Under-Extrusion: Gaps, Thin Walls and Weak Prints, Diagnosed in 4 Questions

The print finished — and it feels like a sponge. Walls are thin and translucent, lines that should touch have gaps, layers look skipped, and the part snaps along a layer line under finger pressure. That is under-extrusion: less plastic laid down than the slicer asked for — a symptom with a family of causes, from a half-blocked nozzle to a slicer quietly asking for too little. In the FixMyPrint community dataset it was the fifth most reported problem: 1,281 of more than 20,000 help posts, and 19.0% of those 1,281 posts had a recorded fix confirmation. Instead of a fix list, this guide asks four triage questions: real prints overlap (one print can click and grind and fade), so the questions set an order of suspicion — cheapest discriminating tests first, and a test narrowing the cause before anything changes.

Compiled from published sources — Prusa's Knowledge Base, Simplify3D's troubleshooting guide, the FixMyPrint dataset — plus clearly labelled community advice. Not personally tested on a printer in our workshop. Manufacturer figures are labelled; where no defensible universal number exists, the guide gives a test instead of a fake-precise setting.

Applies to: any FDM printer · direct-drive and Bowden machines (tests flag where they differ) · PrusaSlicer, Bambu Studio, OrcaSlicer or Cura. Materials scope: the baseline assumes standard PLA, PETG and similar filaments; abrasive or filled filaments (carbon-fibre, wood-fill, glow-in-the-dark) wear brass nozzles fast and may need a hardened or larger nozzle — outside this baseline; nozzle wear is its own suspect there.

In this guide: Step 0: Read the print · The 4-question triage · First: restore the baseline · Branch A: Partial nozzle clog · Branch B: The extruder cannot push · Branch C: The filament itself · Branch D: Too cold, or too fast for the melt · Branch E: The slicer is asking for too little · Branch F: Starts fine, fades halfway · What NOT to do · FAQ

Key Takeaways

• Under-extrusion is a symptom, not a cause. Four triage questions say where to look first.
• Test before you fix. Each branch opens with a test; tests give evidence, not proof — each fix's recovery check is the confirmation.
• A partial clog is the great impersonator. It hides inside complaints of every shape — why the cold pull earns its place in Branch A.
• Never mask a mechanical fault with flow. It postpones the diagnosis and over-extrudes everything else.
• Timing is evidence. Bad from the first layer and bad only after twenty good minutes point at different suspects.

Step 0: Read the Print (60 Seconds)

Look at what the missing plastic is doing — and at what it is not doing:

What you see	What it most likely is	Where to start
Gaps between lines on walls and top faces; parts that snap along layer lines	Under-extrusion — less plastic delivered than requested; material missing in the layers, objects fragile (Prusa's Knowledge Base describes the same picture)	The 4 questions below
Gaps only on the first layer; upper layers solid	Nozzle height / Z-offset — the bead is laid too high to squish together, not under-fed	Our first-layer guide
Extra material — bumps, bulges, fat walls	Over-extrusion or pressure, the opposite fault	Our blobs and zits guide
Domes and sagging patches only on top faces, following the infill grid	Pillowing — a top-skin problem, not a feed problem	Our pillowing guide
Pitted, rough surfaces with popping or hissing at the nozzle	Wet filament is one suspect — moisture flashing to steam disrupts flow (Branch C weighs the full evidence, not the sound alone)	Branch C, then our PLA drying guide

Note when it began — new spool, material change, or out of nowhere? A change you made is a suspect; no change points at wear, dirt and partial blockages.

The 4-Question Triage

A triage sequence, not a decision tree: symptoms overlap, and one fault causes the next (a clog makes the extruder click; a slipping gear makes a print fade). The answers are priority rules — cheapest discriminating tests first (strand test and feed test take minutes and split the machine in half, feed side versus melt side) — and test results overrule the question order whenever they disagree. Throughout: same test print, same filament, same temperatures, one change at a time, each test before its fix.

Question 1: Does it fail from the first layers, or start fine and fade? Bad from the start means it struggles even at low demand. Starts clean, then fades at a similar point in print after print — Branch F first; that timing pattern is its own signature.

Question 2: Is the extruder clicking, ticking or skipping? A clicking extruder is a motor failing against resistance. If yes — Branch B first: clicking says the extruder is losing, not yet why the path resists (Branches A and D hold the usual reasons).

