Reference · print it · pin it by the printers

The Calibration Card

Every target number in the course on one page. Phase 1 is complete; Phases 2 and 3 fill in as those lessons land.

🖨 Designed to print• ↻ Grows with each phase• Reference

▲ Never mix these two

MK3S = Marlin. Linear Advance, M900 K, values ~0.02–0.08 for a 0.4 nozzle.
MK4S / Core One = Buddy. Pressure Advance, M572 S.
Different features, different scales, no conversion between them. One profile per printer family.

1The order — never out of sequence

Changing any step voids every measurement below it.
#	Step	Why it sits here
0	The machine (Phase 1)	Everything below measures filament plus machine
1	Temperature	Sets viscosity — moves every number below it
2	Extrusion multiplier	Measured at the temperature just fixed
3	Pressure / Linear Advance	Pressure depends on viscosity and flow
4	Retraction	Ooze depends on residual pressure, which advance changed
5	Cooling	Modifies temperature's effect after the nozzle
6	Max volumetric speed	Depends on temperature; silently clamps speeds
7	Speeds and accelerations	Bounded by all of the above — and can send you back to 3

The one legitimate loop back: advance must be tuned at production accelerations. Print the advance tower with your real quality profile loaded, not a slow one.

2Phase 1 targets, per machine

Belt tension. Three machines, three units — the numbers do not transfer.
	MK3S	MK4S	Core One
Measure	`Calibration → Belt test` read at `Support → Belt status`	by feel — no number reported	`Settings → Manual Belt Tuning` (fw 6.4.0+) or strum + phone tuner
Target	X ≈ 250 · Y ≈ 275 (±15)	low bass string when plucked	both belts 90–98 Hz
Tolerance	under 240 → loosen over 290 → tighten	pinchable, with resistance	the two within 8 Hz of each other
Direction	lower = tighter	—	tighten bolts evenly and alternately
Verify	re-run `Calibrate XYZ`, read skew	cube at high accel, look for ghosting	circles round, not oval

Selftest, first layer, thermal and motion.
	MK3S	MK4S	Core One
Selftest	`Calibration → Selftest`	`Control → Calibrations & tests`
Probe	PINDA — senses the sheet	Loadcell — senses the nozzle
First layer	`Calibration → First layer calibration` manual, per sheet	automatic before every print (loadcell selftest step must be done)
Z adjust	`Settings → Live adjust Z` range −1.500 … −0.200 CCW = closer · persists	long-press knob during first layer does not persist persistent: `Printer Settings → General → Z offset`
Thermal	`Calibration → PID calibration` `Calibration → Temp Model cal`	no user calibration — observe 1 min at 100 % fan and log
Input Shaper	none	stock values fine; frequency must read 35–70 Hz
Extras	linearity correction: `Settings → Lin.Correction → E correct` (only if you see diagonal moiré)	—	record door + vent state observe thermals hot and cold

⌁ Diagnostic thresholds worth memorising

Hotend swing ≤ ±5 °C → PID tuning. > ±5 °C → fan, thermistor or connector fault, not tuning.
Input Shaper frequency outside 35–70 Hz → go back to belts and assembly.
MK3S axis-length error → suspect belt tension before suspecting the axis.
Defect varies by position on the plate → bed shape, not Z offset.
Defect appears partway up the print → not belts; look at chamber heat or accumulated mass.

3The health check

✓ Per printer — now, and after any hardware change

Firmware current and version recorded
Selftest passed
MK3S: Calibrate XYZ run, skew recorded · MK4S/Core One: loadcell step done
Belt tension measured and in spec
First layer calibrated — per sheet
Thermal stability confirmed
MK4S/Core One: Input Shaper frequency inside 35–70 Hz
Nozzle size and rough age recorded
Test cube printed at production acceleration and judged
Everything above written into tuning-log.md with a date

