⬢ prusa-profile-tuning

Phase 2 · The filament · L11

Extrusion multiplier

One hollow box, a pair of calipers, and the only Phase 2 number you get to compute rather than judge by eye.

⏱ ~45 minutes, twice• 📏 Digital calipers required• Phase 2
◎ Why this lesson

Every other Phase 2 measurement is a judgement — which band looks better, which corner is cleaner. This one is arithmetic. You print a wall of known intended width, measure what actually came out, and divide. That makes it the most trustworthy number in the whole course, and it is worth having early, because advance, retraction and MVS are all judged on the appearance of extruded lines — and lines that are 6 % too fat look wrong for reasons that have nothing to do with what you're tuning.

The extrusion multiplier is a scalar on how much filament gets pushed for a given commanded volume, where 1 = 100 %. It lives at Filament Settings → Extrusion Multiplier, and Prusa note that adjusted values usually land between 0.9 and 1.1.1

It exists because the slicer's arithmetic assumes a filament that is exactly 1.75 mm in diameter, exactly round, and behaves exactly like the reference material. Yours is none of those things — and the deviation is per-brand, per-colour and per-batch.

▲ If your answer lands outside 0.9–1.1, stop

Prusa's stated range isn't a suggestion about taste, it's a sanity check. A computed EM of 0.78 or 1.24 almost never means your filament is unusual. It means something upstream is wrong: firmware flow isn't at 100 %, the extruder is skipping, the nominal width you divided by wasn't the one the slicer used, or the wall you measured wasn't a single perimeter. Go and find that before you write the number down.

1The M221 trap — deal with this first

There are two independent ways to scale the same thing, and neither knows the other exists. Prusa state it plainly: total flow rate = firmware flow multiplier (M221) × extrusion multiplier in PrusaSlicer, and "adjusting one of those two settings will not influence the value of the other".1

The relationship you have to keep straight
total flow = M221 (firmware, %) × EM (slicer, ×)

# firmware at 95 %, slicer at 0.95, and you have
# quietly corrected the same 5 % twice: 0.90 total

So before you print anything: get the firmware side to a known value and leave it there. Measuring with an unknown M221 in play produces an EM that is correct only while that unknown persists — and firmware flow is not saved in your profile, so it will silently change under you.

Open your machine. This is the one place where the three printers genuinely differ.

MK3S · and the legacy line in its start G-code

During a print, the LCD's Tune menu exposes a live flow percentage. Confirm it reads 100 before you judge anything. Exact menu wording moves between firmware versions, so read your machine rather than trusting a screenshot.

The MK3S-specific catch: its stock system printer profile emits a conditional M221 in start G-code — a legacy correction that sets flow to 95 % at normal layer heights and 100 % only at very fine ones:4

M221 S{if layer_height<0.075}100{else}95{endif}
⚠ Verify this on your own machine, don't take it on trust

Slice anything for the MK3S, open the G-code in a text editor, and search for M221. Whatever you find there is the truth for your firmware and profile version. This is the same habit as reading printer_model in Lesson 16: verify the string, don't guess it.

Leave the line alone. It is part of the system profile and Prusa's MK3S numbers are built around it. Just don't correct for it a second time in the slicer. Your measured wall already contains it, so the EM you compute is correct as long as that line stays — which is another reason to record in your log that it was present.

MK4S · Buddy, no legacy correction

During a print, the Tune menu on the touchscreen exposes a flow percentage. Confirm 100 %.

Outside a print, the reliable move is to send M221 S100 explicitly. M221 is documented on Buddy firmware as "Set extrusion percentage", taking S<percent>.2

M221 S100 ; firmware flow to a known state before measuring

There is no MK3S-style legacy M221 conditional in the MK4S start G-code. Slice something and confirm that for yourself — if you find one, it's yours, not Prusa's, and it needs explaining before you measure.

Core One · as MK4S, plus one measurement-hygiene note

Identical to the MK4S: Tune menu for the live value, M221 S100 to set it explicitly.2 No legacy conditional in start G-code.

The enclosure note: a hollow single-wall box in a warm chamber can soften enough on the upper layers to distort slightly, which shows up as a wall that measures thicker at the top. Let the box cool completely to room temperature before touching it with calipers, and if you're measuring PLA, run the test with the vent open.

If the top and bottom of the same wall disagree by more than about 0.02 mm, that's not filament variation — that's the part having moved, and you should reprint it cooler or vented rather than averaging through it.

2Print something with exactly one wall

The whole method rests on the printed wall being one perimeter thick and nothing else. Two perimeters, or a perimeter plus a sliver of gap fill, and you are measuring the slicer's overlap logic instead of your filament.