Question 3: Is the filament feeding cleanly — or grinding, dusty, stalling? Watch a minute of printing. Dust at the drive gear or a ground-flat patch on the strand points to Branch B; the spool jerking or snagging instead of unrolling smoothly points to Branch C.

Question 4: Is it one filament, or every filament? Swap in a spool you trust, same settings, same test print. Clean print: the fault travels with the filament — Branch C. Every filament the same way: machine or profile — Branch A (clog), Branch D (melt capacity) or Branch E (flow).

First: Restore the Baseline (Before Any Flow Change)

Before touching flow or any tuning number, restore a known-good manufacturer profile — stock printer profile plus matching stock filament profile — and verify three settings in it: nozzle diameter matches the nozzle fitted; maximum volumetric speed / flow limit is the maker's default, not a stray edit; filament diameter is the standard value. Reprint the test piece. Clean on stock? The fault was in your tuned profile — diff the two. Still under-extruding? The branches below start from honest ground.

Why It Happens (30 Seconds of Physics)

The slicer commands a volume of plastic per move; under-extrusion is any gap between commanded and delivered. It opens in only three places: the feed side (gear slips or grinds; tangled or undersized filament), the melt side (blockage, cold slow melt, or filament softening above the melt zone — heat creep), or the command side (diameter, multiplier or firmware feed calibration understate reality, so the printer delivers the too-little it was told).

Branch A: Partial Nozzle Clog — the Plastic Cannot Get Through

Signs: under-extrusion on every filament; often begins after a material change (especially dropping from hot to cooler without a full purge — Prusa advises preheating to the previous filament's temperature when loading a cooler one). The purge strand is thin, intermittent, or curls.

The first test — the strand test: heat the nozzle to the loaded filament's printing temperature, raise the head clear of the bed, extrude into the air. A healthy nozzle drops a straight, even strand. A strand that curls sideways and climbs back onto the nozzle, comes out ribbon-thin, or pulses is a clue, not proof: persistent sideways curl is the classic partial-clog signature in Prusa's troubleshooting documentation (the stream deflected by something in its path), but curl can also come from nozzle geometry or residue on the nozzle's face. Treat it as a reason to run the cold pull below and read its tip — the stronger evidence. A clean, straight strand makes a clog unlikely: Branch B or E.

Printer-specific safety gate — read before any cold pull: check your model, hotend and nozzle type first. There is no universal pull temperature: Prusa's own procedures differ by model — roughly 85°C (MK3 series), 90–95°C (MINI), around 100°C (a newer manual), some machines running an automated cycle near 80°C — and Prusa advises against cold pulls on its Nozzle X. If your maker publishes a procedure for your machine (several do, some automated), follow theirs; it outranks everything below. For machines without one:

1. Load light-coloured PLA or cleaning filament — debris shows on a pale pull. Heat to its normal printing temperature and hand-feed until clean plastic flows steadily.
2. Keep gentle forward pressure while the nozzle cools to your maker's pull temperature (maker figures for PLA span roughly 80–100°C by model and method; ≈90°C, where that is your only reference, is a starting point, not a standard).
3. At the pull temperature, grip the filament with pliers and pull — firmly, steadily, in one motion. The plug releases suddenly; keep your free hand and face clear.
4. Read the tip. A good pull ends in a cone mirroring the nozzle's inside. Dark specks, a gritty crust, or a strand of a previous colour in the cone is the blockage, in your hand — evidence, not a guess. Repeat from step 1 until the tip comes out clean.

Safety: during heating and cooling the nozzle sits near printing temperature and will burn skin on contact — never touch the nozzle or heater block; use pliers, not fingertips. Never force a pull against a stone-cold nozzle: if the filament will not move, reheat a few degrees. Do not disassemble the hotend hot. If pulls stop improving the strand, the remaining step is a nozzle swap: let the assembly cool fully before handling it, and follow the manufacturer's removal and installation procedure — including any hot-tightening step it specifies — rather than assuming an entirely cold job.

How you know it worked: re-run the strand test — straight and even — and the reprint's gaps close, with no setting changed. If the tip comes out clean twice yet the strand still misbehaves, the cause may sit above the nozzle (heatbreak or tube) or in the nozzle's geometry: Branch F territory and, beyond its tube reseat, your maker's support.

