Retraction & Temperature Tuning Calculator
Whether your test tower can resolve what you are tuning — 5 degrees is readable, 1 is noise, and 0.5 mm steps are wrong for direct drive.
Retraction step against band count
What a temperature step actually changes
5 °C is readable. 1 °C is reading noise.
Melt viscosity follows Arrhenius, so a 5 °C step changes it by about 21% — plainly visible in stringing and bridging. A 1 °C step changes it by 4.7%, which is inside the run-to-run variation of a printed test. Across the plausible range of flow activation energies the 5 °C figure runs 15–27%, so the conclusion survives the uncertainty even though the exact number doesn't. The received convention is well chosen.
But 0.5 mm retraction steps are right for Bowden and too coarse for direct drive. A direct extruder needs about a millimetre in total, so 0.5 mm steps give four usable bands across the whole range — enough to know roughly where the optimum isn't. Use 0.2 mm and you get ten. A long Bowden needs several millimetres, where 0.5 mm gives twenty and is comfortably right. Most guides quote 0.5 mm regardless of drive, which is half of them wrong.
And the reason Bowden needs so much more is computable. The filament column inside the tube is a spring: at ~5 MPa of melt pressure and a modulus of ~2,500 MPa, a 600 mm column compresses 1.2 mm before anything happens at the nozzle. Add the melt-zone relief of roughly 0.5–1 mm that both drives pay, and the familiar "0.5–2 mm direct, 4–7 Bowden" falls out without having to be looked up.
One number for scale: a visible 20 mm string is 0.157 mm³ — 0.065 mm of 1.75 mm filament. A single 0.5 mm retraction step moves eight strings' worth. The test steps are enormously coarse relative to the thing being measured, which is why the best band is so often ambiguous and why two people reading the same tower disagree.
- Tune temperature first, then retraction. Viscosity determines how readily the melt oozes, so a temperature change moves the optimum retraction while the reverse isn't true. And the temperature tower is the quicker test.
- Retraction speed matters as much as distance and is tested far less. Too slow and the melt keeps oozing during the move; too fast and the drive gear grinds a flat that then feeds unevenly for the rest of the print.
- Too much retraction is its own failure mode. Pulling melt back into the heat break lets it cool against a cold wall and jam — a clog partway through a long print, which is far more expensive than a few strings. If stringing persists at the top of a sensible range, the answer is drier filament or a lower temperature.
- Print two identical towers. If the same band wins on both, the reading is real; if not, your steps are finer than the noise. That's a better test than any amount of squinting at one.
How to use
- Tune temperature before retraction, since viscosity moves the optimum.
- Check the step size gives enough bands before printing the tower.
- Print two identical towers and compare which band wins on each.
- Read the tower in good light at a shallow angle, not straight on.
Frequently asked questions
What temperature step should a tuning tower use?
Five degrees, and the convention is well chosen rather than arbitrary. Melt viscosity follows Arrhenius, so a 5 C step changes it by about 21 per cent — plainly visible in stringing and bridging. A 1 C step changes it by 4.7 per cent, which is inside the run-to-run variation of a printed test, so whichever band looks best is as likely to be chance as signal.
Is a 1 degree temperature tower worth printing?
No. The viscosity difference between adjacent bands is under five per cent, which is smaller than the variation between two prints of the same band. You would be reading noise and calling it a result. If five-degree bands look identical, that is genuine information — it means the setting is not sensitive there, not that you need finer steps.
What retraction step size should I test at?
It depends entirely on the drive, which most guides ignore. Direct drive needs about a millimetre in total, so 0.5 mm steps give only four usable bands and 0.2 mm is the right figure. A long Bowden needs several millimetres, where 0.5 mm gives twenty bands and is comfortably right. The commonly quoted 0.5 mm is correct for one of the two and far too coarse for the other.
Why does Bowden need so much more retraction than direct drive?
Because the filament column inside the tube is a spring. At a melt pressure around 5 MPa and a filament modulus near 2,500 MPa, a 600 mm column compresses 1.2 mm before anything at all happens at the nozzle. Add the melt-zone relief of roughly half a millimetre to one, which both drive types pay, and you get the familiar 0.5 to 2 mm direct against 4 to 7 Bowden.
How much filament is a string, actually?
Startlingly little. A visible 20 mm string at 0.1 mm across is 0.157 cubic millimetres, which is 0.065 mm of 1.75 mm filament. A single 0.5 mm retraction step moves about eight strings worth. The test steps are enormously coarse relative to the thing being measured, which is why the best band is so often ambiguous and why two people reading the same tower disagree.
Should I tune temperature or retraction first?
Temperature, always. Viscosity determines how readily the melt oozes, so changing temperature moves the optimum retraction while the reverse is not true — tuning retraction at the wrong temperature produces a number that stops working the moment the temperature is corrected. The temperature tower is also the quicker of the two tests.
Does retraction speed matter as much as distance?
Very nearly, and it is tested far less often. Too slow and the melt keeps oozing while the retraction is happening; too fast and the drive gear grinds a flat on the filament, which then feeds unevenly for the rest of the print. Direct drives tolerate 40 to 60 mm a second; long Bowden setups usually want less, because the tube has to transmit the movement.
Can too much retraction cause problems?
It causes the worst problem in the list. Pulling molten plastic back into the heat break lets it cool against a cold wall and jam, which shows up as a clog partway through a long print rather than immediately — far more expensive than a few strings. If stringing persists at the top of a sensible range, the answer is drier filament or a lower temperature, not more retraction.
How do I know if my tower reading is real?
Print two identical towers and compare. If the same band wins on both, the reading is real; if a different band wins each time, your steps are finer than the noise and you are choosing at random. That single test settles more arguments than any amount of squinting at one tower, and it costs one extra print.
Why does my stringing come back after I fixed it?
Usually moisture rather than a setting drifting. Wet filament steams in the melt, and the vapour pushes plastic out of the nozzle in a way no retraction distance can prevent — the mechanism is not pressure, it is boiling. If a previously tuned profile starts stringing, dry the filament before touching retraction, because the retraction was probably right.
Does filament type change the retraction setting?
Considerably. PETG and TPU string far more than PLA at the same settings, because they are more viscous and stickier in the melt, and TPU compresses in the drive in a way rigid filament does not. A retraction number is a property of the machine and the material together, so it is worth keeping a tuned figure per filament type rather than per printer.
How should I look at a printed stringing tower?
In good light, at a shallow angle, and preferably against a dark background. Stringing is a surface phenomenon and thin strands are nearly invisible looking straight on — which is one reason two people reading the same tower come to different conclusions. Rotating it slowly under a lamp shows far more than a photograph does.
🔒 This tool runs entirely in your browser. Nothing you enter is uploaded, logged, or stored.