Prusa offer two routes: a Precise Method using a downloadable cube printed in vase mode and measured with calipers, and a Visual Method judged by eye.1 Take the precise one. The visual method is for people without calipers, and you are not tuning three printers by eye.

Prusa's cube, vase mode
# upside
one perimeter, guaranteed
no seam to avoid
designed for this test

# downside
vase mode is a different
extrusion regime to your
real prints
Hollow box, perimeters = 1
# settings
30 x 30 x 12 mm
perimeters 1, infill 0
top 0, bottom 1
ironing off, vase off

# upside
prints the way your
real parts print
⬇ Download · the box

em-single-wall-box.stl — the exact 30 × 30 × 12 mm hollow box this lesson measures. Slice it with perimeters 1, infill 0, top layers 0, bottom layers 1, ironing off, vase off — the settings do the single-wall work, the model is just an honest prism.

Either works. This course leans to the hollow box because it keeps the extrusion regime closer to normal printing, but if you take Prusa's cube, take Prusa's whole procedure with it and don't mix the two.

⚠ Read the nominal width; do not assume 0.45

Prusa's formula divides by an extrusion width of 0.45 mm, which is their default for a 0.4 mm nozzle at 0.1–0.2 mm layers.1 But it is a setting, not a constant. Open Print Settings → Advanced → Extrusion width → Perimeters and read the value your profile actually uses. If it says 0.45, fine. If it says 0.42, or a percentage, or is left at 0 (meaning "derive it"), you need the real number — dividing by the wrong nominal produces a confidently wrong EM.

3Measure, then divide

Prusa's instruction is to make three or more measurements in the middle of each wall and average them.1 Four walls × three points = twelve readings. That is not bureaucracy — extruded walls vary along their length, and one reading is a sample of one.

The formula — Prusa's, and the general form
# Prusa's version, valid when your current EM is 1.00
EM = nominal width / average measured width

# The general form — use this one, it always works
new EM = current EM × (nominal width / average measured width)

# worked: nominal 0.45, measured 0.472, current EM 1.00
new EM = 1.00 × (0.45 / 0.472) = 0.953

Prusa's published formula is the special case where you start from 1.00, which is where a fresh system profile puts you.1 The moment you re-measure a profile that already carries a tuned EM — which is exactly what your second run is — you need the general form, or you will throw the first correction away.

🧮 Twelve readings in, one extrusion multiplier out

Three points per wall, four walls, middle of the wall and away from the seam. Enter them as you measure — the average, the spread and the corrected EM update as you go, and the widget complains in exactly the places the lesson says you should stop.

One row per wall, three readings across it, in mm.

New EM
Average measured
Spread (max − min)
Readings entered
Keep going — the average of four readings is not the average of twelve, and it is the spread that tells you whether to trust any of them.
Outside Prusa's 0.9–1.1 band — don't write this down. Check firmware flow (M221), that the wall really was a single perimeter, and that the nominal width above is the one your profile actually uses. On an MK3S, suspect the legacy M221 S95 line first.
Spread over 0.02 mm. That is usually a mechanical or measurement problem rather than filament variation — look at the part before you trust the average. A wall that varies along its length will not be fixed by a multiplier.
Believable: inside the band, tight spread. Apply it, reprint, and re-measure — mm off nominal now, and you are aiming to land inside ±0.01 mm on the second run.
Already within ±0.01 mm of nominal. This run is signed off — the correction above is a rounding-level tweak, so record the EM you used and stop. Three iterations means something else is wrong.
Nothing here is saved. Paste the twelve raw readings to your tutor, not the average.

Drill it

Work each one out before opening it. Getting this arithmetic wrong is the single commonest way people end up with an EM that makes things worse.

Drill 1 · Nominal 0.45 mm, measured average 0.437 mm, current EM 1.00

1.00 × (0.45 / 0.437) = 1.030

The wall came out thin, so you need more plastic — EM goes up. Sanity check: 1.03 is comfortably inside Prusa's 0.9–1.1 band,1 so this is a believable result. Apply it and reprint.

Drill 2 · Nominal 0.45 mm, measured 0.461 mm, current EM 0.953 (your second run)

0.953 × (0.45 / 0.461) = 0.930

Note what happens if you use Prusa's short form here instead: 0.45 / 0.461 = 0.976, which would undo most of your first correction and send you back toward over-extrusion. Same numbers, opposite direction. Always carry the current EM through.

Also worth noticing: your first correction overshot slightly and you're now correcting back. That is normal and it is why the lesson says one iteration, then stop.