Branch B: The Extruder Cannot Push — Clicking, Grinding, Slipping

Signs: clicking or ticking from the extruder; plastic dust around the drive gear; a flattened, chewed patch on the filament when unloaded.

The first test — the 100 mm feed test (feed movement, not delivered volume): with the nozzle hot and clear of the bed, mark the filament 120 mm above where it enters the extruder. Command 100 mm of extrusion slowly and steadily, into the air — far slower than any printing move. About 20 mm remaining above the mark means the extruder moved the commanded length at that slow feed: feed-side evidence in its favour. Meaningfully more left means it under-moved — the feed side is guilty at any speed. What the test does not measure is the volume landing on the part at print speed: it tracks filament movement past the gear under almost no load, so a partial clog can pass a slow feed test and still starve a fast print — that flow-versus-feed distinction has its own check in Branch D. Clicking during the test is also a result: the path resists even at low demand — clear it via Branch A, then repeat.

Fixes, in order:

1. Clear the resistance first. A slipping gear is usually pushing against something — run Branch A's strand test and cold pull if the strand misbehaves; tension changes against a clog only chew the filament harder. Check: full 100 mm moved, no click or skip.
2. Clean the drive gear. Power down, open the extruder as your maker allows, brush the packed dust out of the teeth — packed teeth grip like a smooth wheel. Check: teeth visibly clean; a test print finishes with no new grinding dust at the gear.
3. Set the idler tension by your manufacturer's guidance for your model — not by a universal "middle". Not every extruder has a meaningful middle setting; some have little real adjustment at all (fixed or spring-set). Where yours adjusts, follow the maker's marks or manual. Too loose slips; too tight flattens the strand oval and adds drag. Check: even tooth marks on the unloaded strand — present, not cut deep or flattened — and the full commanded length moved.
4. Inspect the strand you unloaded. One deep ground-flat hollow means the gear spun against a stalled strand: the stall came first (clog, tangle, cold melt); the hollow is its scar. Fix the stall, then cut away the damaged section before reloading. Check: no new hollow on the strand unloaded after the same print.

If the gear turns visibly without moving filament at any tension, or the drive gear is loose on its motor shaft, you have left adjustment territory — a fasteners inspection by your maker's manual, or their support.

Branch C: The Filament Itself — Tangles, Diameter, Moisture

Signs: the fault arrived with this spool and leaves with it (Question 4's swap test); the spool jerks instead of unrolling; symptoms swing during a print as different parts of the spool feed.

The confirming tests — three quick ones:

1. The unroll watch: print or hand-feed while watching the spool for a full minute. A crossing loop — the loose end tucked under another coil — snags, holds, then releases with a jerk; extrusion starves exactly while it holds. A visible snag is the cause, confirmed.
2. The caliper check: measure the diameter at three or four points along a metre, rotating the caliper between readings. Prusa's Knowledge Base cites ±0.05 mm as the tolerance for the filament it discusses — a cited example, not a universal pass/fail line: compare your readings against the tolerance your own spool's manufacturer states, if any. Readings consistently well under the slicer's set diameter (usually 1.75 mm) mean the slicer computes volume for thicker filament than you own — silent, uniform under-extrusion.
3. The moisture check — more than the sound: popping or hissing at the nozzle, with pitted surfaces, is one sign of moisture flashing to steam. Do not convict on sound alone — weigh storage history (weeks open in a humid room?), material (nylon and TPU absorb fast; PLA slowly), visible surface defects, and the most convincing test: a comparison print with a known-dry spool. Two or more signs together convict the spool.

Fixes, in order:

1. Untangle properly; never print through a snag. Unwind past the crossing, keep tension on the loose end throughout (a crossed loop forms the moment the end goes slack), rewind neatly, and clip the end in whenever the spool leaves the printer. Check: smooth unrolling; starved patches gone from the reprint.
2. Tell the slicer the diameter you measured. If the caliper reads consistently off (within reason), enter the measured average as the filament diameter — Simplify3D's guide puts this check first: every volume calculation scales with it. Check: walls at full thickness on the same test print, no other change.
3. Dry on evidence, not by default. If — and only if — the moisture evidence above convicted the spool, dry it at its maker's published numbers; our PLA drying guide compares every manufacturer's figures. Check: purge smooth and silent; surface defects gone.

