Orca Slicer Calibration: Every Built-In Test, In Order

Orca Slicer ships nine calibration tests, and the order you run them in decides whether the results are worth anything. The project’s own calibration guide sequences them temperature first, then max volumetric speed, pressure advance, flow ratio and retraction, with cornering, input shaping and VFA marked as advanced follow-ups. Tolerance sits outside that chain as a standalone test you run when you print parts that have to fit together. Run them out of order and you will spend an afternoon measuring one variable through the distortion of another.

This is the companion to the twelve Orca Slicer settings that actually change prints. That guide covers what to set; this one covers how Orca works out the numbers for you, and — the part almost nobody explains — which profile each result gets written into, so your careful tuning survives the next profile update.

Where the menu lives

The Calibration menu sits in Orca’s top menu bar, not inside the settings panel. It only populates once a printer and a filament are selected, because every test slices a generated model against the profile you currently have loaded. There is no separate STL to download, which is the single biggest practical advantage Orca has over the older download-a-tower-from-Thingiverse workflow: the test object is generated from your active profile, and for most tests the winning value flows straight back into it.

Two prerequisites before any of this is meaningful:

1. The bed must be level and the Z-offset correct. Calibration prints are mostly thin, wide, first-layer-dependent geometry. On an untrammed bed they fail for reasons that have nothing to do with the variable under test. Fix that first with the first layer calibration guide.
2. The filament must be dry. Wet filament produces stringing, surface pitting and inconsistent extrusion that will be misread as a temperature or retraction result, and you will bake the compensation for a wet spool into a permanent profile.

The order, and what each test writes

#	Test	What it determines	Written back to	Re-run when
1	Temperature	Nozzle temperature for this filament	Filament profile	New brand, colour or batch
2	Max volumetric speed	The real flow ceiling in mm³/s	Filament profile	New brand; after hotend change
3	Pressure advance	Extrusion lead/lag compensation	Filament profile	New brand; after extruder work
4	Flow ratio	Extrusion multiplier	Filament profile	New brand or colour
5	Retraction	Retraction length and speed	Printer profile (extruder)	Stringing persists after step 1
6	Cornering (advanced)	Jerk or junction deviation	Printer profile (motion)	After firmware or belt changes
7	Input shaping (advanced)	Resonance frequency and damping	Printer firmware, not Orca	After any frame or mass change
8	VFA (advanced)	Speeds that produce surface banding	Printer profile (speed caps)	Rarely; diagnostic
—	Tolerance (standalone)	Hole and contour size compensation	Process profile (Quality)	New nozzle or material class

Two things about that list surprise people. The first is that max volumetric speed sits second rather than being an expert-only extra: it caps every speed downstream of it, so measuring pressure advance or flow against a hotend that is already at its melt limit measures the limit rather than the setting. The second is that flow ratio comes after pressure advance, not before — the wiki’s order, and the opposite of most third-party tuning guides.

In practice, temperature and flow ratio are the two most people ever need. Max volumetric speed earns its place if you print anywhere near your machine’s rated speed. Retraction is conditional. Cornering, input shaping and VFA are the wiki’s own advanced group, for people chasing the last few percent or diagnosing a specific artefact. Tolerance is not part of the sequence at all; run it when you print assemblies.

The “written back to” column is the part worth internalising. Temperature, max volumetric speed, pressure advance and flow ratio are per-filament values: they live with the spool, and a profile named “Generic PLA” that you have tuned for one brand will be wrong for the next. Retraction and cornering are per-printer values that stay put across every material. Tolerance compensation is a process value tied to the nozzle and layer height. Mixing these levels up is why tuning sometimes appears to evaporate: a value saved to the wrong profile gets overwritten the moment you switch material or quality preset.

1. Temperature

Orca generates a tower divided into bands and inserts the temperature changes at the correct layers automatically. You pick the lowest band that still shows clean layer bonding and a smooth wall.

Temperature comes first because it changes the behaviour of everything measured afterwards. Plastic extruded 15°C too cold is under-extruding for thermal reasons, and a flow-ratio test run on top of that will “correct” a temperature fault by permanently inflating the flow ratio. The mechanics of reading a tower — what stringing, sagging bridges and curled overhangs each tell you — are covered in the temperature tower and flow calibration walkthrough.

One nuance the tower does not capture: the best temperature for surface finish is often 5°C below the best temperature for layer strength. For a functional bracket, bias upward from the band you would pick for a display model.

2. Max volumetric speed

This test finds the flow ceiling in mm³/s at which extrusion visibly fails, and it is second in the documented order for a reason: it is the real limit on print speed, and every test after it is measured through it. A 500 mm/s speed rating means nothing if the hotend can only melt 12 mm³/s. Orca prints a wedge whose speed rises continuously along its length, and you read the value off the point where the extrusion starts to thin, gap or go matte.

It is a per-filament value and it varies far more between materials than between brands. Flexible filaments and heavily filled materials sit dramatically lower than plain PLA, as the filament types comparison sets out. Set it and the slicer will cap speeds for you rather than commanding a flow the hotend cannot deliver.

