Dry the filament first, then drop the nozzle temperature 5 to 10 °C, and only then touch retraction. Most people do it the other way round, crank retraction to 5 mm on a direct-drive extruder, and trade strings for clogs.
That order is not mine. Bambu Lab’s stringing and oozing guide lists six causes and puts damp filament at number one, ahead of retraction. Prusa’s stringing article names very high printing temperature and incorrect retraction as the two main culprits and tells you to try a different spool because moisture “will cause a lot of stringing”. The OrcaSlicer retraction test page ends with the same advice from the other direction: if the top of your retraction tower is still stringy, stop tuning, dry the filament and check that the nozzle is not leaking.
1. Wet filament
Water in the filament flashes to steam in the hotend. Bambu describes the result as molten material that bursts and flows abnormally, which no retraction setting can pull back. The tell is stringing that arrived on its own, on a profile and a spool that used to print clean.
The fix is a drying cycle at the data sheet temperature. From the maker data sheets in our drying chart: PLA at 50 to 55 °C for 6 to 8 hours, PETG at 60 to 65 °C for 8 hours, TPU 95A at 70 °C for 7 to 8 hours. PETG, TPU and nylon are the usual suspects. The filament drying guide covers methods and why you should not round PLA’s temperature up.
Do this before any tuning. Calibrating retraction on a wet spool gives you settings that are wrong for the same spool once it is dry.
2. Nozzle temperature too high
Hotter plastic is runnier plastic, and runny plastic drips during travel moves. Prusa’s instruction is specific: lower the nozzle temperature by 5 to 10 °C and check whether stringing drops. Bambu says the same without a number.
Stay inside the filament’s printed range while you do it. The data sheet ranges we hold are 190 to 230 °C for PLA and 230 to 260 °C for PETG, with each brand’s own window on the temperature table. If you are printing PETG at 250 °C, try 240 °C. Do not go below the bottom of the range to chase the last wisp: the OrcaSlicer temperature calibration page notes that lower temperatures reduce stringing but higher ones improve layer adhesion, and that too low a temperature brings under-extrusion and weak layers. It recommends printing a temperature tower and, where several blocks look good, taking the middle of that band, or the upper end if you print fast.
3. Retraction length and speed
Now retraction. The published numbers are smaller than folklore.
| Source | Extruder | Retraction length |
|---|---|---|
| Prusa, MK2.5/S and MK3/S/+ | direct drive | maximum 2 mm |
| Prusa, MINI/MINI+ preset | Bowden | 3.2 mm |
| Bambu Lab | direct drive | generally no more than 2 mm |
| OrcaSlicer test defaults | direct drive | sweep 0 to 2 mm in 0.1 mm steps |
| OrcaSlicer test suggestion | Bowden | sweep 1 to 6 mm in 0.2 mm steps |
| OrcaSlicer, PLA or ABS with a clean tower | direct drive | 0.2 to 0.4 mm |
Bambu’s reason for its cap is that too long a retraction “can easily cause clogging”. OrcaSlicer’s retraction page says the same: too short and it does not stop the ooze, too long and you get clogs or under-extrusion.
Retraction speed matters as well. Prusa says a higher speed improves stringing, up to the point where the extruder motor skips steps. The OrcaSlicer retraction settings page adds that PETG and TPU are more prone to stringing and may need longer retraction than PLA or ABS, which you can set per filament instead of changing the printer profile. Prusa notes flexibles need longer retractions because the filament stretches as it is pulled.
Reading a retraction tower
In OrcaSlicer, run Calibration, Retraction test, with the direct-drive defaults: start 0 mm, end 2 mm, step 0.1 mm. The tower is a column of notches, each printed at a different retraction length. Find the lowest section that is free of strings and ooze, and measure its height. Orca’s instruction is to look up that height in the G-code preview and read the Calib_Retraction_tower comment there, not the ordinary ; retract lines, which can understate the real length when part of the retraction happens before a wipe.
Suppose strings are heavy at the base, thin out, and are gone from the band where the comment reads 0.6 mm. Your setting is 0.6 mm. Orca’s rule is that the best value is the shortest one that gives a clean tower, so resist rounding up to 1 mm for luck. If the whole tower is clean from the first notch, Orca suggests 0.2 to 0.4 mm. If it never cleans up by 2 mm on a direct-drive machine, go back to steps 1 and 2.
4. Travel: where the nozzle goes while it is not printing
Every string is drawn during a travel move, so fewer and shorter travels over open air mean fewer strings. The settings that do this, by their names in each slicer:
- Avoid crossing perimeters (PrusaSlicer) or Avoid crossing wall (Bambu Studio, OrcaSlicer) routes travels inside the part. Prusa says it lowers stringing, especially on the Bowden MINI. Bambu recommends it for models with many separate islands.
- Wipe while retracting drags the nozzle back along the line it just printed as it retracts. Prusa recommends leaving it on, along with retract on layer change.
- Z-hop works against you. Prusa’s wording is that a lower lift Z improves stringing, with the caveat that turning it off risks the nozzle hitting the print. The Klipper documentation goes further and recommends disabling z-lift on retract.
- Minimum travel after retraction is 1 mm in PrusaSlicer’s presets. Raising it saves time and increases oozing, so leave it low while you are fighting strings.
Plate layout counts too. Bambu’s advice for multi-part plates is to reduce the spacing between models, because long hops give the nozzle time to drip. Prusa offers the opposite trick for the same problem: sequential printing, one object at a time, so there are no hops between parts at all.
5. Pressure left in the nozzle
Some ooze is not a retraction failure. It is pressure still stored in the melt when the move ends. Firmware pressure compensation addresses that directly. The Klipper pressure advance documentation says a good pressure advance value reduces ooze throughout the print, and that once it is tuned a small retraction such as 0.75 mm with wipe enabled is still useful against ooze caused by the plastic’s stickiness. The Marlin Linear Advance documentation says to recheck retraction distance after calibrating, because it “may even be as low as 0”. The pressure advance guide walks through calibration.
If you tune pressure advance after retraction, run the retraction tower again. The right length will probably be shorter.
6. Hardware
Prusa’s article lists three hardware causes. A film of old filament on the outside of the nozzle, PETG especially, grabs strands and drags them across the print: clean the nozzle before you start. Poor heat transfer between heatbreak and heatsink can cause oozing, fixed by re-applying thermal paste to the threads. And a hotend that was recently taken apart may simply be assembled wrong.
Bambu adds nozzle size. A worn nozzle has a bigger hole than its label and leaks more, and a slicer profile set for 0.4 mm while a 0.6 mm nozzle is fitted will string for the same reason. If you have changed nozzles lately, check the profile matches; the 0.6 vs 0.4 mm nozzle guide explains what else changes with it.
When a few strings are fine
PETG and TPU string more than PLA on a well-tuned printer; OrcaSlicer’s docs say as much. Prusa’s closing tip is a heat gun at around 200 °C passed over the strings for one or two seconds, no longer or the part deforms. For a functional PETG bracket, two seconds of hot air is a better use of your evening than a fourth retraction tower.
