Why Every Calibration Guide Gives You a Different Order

Calibrate in this sequence: temperature, maximum volumetric speed, pressure advance, flow ratio, retraction, cornering, then input shaping. That is the order OrcaSlicer's own wiki recommends  and it puts pressure advance before flow ratio, which is the opposite of what most guides on the first page of Google will tell you.

Both camps have a real argument. The disagreement exists because the word "flow" is being used for two entirely different settings. Separate them and the contradiction dissolves.

The two things everyone calls "flow"

The first is extruder calibration: rotation_distance in Klipper, E-steps in Marlin. It is mechanical and lives at the printer level. It answers one question  when the firmware commands 100 mm of filament, how much actually moves?

The second is flow ratio, also called the extrusion multiplier. It is a slicer setting, specific to a spool, and normally sits within a few percent of 1.0.

Klipper's pressure advance documentation says to tune the extruder rotation_distance and the nozzle temperature before tuning pressure advance. That is the mechanical kind, and it is a genuine prerequisite: if your extruder delivers 8% less filament than commanded, every test downstream measures the wrong thing.

The OrcaSlicer wiki puts Flow Ratio calibration after Pressure Advance. That is the second kind  a small per-filament trim on a machine whose extruder is already correct.

Guides that insist "flow always comes before pressure advance" are, in most cases, repeating the first rule while pointing at the second setting. The advice is not wrong so much as misfiled.

The real ordering rule: each test has to stay legible

The sequence is not arbitrary and it is not about importance. It is about whether you can still read the result. Every calibration test is judged by looking at a printed artefact. If an earlier, uncalibrated variable is writing its own defects onto that artefact, your reading is contaminated.

Temperature goes first because it sets melt viscosity, and viscosity moves everything downstream. OrcaSlicer's max volumetric speed page makes the dependency explicit: if you intend to run higher flow, raise the nozzle temperature toward the top of the filament's range, and use a temperature tower to find where that is.

Maximum volumetric speed goes second because it establishes the ceiling every later test operates under. Orca's defaults run 5 mm³/s to 20 mm³/s in 0.5 steps; you measure the height at which quality collapses and calculate start + (height × step). Their worked example gives 5 + (19 × 0.5) = 14.5 mm³/s. Then take 10–20% off, because the test is a best case that ignores retraction and clog risk. Worth noting: the same wiki warns that brand and even colour shift the maximum flow rate within a single material. Two spools both labelled PLA are not interchangeable here.

Pressure advance goes third because its result is read at corners, and corners are the one feature nothing else has corrupted yet. Klipper's tuning tower deliberately slows the toolhead through corners — SET_VELOCITY_LIMIT SQUARE_CORNER_VELOCITY=1 ACCEL=500 — to exaggerate the pressure effect. Typical values land between 0.050 and 1.000, with the high end usually reserved for long bowden setups. Above roughly 0.200 combined with high acceleration, the extruder may not have the torque to keep up and will skip. Klipper's guidance when the test is ambiguous is blunt and worth following: prefer the lower value.

Flow ratio goes fourth because it is judged on the smoothness of a top surface — and corner blobbing from an untuned pressure advance lands on that same surface. Calibrate flow first and you are grading a top surface that pressure advance is still actively damaging. Orca's current YOLO method sweeps −0.05 to +0.05 in steps of 0.01, with a tighter −0.04 to +0.035 at 0.005 for the perfectionist preset; the new value is simply the old ratio plus or minus the modifier on the best-looking block.

Retraction goes fifth, and OrcaSlicer says why in as many words: doing it after flow and pressure advance "ensures that the printer is already set up for optimal extrusion." There is a second reason. Klipper notes that pressure advance itself reduces ooze during non-extrude moves. Tune retraction first and you will dial in a large retraction to solve stringing that pressure advance was about to fix for free.

Cornering and input shaping come last because they are motion problems, not extrusion problems. They change how the toolhead moves, not how much plastic leaves the nozzle.

The sequence, and what breaks if you invert it

One honest caveat: neither OrcaSlicer nor Bambu Lab publishes a measured magnitude for how much an untuned pressure advance actually shifts a flow ratio reading. Treat this ordering as a well-reasoned dependency chain drawn from the official documentation, not as a quantified experimental result. If you have printed the tests both ways and seen no difference on your machine, that is a legitimate observation, not a mistake.

The Bambu exception is real — and it is not a contradiction

Bambu Lab's wiki says something most third-party guides skip entirely: for a regular filament, you usually will not need to run flow rate calibration at all. Their printers and official filaments are pre-calibrated, mechanical tolerances are tight, and the flow ratio is normally stable.

The wiki goes further. It tells you to consider flow rate calibration only if you have already performed other calibrations  it names Flow Dynamics Calibration specifically — and the defect persists. Flow Dynamics is simply Bambu's term for pressure advance.

So on a Bambu machine the order is: flow dynamics first, then flow rate only if a real defect survives it. That is the same dependency chain as everyone else's, with one step marked optional. Their manual mode runs a coarse pass across 80–120% in 5% steps, then a fine pass across 91–100% in 1% steps.

One practical trap: if you run OrcaSlicer's flow ratio test on a Bambu printer, the Orca wiki warns you to leave the 'Flow calibration' option unticked. Leave it on and the printer's own routine competes with the test you are trying to read.

Where automatic calibration quietly gives you a bad number

This is the part that costs people the most time, because the failure is silent  the machine returns a confident number that happens to be wrong.

Bambu's automatic flow rate calibration works by measuring the printed calibration pattern with the Micro Lidar. Their wiki states plainly that filaments which are transparent or semi-transparent, contain sparkling particles, or have a highly reflective finish may not be suitable for this calibration and can produce less-than-desirable results. It also acknowledges that results may vary between individual calibration runs, and that accuracy is still being improved through firmware updates.

That warning covers a large share of what people actually buy. Sparkle, silk, gradient and translucent filaments are exactly the spools most likely to be handed a bad automatic result  and exactly the spools where a visibly poor top surface gets blamed on the filament instead of the calibration.

What you actually need to re-run, and when

Most people either calibrate obsessively or never again. Neither is right, because the different calibrations have different scopes.

• Pressure advance is per printer, per nozzle, per filament. Klipper's documentation is explicit that filament from different manufacturers, or with different pigments, can require significantly different values - so a new brand or even a new colour is a legitimate reason to re-run it.

• Flow ratio is per filament, and on a pre-calibrated machine with a well-made spool it may never need touching.

• Max volumetric speed is per filament, and matters most when you are pushing speed. It is also brand- and colour-sensitive.

• Input shaping and cornering are per machine. Re-run them after mechanical work  a belt change, a new toolhead, moving the printer onto a different surface - not after a filament change.

• Temperature is per filament type, and worth revisiting if you have changed nozzle size or moved to a much higher flow rate.

A sensible working schedule: full sequence once on a new printer with your primary filament. After that, pressure advance on any new brand or colour, flow ratio only when top surfaces look wrong, and motion calibration only after you have physically changed something.

The short version

Temperature, then max volumetric speed, then pressure advance, then flow ratio, then retraction, then cornering and input shaping. Make sure your extruder's mechanical calibration is correct before any of it - that is the "flow" the older guides were talking about.

If you own a Bambu printer, run flow dynamics and stop there unless a defect survives. And on any sparkle, silk, translucent or gradient spool, do not trust the automatic flow number - Bambu's own documentation tells you not to. Ten minutes with a manual calibration block will beat a confident bad reading every time.