Tick as you go

4The arithmetic

The extrusion equation

volumetric flow (mm³/s) = layer height × extrusion width × print speed

# rearranged, to find the speed that saturates a given flow
speed = flow ÷ (layer height × extrusion width)

# extrusion multiplier, from a measured single wall
new EM = current EM × (nominal wall width ÷ measured wall width)

# total flow is the product of two independent scalars
total flow = M221 (firmware, %) × extrusion multiplier (slicer)

Practical flow ceilings, 0.4 mm nozzle — starting figures, to be replaced by measurements.
Machine	mm³/s	Note
MK3S	~11–13	the slow machine — accept 45–60 mm/s perimeters
MK4S	~22–28	MK4S hotend; bedslinger, so motion limits before flow does
Core One	~22–28	same hotend class as the MK4S; CoreXY, so it can actually use it

5Phase 2 starting values — 0.4 nozzle, 0.2 mm layer

⚠ Entry points, not answers

Every number below is where you start a measurement, not where you stop. Compare against the stock Prusa system profile for that printer before deviating — if a value here differs a lot from stock, trust stock and tune from there. Core One chamber figures are community consensus, not Prusa specification.

PLA
Setting	MK3S	MK4S	Core One
Nozzle, first / other	215 / 210	230 / 225	230 / 225
Bed, first / other	60 / 60	60 / 60	60 / 55–60
Chamber	—	—	vent open, ≤ 30 °C
Extrusion multiplier	0.95–1.00	0.95–1.00	0.95–1.00
Advance	K 0.03–0.05	S 0.03–0.05	S 0.03–0.05
Retract length / speed	0.8 / 35	0.6–0.8 / 35	0.6–0.8 / 35
Fan min / max, bridges	100 / 100, 100 %	100 / 100, 100 %	100 / 100, 100 %
Disable fan, first N	1	1	1
Max volumetric speed	13	24	24
External perimeter	30	45	45
Perimeters	45	120	130
Solid / infill	50 / 70	150 / 200	150 / 220
Travel	180	400	450

PETG — on smooth PEI use a separator or switch to satin/textured, and squish slightly less than PLA.
Setting	MK3S	MK4S	Core One
Nozzle, first / other	240 / 240	245 / 240	245 / 240
Bed, first / other	85 / 90	85 / 85	85 / 85
Chamber	—	—	closed, 35–40 °C
Extrusion multiplier	0.93–0.98	0.93–0.98	0.93–0.98
Advance	K 0.05–0.09	S 0.04–0.06	S 0.04–0.06
Retract length / speed	1.0–1.2 / 35	0.8–1.0 / 35	0.8–1.0 / 35
Z-lift	0.4–0.6	1.5 (stock)	1.5 (stock)
Fan min / max, bridges	30 / 50, 100 %	40 / 60, 100 %	30 / 50, 100 %
Disable fan, first N	2–3	2–3	2–3
Max volumetric speed	8	10–13	13
External perimeter	25	40	40
Perimeters	40	100	110
Solid / infill	40 / 55	120 / 160	120 / 170
Travel	180	400	450

TPU 95A — speed is limited by the extruder, not the hotend. Ironing off; approach retraction from below.
Setting	MK3S	MK4S	Core One
Nozzle, first / other	235 / 230	235 / 230	235 / 230
Bed, first / other	55 / 50	55 / 50	55 / 50
Chamber	—	—	closed
Extrusion multiplier	1.00–1.05	1.00–1.05	1.00–1.05
Advance	K 0.00–0.03	S 0.02–0.04	S 0.02–0.04
Retract length / speed	0.4–0.8 / 20	0.4–0.8 / 25	0.4–0.8 / 25
Fan min / max	50 / 80	50 / 80	40 / 70
Disable fan, first N	1	1	1
Max volumetric speed	1.8–2.5	3–4	3–5
External perimeter	15	25	25
Perimeters	20	35	40
Solid / infill	20 / 25	35 / 45	40 / 50
Travel	120	250	300

The three test blocks · Printer Settings → Custom G-code → After layer change

# L10 · temperature tower — 0.2 mm layers, 50 layers = 10 mm bands.
# Set the profile to the TOP of the range and step DOWN. M104 does not wait.
{if layer_num == 50}M104 S245{endif}
{if layer_num == 100}M104 S240{endif}   ← judge the MIDDLE of each band

# L12 · advance tower. MK3S uses M900 K; MK4S and Core One use M572 S.
# Print with the print profile you actually intend to use.
{if layer_num == 20}M900 K0.02{endif}    ← MK3S only
{if layer_num == 20}M572 S0.02{endif}    ← MK4S / Core One only