If two spools from the same maker both measure far off their labelled diameter, that is a purchasing problem, not a printer problem.

Branch D: Too Cold, or Printing Too Fast for the Melt

Signs: matte, rough walls; layers that peel apart easily; keeps up at slow speeds and starves in fast infill; clicking only during the fastest moves.

The first test — the temperature step: on the same small test print, raise the nozzle temperature by 5°C, staying inside the range printed on your spool's label (Prusa's guidance works in ±5–15°C adjustments). If gaps close and bonding improves, temperature (or melt rate) was at least part of the constraint. If nothing changes after two steps up, restore the original temperature and stop climbing — but that does not clear the melt side: a volumetric-flow limit survives a temperature test, because the hotend may be unable to melt fast enough at print speed whatever the setpoint. Next check:

The flow-vs-feed check: print the same small test piece twice — once at your normal speeds, once at half speed (or with the slicer's volumetric flow limit halved). If the slow print comes out full and the fast print starves in the same places, the machine feeds correctly but cannot melt fast enough: a flow-capacity problem, not a feed problem — the mirror image of Branch B's feed test; the two are a pair. Check: at reduced demand, fast infill and slow walls come out at the same fullness on the same print.

Fixes, in order:

1. For a confirmed flow limit: print at the warm end of the spool's range (brands and colours melt differently; the spool label sets the ceiling), slow the fast features, or lower the maximum volumetric flow. Beyond that, melt capacity is hardware — a high-flow hotend is a purchase, not a setting. Check: the test's gaps close and a finished wall resists a layer-peel instead of unzipping.
2. Check the mechanical thief: a first layer set too close. An over-squished first layer chokes the nozzle outlet and clicks the extruder for the first minutes, then recovers. Clicking that lives only at the start is a first-layer height question — our first-layer guide covers the height checks. Check: the first layer lays as connected, even beads instead of a starved sheet.

Branch E: The Slicer Is Asking for Too Little — Flow and Multiplier

Signs: uniform, even thinness everywhere; the feed test moves the full commanded length; the strand runs reasonably straight; the printer sounds happy. That points at what the machine was told — but the signs are not unique to slicer settings: a wrong firmware E-steps / rotation-distance value, or a feed fault that appears only under load, wears the same face. Keep the calibrations separate: firmware feed calibration (E-steps / rotation distance — filament moved per command; Branch B's test) is a machine setting; slicer flow (extrusion multiplier) is a per-profile volume trim. Never patch one with the other.

The confirming test — the single wall (the standard flow-calibration check): print a small open box or vase-mode cube with one perimeter and measure the wall with calipers; compare it with the line width your slicer says it laid. Treat the result as evidence, not proof: width also depends on line shape, caliper placement, and the slicer's width strategy. A wall consistently and clearly thinner than stated supports under-delivery in volume — enough to calibrate flow by. Full width with gaps between walls sends you back to Branches A–D.

The fix — calibrate the extrusion multiplier in small steps:

1. Rule out the firmware side first. If the Branch B feed test under-moved, fix feed calibration (E-steps / rotation distance) by your maker's procedure — a multiplier tuned over a feed error follows you into every profile.
2. Verify the diameter setting (Branch C's caliper check). A multiplier tuned to hide a wrong diameter changes with every spool.
3. Adjust the multiplier in one small step. Simplify3D's scale: 1.00 → 1.05 extrudes 5% more plastic. Steps of that size are the calibration unit; reprint the wall after each.
4. Stop at the measured width. Check: wall matches stated width across readings; walls touch; tops close; no over-extrusion in exchange.

The warning that defines this branch: the multiplier calibrates a healthy machine; it never overpowers a sick one (see What NOT to Do).

Branch F: Starts Fine, Fades Halfway — Heat Creep and Bowden Gaps

Signs: the Question 1 pattern — good printing for a predictable stretch, then thinning, clicking, sometimes a full stall; a cool-down and restart prints well again for about the same stretch; the unloaded filament's tip looks swollen or mushroomed. The lead suspect is the hotend's thermal boundary: heat climbs above the melt zone (heat creep) and filament softens, swells and drags — or, on Bowden machines whose tube seats against the nozzle, a tube-end gap fills with melt that half-sets into a plug.

