3D Printer FYICalibration answers “does the machine do what the g-code asked?” Tuning answers “does this filament look right on this machine?” If you invert the order, you will save a slicer profile that encodes a loose belt.
The two free curricula most of the community actually uses are Ellis’ Print Tuning Guide and Teaching Tech’s calibration page. OrcaSlicer, PrusaSlicer, and Bambu Studio expose the same ideas with different labels.
Calibration vs tuning: what each means
| Word | Example |
|---|---|
| Calibration | 100 mm requested extrusion is ~100 mm at the gear; probe offset is saved |
| Tuning | This Polymaker PLA wants a slightly different flow than the generic PLA profile |
Mechanical baseline before touching slicer settings
Belts seated, pulleys tight, sheet clean, no crash damage. See maintenance and first setup. If the first layer cannot repeat, you are not ready for a retraction tower.
First layer / Z offset / mesh
Do Ellis’s first-layer squish procedure for your firmware (First Layer Squish). Save the offset. Then leave it alone while you tune flow.
Extrusion sanity: diameter, flow/extrusion multiplier
- Measure filament diameter in a few spots if you care about flow accuracy; cheap spools wander.
- Calibrate e-steps / Klipper
rotation_distance(Ellis extruder calibration). - Then tune extrusion multiplier / flow per filament (Ellis EM). Ellis treats EM as aesthetics-first on top surfaces and notes it is per filament, not once per printer.
In Orca this is often “Flow”; in PrusaSlicer “Extrusion multiplier”; in Cura “Flow.” They are the same knob with different names.
Temperature test
Use a temperature tower for that filament, after flow is close. The “best” temperature is the one that bonds layers without sagging the tower’s letters — it will not match a YouTube overlay.
Retraction only after extrusion/temp are sane
Retraction hides extra filament during travel. If flow is high, you will “need” huge retraction and still blob. Tune retraction on a stringing test after EM and temp. Bowden needs more length than direct; do not copy a number across architectures.
Pressure advance / linear advance where supported
Klipper Pressure Advance and Marlin Linear Advance reduce corner blobs. Ellis documents pattern / tower methods (Pressure Advance). Vendor printers may expose this as “pressure advance” in the filament profile (Bambu, Prusa). Calibrate it after EM.
Speed and acceleration without chasing benchmark numbers
Speed is limited by volumetric flow (hotend melt rate) and by ringing. Input shaper can raise acceleration without turning the printer into a drum (Klipper Resonance Compensation). A printed “max speed” flex video is not a calibration.
Dimensional accuracy
Print a calibrated cube, measure X/Y/Z, then use slicer shrinkage / scale per axis if your slicer has it. Do this after flow and temperature, not before.
Bridging/overhangs/supports
Cooling and speed dominate bridges. If PLA bridges fail with the part fan at 0%, that is a test problem. If they fail at full fan, look at geometry and support interface settings — not a new extruder.
Saving known-good material/printer profiles
Duplicate Printer-PLA-baseline before every experiment. When a test wins, rename it with filament brand and date.
When tuning becomes cargo cult
Stop if you are changing five sliders because a screenshot of someone else’s Orca “maxed” profile looked faster. If the defect has a name, look it up in troubleshooting and change the matching control only.