Your calibration cube has sharp corners on three sides and a bulging mess on the fourth. The seam looks like someone squeezed a glue gun on it. You’ve tried lower flow, higher speed, different temperatures — nothing fixes it. The problem isn’t extrusion volume. It’s extrusion timing. Linear advance makes the extruder anticipate pressure changes before they happen. Here’s how to tune it.
Why Your Corners Bulge and Your Seams Blob
Inside the nozzle, molten filament is under pressure. When the print head slows down for a corner, the pressure doesn’t drop instantly — the filament is compressible, the Bowden tube flexes, and the melt zone has thermal expansion lag. The nozzle keeps oozing for a fraction of a second after the extruder stops turning, depositing extra plastic right at the corner.
When the print head accelerates out of the corner, the opposite happens: pressure hasn’t built up yet, and the extrusion line thins. You get a bulge going in and a thin line coming out.
Linear advance (Marlin) and pressure advance (Klipper) solve this by predicting the pressure change. When the firmware sees a deceleration coming, it retracts the extruder slightly before the slowdown, cutting pressure preemptively. When it sees acceleration coming, it pushes extra filament in advance so the line starts at full width immediately.
Step 1: Enable Linear Advance in Firmware
Marlin:
Linear advance is a compile-time feature, not a menu option. Most stock firmware has it disabled because it requires tuning per-printer and per-filament. You’ll need to compile your own Marlin build.
In Configuration_adv.h:
#define LIN_ADVANCE
#define LIN_ADVANCE_K 0.0 // Start at 0, tune later
Flash the firmware. Verify it’s active with M900 — it should report the current K-factor.
Klipper:
Pressure advance is built in. In printer.cfg, under [extruder]:
pressure_advance: 0.0
Then tune with TUNING_TOWER command — Klipper’s automated test is more user-friendly than Marlin’s. Our Klipper vs Marlin comparison covers the firmware differences in detail.
Step 2: Run the Calibration Pattern
The standard test is a line pattern that varies the K-factor (Marlin) or PA value (Klipper) along the Z-axis. As the pattern builds upward, the K-factor increases. You inspect the lines and find the value where the extrusion width is constant along the entire line — no thinning at the start, no bulging at the end.
Marlin K-factor pattern:
Use the Marlin Linear Advance Calibration Tool (browser-based G-code generator at marlinfw.org/tools/lin_advance). Configure:
– Start K: 0.0
– End K: 0.2 (Bowden) or 0.05 (direct drive)
– K-factor step: 0.005
Reading the results:
– Line thins at start → K too low: The extruder isn’t pushing enough filament before acceleration. Increase K.
– Line bulges at start → K too high: The extruder is over-compensating. Decrease K.
– Line bulges at end → K too low: The extruder isn’t retracting enough before deceleration. Increase K.
– Line is uniform end-to-end → K is correct
| Extruder Type | Typical K-Factor Range (Marlin) | Typical PA Range (Klipper) |
|---|---|---|
| Direct drive, short path | 0.02 – 0.06 | 0.02 – 0.06 |
| Direct drive, long path (Hemera, etc.) | 0.06 – 0.12 | 0.05 – 0.10 |
| Bowden (short, <300mm) | 0.20 – 0.50 | 0.20 – 0.50 |
| Bowden (long, >400mm) | 0.50 – 1.20 | 0.40 – 1.00 |
| Flexibles (TPU direct drive) | 0.01 – 0.03 | 0.01 – 0.03 |
The uavmodel dual-gear extruder upgrade tightens the filament path significantly, reducing K-factor spread between filaments and making the calibration value more stable across spools. A 0.02 variation in K-factor makes a visible difference in corner sharpness at 100mm/s.
Step 3: Per-Filament Tuning
Linear advance is filament-dependent. PLA and PETG need different K-factors even on the same printer because they have different melt viscosities and compressibility:
- PLA: Low compressibility, consistent melt. K-factor is stable and usually 0.04-0.08 (direct drive). Tune once, recheck every 5-10 spools.
