Learning from Failed Prints
Use a repeatable “symptom → category → one-change test” loop: stop early when the first layer is wrong, identify the first bad layer/feature (not the aftermath), match it to a likely cause, then validate the fix with a small targeted test before you commit to another long print.
TL;DR
When a print fails, find the first bad layer/feature and classify it (first-layer/adhesion, flow, cooling/overhangs, mechanics, supports, or model). Change one major variable, then re-test on a small print (first-layer square, overhang/retraction test, or a cropped model section) before re-running the full job.
What to capture before you touch settings
Your goal is to preserve clues that disappear once you peel the part up or restart the job. Note: (1) first bad layer number or timestamp, (2) where it starts on the part (first layer, a corner, a specific overhang, a thin wall, near the seam, at a certain height), and (3) what the symptom looks like (lifting, gaps, blobs, stringing, droop, layer shift, delamination, “spaghetti”). Take a clear photo from two angles if you can, then record a quick settings snapshot: material and brand/type, dry/not dry, nozzle size, layer height, nozzle/bed temps, speed/accel (if you changed it), fan %, and any recent hardware changes (new nozzle, new bed surface, moved printer, tightened belts).
Fast triage order (don’t skip the order)
- If the first layer is failing, stop early and fix that first. Almost every other problem gets worse on a bad first layer.
- Find the first bad layer or first bad feature. Ignore the mess that happens after; it’s often just a cascade.
- Map the symptom to a category: adhesion/first layer, extrusion/flow, cooling/overhangs, mechanics (shifts/ringing), supports, or model/design.
- Pick the single most likely cause you can test quickly (one knob).
- Run the smallest test that reproduces the failing feature, not the whole part.
- Only after a fix repeats successfully should you restart a long job.
Common failure patterns and the best first fix to try
Corners lifting or part warps off the bed
Likely cause: Weak bed adhesion or uneven cooling/drafts
Fix: Clean bed; verify level/mesh; add brim; block drafts; raise bed temp slightly
First layer not sticking or looks patchy/transparent
Likely cause: Z offset too high, bed contamination, or bed too cool
Fix: Clean bed; lower Z offset slightly; slow first layer; increase first-layer temps
Elephant foot (bulged bottom edge)
Likely cause: Nozzle too close and/or bed too hot (first layer over-squished)
Fix: Increase Z offset slightly; lower bed temp; use elephant-foot compensation
Gaps/weak walls (under-extrusion)
Likely cause: Partial clog, wrong flow, or printing too cold/fast
Fix: Do a cold pull or swap/clean nozzle; verify flow/extrusion; raise nozzle temp or slow down
Blobs/zits on outer walls
Likely cause: Pressure/retraction issues, wet filament, or inconsistent extrusion
Fix: Dry filament; tune retraction; reduce temp slightly; use wipe/coast if available
Stringing between features
Likely cause: Too hot, wet filament, or insufficient retraction/travel strategy
Fix: Dry filament; lower temp; increase retraction within safe limits
Layer separation (delamination)
Likely cause: Poor interlayer bonding from low temp, too much fan, or drafts
Fix: Raise nozzle temp; reduce fan; shield from drafts/use enclosure (as appropriate)
Overhangs droop or curl upward
Likely cause: Not enough cooling for the geometry, or printing too fast/thick on that angle
Fix: Increase part cooling for PLA; slow outer walls; reduce layer height on overhangs
Top surface has holes/rough patches
Likely cause: Not enough top thickness or weak support from infill underneath
Fix: Add more top layers; increase infill density; slow top skin
Layer shift or sudden offset
Likely cause: Belt/pulley slip, collision, or accel/speed too aggressive
Fix: Check belt tension and pulley/grub screws; reduce acceleration/speed; ensure part isn’t curling into the nozzle
How to pick “one change” (and avoid chasing your tail)
During diagnosis, hold everything else steady: same filament spool, same slicer profile, same nozzle, same bed surface. Change one major variable per test: temperature, speed/acceleration, retraction, fan, or Z offset (not two at once). If you change hardware (nozzle, extruder parts, hotend), treat that as a separate experiment and revert settings to a known profile first so you can tell what actually fixed it.
Small tests that save hours
- First-layer square: for adhesion, Z offset, bed cleanliness, and elephant foot.
- Cropped model section: slice only the region that contains the failing feature and print it where it failed on the bed.
- Overhang test: for droop/curl, fan effectiveness, layer height, and outer-wall speed.
- Retraction or temperature tower: for stringing, blobs, and layer bonding tradeoffs.
- Single-wall cube/vase mode segment: for flow/extrusion consistency and dimensional sanity checks.
Debug log template (keep it short, but consistent)
- Part + requirement
- What the part must do; what “good” means
- Symptom
- What you see; attach photos if possible
- First bad layer/time
- Layer number or timestamp when it starts
- Location on part
- Corner, seam, overhang, thin wall, top skin, etc.
- Settings snapshot
- Material, nozzle, layer height, temps, speed, fan
- One change tested
- Exactly one adjustment (include old → new)
- Result
- Better/worse/unchanged; what changed visually