- Post-mold — handling, packing parts too hot, stacking that warps them, moisture pickup before assembly. These causes produce parts that were good at the press and bad at the customer.
The single most useful question on the floor is still the cheapest one: what changed? A defect that appeared at a startup, after a lot change, or right after someone “improved” a setting has already narrowed itself to a source before you’ve picked up a tool.
The change-one-variable rule
This is the rule that separates troubleshooting from gambling, and it is the one most often broken under schedule pressure:
Change one variable. Move it a meaningful amount. Let the process stabilize for enough shots to read a real signal. Write down the change and the reason. Then decide.
“Enough shots” matters because a barrel and tool take time to reach a new steady state — reading the third shot after a temperature change tells you almost nothing. “Write it down” matters because a change you can’t reconstruct is a change you can’t undo, and three undocumented changes stacked together are how a process drifts into a state nobody designed.
If you genuinely have to make several changes to keep a job running, that is a decision to abandon diagnosis for the moment and just produce — which is sometimes the right call. But be honest that you’ve made it, and plan to come back and find the actual cause when the schedule allows. A process held together by stacked compensating adjustments is a process waiting to fail at the worst possible time.
A cause-and-remedy quick reference
Once you’ve sorted a defect into its family and isolated the source, it helps to have a starting list of the usual suspects and the process-side moves that address them. Treat this as a starting point for one-variable-at-a-time testing, not a recipe to apply all at once — and remember that some causes are design or material problems that no setting will fix.
| Defect | Common causes | Process-side remedies to test |
|---|---|---|
| Streaks / flow marks | Melt temp too low; injection speed too high; mold temp too low; poor gate location | Raise melt temp; lower injection speed; raise mold temp; (design) relocate or add a gate |
| Silver streaks (splay) | Insufficient drying; resin decomposition; excess residence time; injection too fast; trapped air; poor venting | Dry thoroughly first; lower melt temp; reduce cycle/residence; lower injection speed; raise back pressure; add venting |
| Sink / shadow | Insufficient hold pressure or time; thick sections; melt/mold temp issues | Raise or extend hold; tune temps; (design) core out thick areas |
| Short shot | Insufficient pressure or speed; poor venting; melt temp too low | Raise pressure/speed; improve venting; raise melt temp |
Two disciplines keep this table from becoming a knob-twisting exercise. First, change one variable at a time and confirm against the process window before moving on. Second, separate the design and material causes from the process causes — splay from wet resin clears in the dryer, not at the press, and sink from a section that’s simply too thick is a tooling conversation, not a hold-pressure one.
It also pays to make sure everyone is naming the defect the same way before hunting its cause. Terms like sink, gas streak, weld/knit line, short fill, stress whitening, flash, and discoloration should mean one agreed thing on the floor — otherwise two people “fixing splay” may be looking at two different defects.
A structured diagnosis sequence
When a defect appears on a running job, this sequence finds the cause faster than reaching for the usual knobs:
- Name the defect and the family. Sink, short, splay, flash — which family does it belong to?
- Ask what changed. Startup, lot change, color change, mold pull, shift change, a logged setting change in the last 24 hours.
- Compare to the last approved run. Pull the setup sheet. Do current settings match? Is cushion stable over the last 10–20 shots? Is melt temperature verified by an actual shot check, not just set point?
- Check the cheap material and mold causes. Drying status, regrind ratio, vent and parting-line condition, water flow. Many “process” defects are a plugged vent or a wet hopper.
- Isolate the source. Material, mold, machine, or post-mold — commit to one before adjusting.
- Change one variable, document it, read the result. Only after steps 1–5.
Plants that work this sequence stop trading one defect for another. The reflex fix — drop pack to kill flash, then chase the short shots and sink it creates — is exactly what the sequence is built to prevent.
It’s worth remembering that the same defect reads very differently from the other side of the transaction. A buyer sees the symptom on a finished part on an incoming inspection report; the molder sees the root cause on the floor. Those two perspectives rarely meet, and both diagnoses are stronger when they do.
For structured, vendor-neutral training that goes deeper on scientific troubleshooting method, the Society of Plastics Engineers (4spe.org) remains the standard reference point in the industry.
FAQs
What’s the difference between a defect family and a root cause?
A defect family is the category a symptom belongs to — material, filling, pack/hold, cooling, or post-mold. The root cause is the specific thing inside that family that’s actually producing the defect. The family tells you where to look; the root cause is what you find when you look there. Sorting into the family first is what stops you from chasing a pack-pressure fix for a cooling problem.
Why is “change one variable at a time” so important if it’s slower?
Because if you change three things and the part improves, you’ve learned nothing you can reuse — you don’t know which change mattered, and you don’t know what the other two did to dimensions you weren’t watching. One documented change at a time is slower for ten minutes and far faster over the life of the job, because it builds knowledge you can write down, transfer to the next shift, and carry to another press.
How do I tell a mold problem from a process problem?
Look at location and history. A mold cause usually appears in the same spot shot after shot and gets worse the longer the tool runs — worn vents, a tired shut-off, scaled water lines. A process cause tracks with what was changed and when: a setting that drifted, a lot that switched, a check ring that stopped sealing. When a defect shows up right after a startup, a material change, or a setting “improvement,” start with the process and material before you blame the tool.
Where does scientific molding fit into this?
Scientific (or decoupled) molding is the toolset that makes the diagnosis precise — separating fill from pack, finding the gate-seal point, establishing a real process window so you can tell drift from noise. The defect-family method is the map; scientific molding is how you read the terrain accurately once you know which part of the map you’re standing on.