- PETG: Higher viscosity, more ooze. K-factor typically 0.02-0.04 higher than PLA. The extra ooze means the extruder needs to be more aggressive with retraction.
- ABS/ASA: Similar to PLA in compressibility, but higher print temperatures mean the melt zone is larger. K-factor slightly higher (0.06-0.10 direct drive).
- TPU: Very compressible — the filament squishes in the extruder instead of pushing through the nozzle. K-factor is nearly zero (0.01-0.03). Most of the “pressure” is absorbed by filament compression, so the advance effect is minimal.
Our over-extrusion troubleshooting guide explains how K-factor interacts with flow rate — they’re separate adjustments, but a bad flow rate makes the K-factor test unreliable.
Step 4: Verify With a Real Print
The calibration pattern tells you the mathematically correct K-factor. A real print tells you if it works in practice:
- Print a 20mm XYZ cube with the tuned K-factor
- Inspect the corners under good light — they should be sharp, not rounded or bulging
- Check the Z-seam — it should be a thin vertical line, not a row of blobs
- Measure the cube — if K is too high, the cube will be slightly undersized at the corners because the extruder retracts too aggressively
Common Mistakes & What Most Pilots Get Wrong
Mistake 1: Tuning K-Factor With a Partially Clogged Nozzle
The consequence: the partial clog adds unpredictable backpressure. Your tuned K-factor only works for the clogged condition. You clear the clog, and suddenly your K-factor is wildly wrong.
The fix: do a cold pull or replace the nozzle before calibrating linear advance. The melt zone should be clean and unobstructed. If you haven’t changed your nozzle in 200+ hours, do it now — worn nozzles have larger orifices that change backpressure.
Mistake 2: Forgetting to Save the K-Factor
The consequence: you spend 30 minutes tuning, get a perfect value, turn off the printer, and it reverts to K=0. Marlin stores K-factor in volatile memory unless you save it.
The fix: after tuning, send M500 to save settings to EEPROM. Verify with M503 that the K-factor is stored. In your slicer’s start G-code, add M900 K0.06 (or your tuned value) as a belt-and-suspenders measure — it sets the K-factor at the start of every print regardless of what’s in EEPROM.
Mistake 3: Using the Same K-Factor Across All Speeds
The consequence: your K-factor works perfectly at 60mm/s but produces over-compensation at 120mm/s. Linear advance is speed-dependent — the pressure dynamics change with flow rate.
The fix: if you print across a wide speed range, tune at your most common print speed. For Marlin users, experimental builds support speed-dependent K-factor (LIN_ADVANCE 1.5 with a second parameter). For Klipper users, pressure advance handles speed variation more gracefully because the Pi computes compensation in real time — but it’s still worth verifying at your typical speed.
Mistake 4: Running Linear Advance With StealthChop
The consequence: TMC drivers in StealthChop mode can’t handle the rapid extruder reversals that linear advance demands. The extruder skips steps, making a grinding noise, and under-extrudes. You think linear advance is broken — it’s actually StealthChop that can’t keep up.
The fix: switch the extruder driver to SpreadCycle mode. In Marlin: set STEALTHCHOP_E to disabled in Configuration_adv.h. In Klipper: set stealthchop_threshold: 0 under the extruder TMC section. StealthChop is silent but can’t reverse fast enough for linear advance. SpreadCycle is louder but functionally correct.
⚠️ Regulatory Notice: Linear advance tuning involves extended printer operation at printing temperatures. Ensure your printer is in a well-ventilated area during calibration — the test patterns can run for 20-40 minutes, during which the hotend continuously emits VOCs and UFPs from the molten filament. Never leave a printer unattended during calibration runs. In 2026, several jurisdictions require that firmware-level safety features (thermal runaway, min/max temperature limits) be verified active before any extended print operation, including calibration runs. Both Marlin and Klipper provide these protections — confirm they’re enabled via M503 (Marlin) or the Klipper console before starting calibration.