Almost every FDM print failure traces back to one of six root causes: poor bed adhesion, the wrong temperature, wet filament, a nozzle clog, a mechanical motion fault, or a mismatched slicer profile. The hard part of 3D print failure diagnosis is not fixing any one of those — it’s telling them apart, because several of them produce near-identical symptoms. Thin walls and a clicking extruder can mean a partial clog, wet filament or plain under-extrusion. Get the diagnosis wrong and you spend a week “fixing” the wrong thing.
The reliable method is triage by when and where the print failed, then changing one variable at a time. This guide walks the common defects by their signature, the mechanism behind each, and the diagnostic tests that separate look-alike faults — including the traps that send people down the wrong path.
Triage: match the symptom to the layer it appeared on
Before touching a setting, look at the print and ask where in the build it went wrong. A failure in the first three layers is a different animal from one that appears 30 minutes in. This table is the fast lane:
| What you see | Likely cause | First thing to check |
|---|---|---|
| Not sticking at all, corners curling off the bed | First-layer adhesion / warping | Z-offset and plate cleanliness |
| Bottom 2–3 layers visibly wider than the rest | Elephant’s foot | Over-squished first layer / bed too hot |
| Nozzle dragging through plastic it already laid | Z-offset too low | Z-offset — rule this out first |
| Layers split cleanly along a layer line | Layer separation / delamination | Nozzle temp, cooling |
| Everything above a height offset sideways | Layer shifting (mechanical) | Belt tension |
| Thin, gappy walls; sparse infill; weak bonding | Under-extrusion | Manual extrusion test |
| Lumpy, oversized surfaces | Over-extrusion | Flow rate / filament diameter |
| Hair-like strands between parts | Stringing / oozing | Retraction, temperature, wet filament |
A Z-offset problem deserves its own warning: a nozzle set too low masquerades as half a dozen other failures, so it’s worth ruling out before anything else. If the tip is ploughing through the previous layer, no amount of temperature or retraction tuning will save you.
If you’d rather have this done for you, our Diagnose tool identifies the defect from a photo and hands back concrete slicer changes — but understanding the mechanism below means you’ll know why the recommendation is right.
First-layer adhesion and the Z-offset trap
An unlevel bed or wrong Z-offset is the single most common root cause of early failure. A nozzle too far from the bed, a dirty surface, or wrong temperatures starve the first layer of grip. Work through it in order:
- Clean the plate with IPA — oils from fingers are invisible and lethal to adhesion.
- Lower the Z-offset in 0.02 mm steps until the first layer squishes into a continuous surface.
- Raise the bed temperature if needed.
- Slow the first layer to 20–25 mm/s, with a first-layer height of 0.2–0.3 mm.
Temperature is a double-edged sword here: a cold bed can release a part, but an overly hot bed softens the base and causes elephant’s foot. If your bottom layers bulge outward, lower the bed temperature 5–10 °C once the first few layers are down, or raise the Z-offset slightly — an over-squished first layer is the usual culprit. For exact numbers in a Prusa workflow, see our guide to first layer settings in PrusaSlicer.
Warping: differential cooling, not bad luck
Warping happens when differential cooling induces residual stress, pulling the print off the plate. It is driven by nozzle temperature too low, part cooling too strong, cold ambient air, or a material that shrinks heavily as it cools. ABS and ASA are the classic offenders because they shrink significantly on cooling.
Dial the environment and the material together:
| Material | Bed temp | Part cooling fan |
|---|---|---|
| PETG | 70–85 °C | Moderate |
| ABS / ASA | 100–110 °C | 0% (or ≤20%) |
| PC | 110–120 °C | Low |
Add a brim or raft, use an enclosure for high-shrink materials, and keep the room stable at around 20–25 °C with low humidity. Two mistakes to avoid: raising every temperature at once (or adding adhesive) before you’ve calibrated, and opening an enclosure mid-print to peek — the thermal shock can warp the next layers. Calibrate one variable at a time or you’ll never know which change worked.
Stringing: oozing during travel moves
Those fine hairs between parts are molten filament oozing from the nozzle during travel, most often from insufficient retraction, excessive print temperature, or slow travel. The fixes are well established:
| Setting | Direct drive | Bowden |
|---|---|---|
| Retraction distance | 0.5–2.0 mm | 4–7 mm |
| Retraction speed | 20–45 mm/s (up to 60 if quality allows) | |
| Travel speed | 150–200 mm/s | |
Drop the print temperature in 5 °C steps until the strings disappear. Enable “Avoid Printed Parts When Traveling” and, for tall parts, Z-Hop When Retracted to cut exposed travel moves. PETG is a special case — it’s viscous, so lower the retraction speed; pulled out too fast it strings worse.
The critical direct-drive mistake is over-retracting. Anything above 2–3 mm on a direct-drive setup pulls cooled filament into the heat break, causing heat creep and eventual jams — you trade stringing for a clog. And if retraction, temperature and travel speed are all optimised and stringing persists, the filament is almost certainly wet. Dry it before you change another thing.