The confirming tests:

1. The restart test: cool the printer fully and restart the same print untouched. If it fades again at roughly the same duration or layer height, the pattern is consistent with a time- or heat-dependent fault — not proof of heat creep. Rivals wear the same timing: spool drag building as the roll turns, retraction zones that occur only at certain layers, a heating stepper motor, an intermittent feed fault. The fan and tube checks below separate hotend cooling from those rivals: a failed fan or heat-damaged tube end incriminates the hotend; clean cooling plus a fade tracking one feature or layer band points back at retraction and feed.
2. The fan check: with the hotend at printing temperature, confirm the hotend (heatsink) fan actually spins. Starting only above a threshold temperature is normal; a fan that never starts, rattles, or is choked with dust is not — a silent heatsink fan is heat creep's most common accomplice.
3. The tube check (Bowden, done cold): remove the tube by your maker's method and inspect its hot end. A burnt, shrunk or melted-back end — or a tube never pushed fully home — leaves the gap where plugs form.

Fixes, in order:

1. Restore the cooling. Clean the fan and heatsink fins; replace a fan that does not run at temperature. Print PLA with the enclosure door open where your maker advises it. Check: the print passes its previous failure point at full thickness.
2. Reseat the Bowden tube — cold, square, fully home. This applies only to hotend designs where the PTFE tube runs down and contacts the nozzle; all-metal and other designs seat differently — follow your maker's design. Trim a heat-damaged end square (or replace the tube) and refit it pushed firmly against the nozzle seat, as your maker's hotend guide describes. Check: next unload's tip is a clean taper, and the print passes its previous fade point.
3. Reduce the heat load from retraction. Heavy retraction over one small area pumps heat upward; if the fade tracks dense retraction zones, reducing retraction frequency is labelled community practice worth one test print. Check: the dense zone prints through without thinning — and stringing does not return; if it does, restore the setting.

If the fan runs, the tube seats perfectly, and the fade still arrives on schedule, the fault sits beyond cleaning-and-settings territory: stop, and take the restart-pattern evidence to your maker's support.

About the Community Numbers in This Guide

The FixMyPrint dataset behind this guide is one analyst's scrape of 20,000+ help posts and roughly 60,000 fix suggestions from one of Reddit's largest 3D-printing help communities, collected over one year and classified by problem type, printer model, and whether the asker confirmed a fix. Under-extrusion accounts for 1,281 of those posts (#5 overall); 19.0% of the 1,281 had a recorded fix confirmation — a confirmation count, not a success rate: many askers fix a print and never return to report it. Clogging is filed separately at 764 posts, with clogs noted to also hide inside under-extrusion and stringing complaints. In the companion stringing dataset, cold pulls were confirmed unusually often in reported-back cases, while suggested far less often than weaker-confirming fixes — why Branch A teaches the procedure properly. (Our notes record that rate as a bare percentage without numerator or denominator, so this guide describes the finding rather than quoting a number it cannot stand behind.)

What NOT to Do

• Do not crank the extrusion multiplier to 120–130% to "push through" the problem. Against a clog or slipping gear the extra plastic has nowhere to go — pressure builds, the gear grinds deeper, a partial clog can pack into a complete one. Flow calibrates a healthy machine (Branch E); it cannot heal one.
• Do not force a cold pull. The pull works inside its temperature window, with pliers, in one firm motion — never by escalating force against a stone-cold or fully molten plug.
• Do not improvise temperatures above the spool's range to melt through a blockage. Overheated plastic degrades and carbonises into a harder clog, and the heat feeds heat creep (Branch F). One exception: some manufacturers publish directed higher-temperature clearing procedures for specific clogged hotends — maintenance done exactly by the maker's instructions for that machine, not a licence to blast any hotend.
• Do not change flow, temperature and tension in the same print. If it improves, you own three unverified numbers; if it worsens, you cannot retreat. One change, one test, one verdict.
• Do not keep printing through persistent clicking. Every click chews the strand flatter and guarantees worse feeding. Pause, run the questions, fix the resistance.
• Do not disassemble the hotend hot, and do not touch the nozzle.