3. Pressure advance

Pressure advance compensates for the lag between commanding extrusion and plastic actually arriving at the nozzle. Without it you get bulges just after corners and thin spots just after them, because the pressure in the melt zone does not track the commanded flow instantly.

Orca offers three test shapes and the right one depends on your firmware:

• PA Line — a set of lines at increasing PA values. The default choice for Marlin and Bambu machines.
• PA Pattern — a denser grid pattern, and the more precise option on Klipper and RepRapFirmware.
• PA Tower — a tall test where you read the value off the height at which corner artefacts disappear. Slower, but easier to judge for some.

Klipper users tuning past what the tower reveals should read Klipper’s own pressure advance documentation, which explains the underlying model, and Ellis’ Print Tuning Guide for the wider tuning sequence.

Note that pressure advance values are firmware-scaled. A Klipper value and a Marlin M900 K value are not interchangeable numbers, so do not copy one across from a forum post about a different firmware.

4. Flow ratio

Orca now recommends the Archimedean-chords “YOLO” test: a single pass that prints eleven blocks across a flow-ratio modifier range of −0.05 to +0.05 in steps of 0.01, added to your current flow ratio. You pick the flattest, most uniform top surface and apply that number. The older monotonic-line method is still available and runs in two passes — a coarse nine-block pass followed by a ten-block fine pass — but the single-pass test gets there in one print, and it is the one the wiki now points at first.

Judge the blocks by top-surface appearance, not by eye-balling the sides. Over-extrusion shows as raised ridges where the top solid infill lines are pushed into each other; under-extrusion shows as visible valleys or pinholes between lines. The two methods also apply their result differently: the single-pass modifier is added to the existing flow ratio, while the two-pass modifier is a percentage the existing ratio is multiplied by. Do not mix a number read off one test into the other’s formula.

An important exception: on Bambu Lab machines, the wiki is explicit that you should not enable the printer’s own flow calibration when running this test, because the firmware handles flow dynamics itself and the two results will fight. Let the printer do it, or turn the on-printer calibration off before testing in the slicer, but do not run both and expect them to agree. This applies to the A1 Mini and its siblings.

5. Retraction

Run this only if stringing survives a properly calibrated temperature. Most stringing that people attribute to retraction is either wet filament or a nozzle running 10°C hot, and adding retraction distance to compensate introduces its own failure mode: clogs from repeatedly pulling molten plastic back into the heat break.

The test prints a tower with retraction length increasing per band; you select the shortest length that eliminates strings. Direct-drive extruders usually land somewhere under 1 mm. Bowden setups need several times that. The full diagnostic order — dry, then temperature, then retraction — is in the stringing and retraction tuning guide.

6 to 8: the advanced three

Cornering tunes classic jerk or junction deviation, controlling how the toolhead handles direction changes. Raise it and corners get faster and rounder; lower it and they get sharper and slower.

Input shaping measures the frame’s resonant frequency and damping to cancel ringing. The critical caveat: this one is not stored in Orca. The result goes into printer firmware, and it must be re-measured after any change to the moving mass — a new toolhead, a heavier bed, even a different spool holder on a bed-slinger.

VFA (vertical fine artefacts) hunts the specific speeds that excite mechanical resonance and produce fine vertical banding on walls. It is a diagnostic rather than a routine step; run it when you have a surface artefact that survives input shaping.

Tolerance: the test outside the sequence

The Orca Tolerance Test prints a set of pins and matching holes at graduated clearances. It answers a question no other test does: what actual clearance do parts from this printer need to fit together? Holes print undersize on almost every FDM machine because the extruded line is laid on the inside of a curve, and the amount varies with nozzle, layer height and material.

The result feeds the hole and contour compensation fields in the Quality section of the process profile. It is worth running once per nozzle size if you print assemblies, bearings, or anything press-fit, and is safe to skip entirely if you print display models.

A realistic schedule

Nobody runs nine tests per spool, and the wiki does not ask you to. In practice:

• Every new brand or colour: temperature, then flow ratio. Fifteen minutes of printing.
• Every new brand, if you print fast or print functional parts: add max volumetric speed and pressure advance, in that order.
• Once per printer: retraction, cornering, input shaping.
• Once per nozzle size, if you print assemblies: tolerance.
• Never, unless you have the symptom: VFA.

Name the filament profile with the brand and colour before you start, or the values you just spent an hour finding will be saved over the next time you load a different spool of nominally the same material.

Where calibration is the wrong tool

Calibration corrects predictable, repeatable offsets. It cannot correct a fault that changes print to print. If results wander between runs, stop calibrating and look at hardware: a partially clogged nozzle, a loose belt, an extruder gear slipping on wet filament, or a bed that moves when the gantry accelerates. Tuning a profile against an intermittent fault produces a profile that is wrong the moment the fault clears.

If you are not sure which category your symptom falls into, the FDM print troubleshooter walks the common failures — stringing, warping, poor adhesion, under-extrusion, layer shifting — down to a ranked list of causes with the specific setting or repair each one calls for.

Calibrate in the documented order, save each value to the profile level it actually belongs to, and the numbers stay correct for the life of the spool.