# L15 · MVS ramp. Set MVS to 40 first so it can't clamp what you're measuring.
{if layer_num == 30}M220 S150{endif}
{if layer_num == 60}M220 S200{endif}   ← take 85 % of the failure band

### DELETE EVERY ONE OF THESE WHEN THE TEST IS OFF THE BED. ###
### They live in the PRINTER profile and fire on every later print. ###

Diagnostic shorthand for Phase 2.
What you see	Which step	Which way
Clean flat split along a layer line	L10 temperature	too cold — go up a band
Strings, sag, curl, gloss	L10 temperature	too hot — go down a band
Wall measures over nominal	L11 flow	EM down
Pinholes in top surface	L11 flow	EM up
Blob on the outside of a corner	L12 advance	value up
Gap just after a corner or travel	L12 advance	value down
Strings, everything else ruled out	L13 retraction	length up, 0.1 mm at a time
Extruder clicking, ground filament	L13 retraction	length down, immediately
Overhangs droop, bridges sag	L14 cooling	fan up
Layers delaminate, part is brittle	L14 cooling	fan down
Wall goes matte and thin at speed	L15 MVS	ceiling down to 85 % of failure
Infill far slower than the profile says	L15 MVS	at the ceiling — lower layer height
Ghosting downstream of a feature	L16 accel	down — and re-check belts

6Method rules

The habits, compressed.
Rule	Because
One variable between two test prints	Two changes give a result you can't attribute — worse than none, because it feels like data
Banded test prints over separate prints	Same spool, nozzle, hour and humidity — only the variable differs
Dry the filament first	Moisture makes prints non-reproducible, which corrupts temp, flow and retraction at once
Tune advance at production speed	A slow-test value over-compensates and starves corners at real speed
Delete test G-code afterwards	It lives in the printer profile and keeps firing on unrelated prints forever
Log printer, spool, colour, date	Pigment changes flow within a brand; an unlogged number is indistinguishable from a guess
Snap the part to judge bonding	Granular fracture = fused. Clean plane along a layer = laminated. Surface tells you nothing

7Shipping a profile

Naming and scoping

# <Vendor> <Material> @<PRINTER> <nozzle>
eSun PETG @MK3S 0.4      eSun PETG @MK4S 0.4      eSun PETG @COREONE 0.4

# Filament Settings → Dependencies → Compatible printers condition
printer_model=~/.*MK3.*/     and nozzle_diameter[0]==0.4
printer_model=~/.*MK4.*/     and nozzle_diameter[0]==0.4
printer_model=~/.*COREONE.*/ and nozzle_diameter[0]==0.4

# Filament Settings → Custom G-code → Start G-code — NEVER printer start G-code
M900 K0.07   ; MK3S only
M572 S0.05   ; MK4S / Core One only

# Backup, after every session
File → Export → Export Config Bundle

Verify printer_model by slicing and reading the config comments at the end of the G-code. Never tighten a regex you haven't confirmed — an over-strict condition hides the profile silently.

Validation — each failure names the step to go back to.
Criterion	Failure means
Top surface uniform	pinholes or ridges → L11 extrusion multiplier
Corners square	bulge → advance low · gap after corner → advance high → L12
Seam a fine line	L12 advance + L13 retraction, then scarf seams
Layer lines even	degrading upward → L14 cooling / L15 MVS (Core One: chamber)
Snap test granular	clean layer plane → L10 temperature or too much fan

Three prints: dimensional cube (±0.15 mm, and X must agree with Y) · overhang/bridge/stringing combo · a real part of your own. Elephant foot compensation last — Print Settings → Advanced, ~0.2 mm for a 0.4 nozzle. Record the scope: material, machine, speeds, enclosure state, date.

Log every value (m) measured or (i) inherited. Record date, printer, spool including colour, and when it was dried. A profile of (i) values is a hypothesis; a profile of (m) values is a measurement.

⌁ Complete as of 2026-08-05

All four phases are on this card. What it deliberately does not carry is your own measurements — those live in tuning-log.md and in the profiles themselves. When one of your measured numbers disagrees with a starting value here, your number wins; this page is where you began, not where you ended up. Fuller context in the original guide.

The Calibration Card · all lessons · glossary