Under-extrusion: an outcome, not a single fault
Under-extrusion means the nozzle lays down less plastic than the slicer planned — gaps between lines, thin walls, sparse infill, layers that barely bond. It’s an outcome with several possible root causes: a partial clog, wrong slicer inputs, feed slip, heat creep, or a first-layer clearance problem.
When under-extrusion appears suddenly, the most common cause is a partial nozzle clog — it restricts flow without stopping it, producing exactly that starved pattern. The quickest diagnostic is a manual extrusion test: heat the nozzle and push filament through by hand. If it comes out straight and consistent, the nozzle is clear. If it curls, comes out thin, or needs significant force, it’s partially clogged.
The classic trap: a temperature tower shows less under-extrusion at 220 °C than at 200 °C, so you conclude “I need to print hotter.” But if the real problem is a partial clog, printing hotter just masks the symptom while the blockage keeps getting worse.
Other contributors worth checking: Bowden tube friction (gives position-dependent under-extrusion), wrong filament diameter (measure with calipers at ~10 points along a length), and a worn brass nozzle (gradual decline over months — replace it, ideally with hardened steel). Our defect-by-defect repair guide covers each in depth.
Clog versus heat creep: let the clock decide
These two feel identical in the moment but need opposite fixes, and the timeline tells them apart. A partial clog is immediate — the print under-extrudes from the start or worsens steadily. Heat creep is time-dependent: the print starts perfectly, under-extrudes 20–40 minutes in, and recovers after the hotend cools down.
The mechanism behind heat creep is heat travelling up the filament before it reaches the melt zone; the filament swells in the cold zone and jams. Confirm it by touching the top of the heatsink where the heat break enters — if it’s uncomfortably warm (above ~50 °C), you have heat creep. The usual trigger is printing low-temp material like PLA in a fully enclosed, hot chamber.
Why the distinction matters: the standard clog fix does nothing for heat creep. For a partial clog, do a cold pull — heat to 170 °C for PLA, cool to 90 °C, then pull firmly. For heat creep, the first check is the hotend/heatsink fan, not the nozzle. And don’t swap the nozzle without clearing the heat break; debris often backs up into it, so a fresh nozzle alone won’t fix a deeper blockage.
Layer shifting: a mechanical fault, not an extrusion one
Layer shifting is a whole-axis displacement — the nozzle still deposits material, but later layers no longer align with earlier ones. That makes it distinct from under-extrusion (which changes line volume), ringing (repeated ripples rather than a clean offset), and Z-banding (periodic vertical variation rather than a sudden sideways jump).
The number-one cause is belt tension — loose or misaligned motor pulleys produce the classic staircase shift. Pluck the belt like a guitar string: it should give a low, clear tone (roughly 60–90 Hz), not a floppy rattle. Secondary causes:
- Stepper driver current (VREF): undervolting causes missed steps.
- CoreXY imbalance: both belts must be at identical tension — a mismatch torques the gantry, causing binding and shifts.
- Bed-slingers: prone to Y-axis shifts because the heavy bed must accelerate and decelerate.
- Nozzle collision with a warped, curled part.
- Corrupted G-code or a failing SD card: if the shift always lands at the same layer of the same file, reslice and print from different media.
One mindset note: if layer shift only appears on faster profiles and vanishes when you reprint slowly, the real fix is belt tension or stepper current — not printing everything slowly forever.
When live failures matter more than post-mortems
Everything above is root-cause diagnosis after the fact. On a single machine that’s fine. Once you’re running several printers or selling parts, a failure you don’t catch for six hours is wasted filament and a missed despatch. That’s where Printer Farm earns its keep: camera-based AI failure detection raises an alert the moment a print goes wrong, and the job queue only dispatches to a printer with a confirmed clear bed and the right filament loaded — so a failure doesn’t cascade into the next order. It runs the whole job end to end, from a web order through works orders, costing and a numbered delivery note. For the difference between catching a failure live and diagnosing its cause, see our breakdown of live detection versus root-cause diagnosis.
FAQ
How do I tell under-extrusion from wet filament?
They look identical — thin walls, a clicking extruder, gaps. Run the manual extrusion test first: if filament pushes through straight and consistent, the nozzle is clear and the cause is elsewhere. If your retraction, temperature and travel speed are all optimised and you still get stringing or starved layers, the filament is almost certainly wet. Dry it and retest before changing settings.
Why does my print fail 30 minutes in but start perfectly?
That timeline points to heat creep, not a clog. Heat soaks up the heat break over 20–40 minutes, swelling the filament until it jams, then recovers once the hotend cools. Touch the top of the heatsink — above ~50 °C confirms it. Check the hotend fan and improve cooling rather than doing a cold pull.
What causes layer shifting and how do I stop it?
Most often loose belt tension or an insecure pulley. Pluck the belt for a low, clear tone around 60–90 Hz. Also check stepper current, confirm both belts match on a CoreXY, and — if the shift lands at the same layer every time — reslice and print from different media in case the G-code is corrupt.
Can I diagnose a failed print from a photo?
Yes — our Vision-AI Diagnose tool identifies the defect from an image and returns specific slicer changes. For catching problems before they waste filament, run your file through the G-code pre-flight checker first.
Related: 3D Print Farm Cost Per Part: The Formula, the Numbers and Where the Money Actually Goes