When to Stop and Ask for Help

If a trusted spool (Question 4), a clean strand test, a feed test that moved the full commanded length and all six branches leave the same gaps, stop tuning and gather the evidence you now own: your test results and numbers, what the pull tips showed, and the restart pattern if Branch F applied. A support ticket carrying measurements gets a different conversation than "it under-extrudes". Mechanical suspects beyond this guide's reach — a drive gear loose on its shaft, a worn nozzle bore, gearbox alignment where fitted (Prusa names it for its Nextruder machines) — belong to the maker's manual and support.

FAQ

Q: Only my top layers have gaps. Is that under-extrusion?
A: Sometimes — mild under-delivery shows first where lines are longest and fastest. Domed, puffy tops are pillowing (Step 0 table); a flat top with gaps beside full walls deserves Branch E's single-wall check first.

Q: The first layer is gappy but everything above it is fine. Same fault?
A: Almost never: under-extrusion starves the whole print, while a starved-looking first layer under healthy layers is a bead laid too high to merge — a nozzle-height problem (see our first-layer guide).

Q: Will raising the flow just fix it? It is one number.
A: It fixes one cause only — a miscalibrated multiplier on a healthy machine — and Branch E's wall check is the evidence for that cause. Against a clog, tangle or slipping gear it is a mask, and masks come off mid-print.

Q: How often should I do a cold pull?
A: On evidence, not on a calendar: when the strand test curls or thins (safety gate first), and when stepping down from a hot material to a cooler one, so residue does not become tomorrow's partial clog.

Q: My printer is nearly new. Can it really have a hardware cause already?
A: Yes — early-life causes are usually assembly and configuration, not wear: a nozzle seated incorrectly from the factory or after a swap, a crossed spool from first loading, an idler at its shipping position, or a factory profile mismatched to the bundled filament. Start with the baseline step and Branch C; if a new machine fails all six branches, that is what warranty support exists for.

Summary: The Six Branches at a Glance

Diagnose before you tune: read the print (Step 0), run the four triage questions, restore the stock-profile baseline — then work the branch the evidence points to:

Branch	Lead signal	Confirming test	First fix
A — Partial clog	Strand curls or thins; every filament	Strand test; read the cold-pull tip	Cold pull by your maker's method (safety gate first)
B — Extruder feed	Clicking, grinding, dust at the gear	100 mm feed test, run slowly (movement, not volume)	Clear resistance; clean gear; maker-set tension
C — Filament	Fault travels with the spool	Unroll watch; caliper check; moisture evidence	Untangle; enter the measured diameter; dry only on evidence
D — Melt capacity	Starves only at speed; clicking on fast moves	5°C temperature step; flow-vs-feed slow/fast comparison	Warm end of range; lower volumetric demand
E — Slicer flow	Uniform thinness; feed test moves full length	Single-wall measurement (evidence, not proof)	Multiplier in ~5% steps — after feed and diameter checks
F — Fades halfway	Good start, fades on a schedule; restart repeats it	Restart pattern; fan check; tube check	Restore cooling; reseat tube (contact-seat designs only)

Every fix ends with its verification — if the named sign did not move, that was not your cause.

Honesty note: Compiled from manufacturer and community documentation — not personally tested on a printer in our workshop, and it does not pretend otherwise. Figures are labelled as such; dataset figures are stated with their denominators where the research records them — where a denominator is not recorded (the cold-pull rate), the finding is described qualitatively instead of quoting a percentage the guide cannot stand behind. Where the evidence ran out — a universal tension setting, a universal pull temperature — the guide says so instead of inventing a number.

Sources: FixMyPrint — dataset analysis (1,281 posts, #5; 19.0% of posts had a recorded fix confirmation); FixMyPrint — stringing dataset (cold pull confirmed unusually often in reported-back cases; denominator not recorded in our notes, so no percentage is quoted); Prusa Knowledge Base — Under-extrusion (Nextruder printers); Prusa Knowledge Base — Cold pull; Simplify3D — Print Quality Troubleshooting Guide (Not Extruding Enough Plastic; Grinding Filament); community advice: Prusa forum — cold pull explainer thread.

Related guides: Stringing after changing PLA filament? Check these 5 things first · How to dry PLA filament: every manufacturer's numbers, compared · First Layer Won't Stick on Your Ender 3? 8 Things to Check Before Touching Settings · Why Is My Top Surface Rough and Puffy? Pillowing, Diagnosed in 3 Questions · 3D Print Blobs and Zits: Seam, Moisture, or Pressure?