Drill 3 · Nominal 0.45 mm, measured 0.402 mm, current EM 1.00 — on an MK3S

1.00 × (0.45 / 0.402) = 1.119

That is outside 0.9–1.1, so do not write it down. The likeliest explanation on this specific machine is the legacy M221 S95 line in start G-code: firmware is already delivering 95 %, so the wall is thin for a reason that has nothing to do with the filament.

The fix is not to stack an EM of 1.12 on top of it. Confirm what M221 is actually doing, decide deliberately which of the two layers is carrying the correction, record that decision in your log, and re-measure. An EM of 1.12 that silently depends on a firmware line you've forgotten about is a trap you set for yourself six months from now.

4What this invalidates

EM is step two of eight. Re-running it voids advance, retraction, cooling and MVS — all four are judged on how deposited lines look and behave, and you have just changed how much plastic is in every line. It does not void your temperature, which is upstream.

The reverse dependency is the one to internalise: if you change temperature, EM must be re-measured. Viscosity changes how much material a given push delivers, so a 10 °C move is a real change in flow, not just in appearance.

5Check yourself

From memory. Question 2 is the one people get wrong in practice.

Question 1 of 3
Firmware flow sits at 95 % and you set the slicer's EM to 0.95. What is the total?
Question 2 of 3
Second run, current EM 0.95, nominal 0.45, measured 0.458. What is the new EM?
Question 3 of 3
Why does the lesson insist you read extrusion width out of the profile first?
✎ Your task · do this with your tutor

Measure flow on the machine you towered

Same machine, same spool, same temperature you chose in Lesson 10. One variable at a time.

  1. Slice anything for that printer and report what M221 is doing in your start G-code — quote the literal line, or say there isn't one.
  2. Read and report Print Settings → Advanced → Extrusion width → Perimeters.
  3. Print the single-wall box. Take twelve caliper readings — three per wall, middle of the wall, away from the seam — and paste me all twelve, not the average.
  4. Compute the new EM with the general form, showing the arithmetic.
  5. Apply it, reprint, re-measure, and report whether you landed within ±0.01 mm.

→ Twelve raw readings, please, not one average. Their spread tells me things the average hides — a wall that varies by 0.05 mm along its length is a machine problem, and I'd rather catch that now than have you tune advance on top of it.

What you can now say

  1. Total flow is M221 × EM, and the two are independent — correcting in both places corrects twice.
  2. The MK3S ships a legacy M221 conditional in start G-code; verify it in your own sliced file rather than assuming.
  3. Measure three points per wall on four walls and average — twelve readings, cooled, away from the seam.
  4. Use the general formula, new EM = current EM × (nominal / measured); Prusa's short form only works from 1.00.
  5. An answer outside 0.9–1.1 is a fault report, not a filament property.
📖
Primary source · read this next
Extrusion multiplier calibration — Prusa Knowledge Base
The whole procedure in Prusa's own words, both the precise and visual methods, plus the downloadable cubes. Read it alongside this lesson and note where the two differ: Prusa assume you start from an EM of 1.00, and this course doesn't.
💬 I'm your tutor — ask me anything. Paste me your twelve readings and I'll tell you whether the spread is filament, machine, or measurement technique — that distinction is genuinely hard to make alone, and getting it wrong sends you chasing the wrong fault for an evening. Confused beats stuck.
Linked: Glossary → extrusion multiplier, extrusion width, extrusion equation, volumetric flow rate, calibration order, one-variable rule, Pressure Advance, max volumetric speed, system profile, tuning log. Keep it close with → The Calibration Card.

§References

  1. Prusa Research. Extrusion multiplier calibration. The setting's location in Filament Settings; that adjusted values are "usually … between 0.9 and 1.1"; the Precise and Visual methods and their cube models; "make three or more measurements in the middle of each wall and calculate the overall average thickness"; the formula Extrusion multiplier = (Extrusion width (0.45) / Average measured wall thickness); and that "Total flow rate = Flow multiplier in firmware (M221) x Extrusion multiplier in PrusaSlicer" with adjusting one not influencing the other. Link.
  2. Prusa Research. Buddy firmware specific G-code commands. M221 — "Set extrusion percentage", parameters T (tool number) and S<percent>. Link.
  3. Prusa Research. Max volumetric speed. PrusaSlicer's stadium-shaped cross-section model — the reason a measured wall width is a meaningful proxy for delivered volume. Link.
  4. This course. Master tuning guide, §2.2 — the single-wall box setup, the general correction formula, and the MK3S start-G-code M221 conditional. Verify the last of these against your own sliced G-code rather than relying on the guide. Link.

Lesson L11 of the prusa-profile-tuning course · all lessons · The Calibration Card · glossary