A failed print is not a verdict; it is data. Read where and how it failed, decide whether the cause sits in the design, the material or the machine, and change one thing at a time.
Lead brief
The fastest way to waste a spool is to change five slicer settings after one bad print. The fastest way to fix it is to measure the failure, form one hypothesis and print one small test.
Observe
Photograph, measure, locate.
Classify
Design, material or machine?
Test
Smallest print that proves it.
Record
Log the change and the result.
Collect the Evidence First
Before touching the slicer, look closely. Where did the problem start — first layer, a specific height, one corner, every overhang? Is it consistent across prints, or random? Measure the part with calipers and compare with CAD. Photograph fracture surfaces under raking light. A clean, flat split along a layer line points to bonding; a jagged break through several layers points to genuine overload; a crack starting at a sharp internal corner points to geometry.
Then classify. A design fault reproduces on any well-tuned printer: a sharp root, a load across layers, a wall too thin. A material fault changes with the spool: wet filament, a brand that bonds poorly at your temperature. A machine fault shows up on every model: loose belts, a partially clogged nozzle, wrong flow.
Failure: layer splitLoad across ZCause: orientation
A flat, glossy split along one layer usually means the load crossed the layers or the bond was weak.
Symptom matrix — likely causes and the smallest useful test
Speed and acceleration, frame resonance, loose belts
Ringing tower
Tension belts, reduce acceleration, input shaping
Drooping overhangs
Angle too steep, insufficient cooling
Overhang angle test
Redesign with chamfers, increase fan
Change One Variable at a Time
If you raise the temperature, slow down and switch filament in one go, a good result tells you nothing about which change mattered. Print the smallest test that isolates your hypothesis — a 20 mm tower, a coupon, a single-wall cube — rather than the full part. A ten-minute test that answers one question beats a four-hour print that answers none.
Dimensional Problems That Look Like Design Problems
When parts don't fit, it is tempting to adjust CAD immediately. First rule out the machine: check flow calibration, belt tension and that steps-per-millimetre values are correct. Only once a calibration cube measures true should you adjust clearances in the model. Otherwise every design carries a hidden correction for a printer fault, and the files stop working when the printer is fixed or replaced.
Lift 1.2 mmABS · open frameDraught side
Strength Failures: Read the Fracture
If a functional part breaks in service, keep both halves. Note the load, how long it lasted and the crack origin. Cracks starting at sharp corners call for fillets. Splits along layers call for reorientation or better bonding. Slow bending before breaking suggests creep and a material change. Sudden brittle snaps in a normally tough material suggest moisture or underheating.
Calibration resources worth bookmarking: Ellis' Print Tuning Guide and the Teaching Tech calibration site both walk through flow, temperature and retraction methodically. PrusaSlicer, UltiMaker Cura, Bambu Studio and OrcaSlicer each include built-in calibration or test objects.
Example Diagnosis Log
Observed — enclosure lid too tight on all four sides; walls measure +0.15 mm.
Hypothesis — over-extrusion rather than design error.
Test — single-wall cube: 0.51 mm against 0.45 mm expected.
Change — flow reduced 12 %; cube re-measured 0.46 mm.
Result — original lid now fits with no CAD change.
“Write down the failure before you fix it. Next month, that note is the fix.”Bench rule
