TPU Max Pressure Advance: Tuning Flexible Filament Without Clicks or Blobs

Why TPU fights back when you tune pressure advance

Tuning tpu max pressure advance values is the single most frustrating calibration exercise in the hobby. The standard procedure that works beautifully for PLA and PETG — print the classic pressure advance test pattern, look at the corners, pick the cleanest transition — produces numbers that are either obviously wrong or stubbornly inconsistent for flexible filament. Users swap extruders, re-seat the filament, and re-run the test three times before concluding that their printer is broken. It is not broken. TPU is physically different from rigid filament in ways that break the standard assumptions of pressure advance, and once you understand why, the tuning procedure has to change with it.

This article walks through why TPU is different, what “max pressure” actually means for a flexible material, and how to land on working values for both Bowden and direct-drive setups. I will also cover where to stop tuning and accept what you have, because there is a point of diminishing returns specific to TPU that does not exist for rigid filaments.

The physical model: why pressure advance exists at all

Pressure advance compensates for the fact that melted filament inside a nozzle behaves like a slightly compressed spring. When the extruder pushes filament, pressure builds in the melt chamber before extrusion actually speeds up. When the extruder stops, pressure keeps pushing melt out for a moment. This lag manifests as corners printed too light at their start and too heavy at their end — extrusion lags what the G-code commanded.

The pressure advance algorithm predicts the pressure curve and pre-adjusts the extruder to counteract it. A higher PA value means the extruder retracts harder and pre-builds pressure more aggressively. A lower PA value is more conservative. For PLA, typical values are 0.02-0.05 on Bowden and 0.02-0.04 on direct drive. For TPU, everything changes because TPU itself has material properties a rigid filament does not.

TPU compresses along the extruder path before it even reaches the melt zone

The single most important fact about flexible filament is that the filament itself is springy. When a direct-drive extruder pushes a 95A TPU filament forward, the filament compresses axially inside the Bowden tube or between the extruder gears and the melt zone. The extruder has pushed 1 mm of filament; only 0.6-0.8 mm has actually reached the hot end. The rest is stored as compression in the filament itself, which will release at unpredictable moments.

This is the effect that makes TPU stringy and blobby when you apply rigid-filament tuning methods. Pressure advance assumes the filament is incompressible. For PLA this is approximately true. For TPU it is false, and the falsehood scales with how flexible the filament is. A 98A TPU behaves almost like PETG. A 85A TPU behaves like a rubber band. Every point between has its own pressure-advance ceiling.

Why the “max” in tpu max pressure matters

The concept of maximum pressure advance for TPU is a ceiling, not a target. Above a certain PA value, the extruder cannot apply retraction or pre-pressurisation fast enough because the filament simply buckles or skips in the gears before the pressure transfers to the melt. For direct-drive extruders on 95A TPU, that ceiling is typically around 0.08-0.12. For Bowden setups it is more like 0.03-0.06, because the long path amplifies compression dramatically.

When you tune TPU you are not trying to hit the “right” PA number — you are trying to find the highest PA value where corners look clean without triggering skip, clicks, or grinding at the extruder. Once you hit the ceiling, lower PA is always safer than higher PA because the failure mode of over-tuned TPU is not ugly corners, it is a skipped step that ruins the print hours later.

The modified tuning procedure: slow, short, repeated

Here is the procedure that actually works for TPU, refined from hundreds of reports in the Klipper and Marlin communities.

Step one: print a pressure advance tower but at half the usual speed. PLA tuning towers run at 100 mm/s or higher. For TPU, print the same tower at 40-50 mm/s. The whole point of high-speed tuning is to amplify the PA signal, but TPU cannot be run that fast reliably, so tuning must happen at the same speed as production prints.

Step two: use short test segments. Long linear moves mask PA problems because the extruder has time to stabilise. Short alternating moves with sharp corners expose PA issues immediately. Tune on a test pattern with many small corners rather than a long tall tower.

Step three: run the test three times, not once. TPU tuning is noisy. A run that gives PA=0.08 the first time often gives PA=0.04 the second time depending on how the filament was sitting on the spool. Run the test three times with 10-minute intervals and take the median, not the best-looking single run.

Step four: validate with a real part, not another test print. The final PA number only matters if a real functional part prints cleanly. Print a small TPU part from your normal workflow — a phone case, a strap, a gasket — and look for the same failure modes you were originally trying to fix. If they are gone, you are done.

Direct-drive versus Bowden: different tuning ceilings

Direct-drive extruders dominate TPU printing for a reason. The short distance between the gears and the melt zone means less filament compression, which means PA tuning is more meaningful. Typical working PA values on direct drive:

• 95A TPU: 0.06-0.10
• 90A TPU: 0.04-0.08
• 85A TPU: 0.02-0.05

Bowden setups can print TPU but with compromises. The long PTFE tube between the extruder and hot end amplifies compression, so the effective PA range collapses to 0.015-0.04 even for 95A TPU. Going higher produces no improvement and often causes skips. Bowden users should consider retraction tuning and print-speed reduction before trying to chase higher PA — those adjustments give more return on effort.

If your printer supports both modes (some do via conversion kits), the decision tree is simple. Printing 95A TPU occasionally: Bowden is fine. Printing flexible TPU regularly at 85A-90A: direct drive is worth the conversion.

Common failure modes and what they mean

When TPU tuning goes wrong, the symptoms fall into a few recognisable categories.

Stringing between separate parts: PA is too low and retraction is compensating by running too high. Lower retraction distance, then try raising PA by 0.01 increments until the strings come back, then back off one step.

Thick blobs at corner starts: PA is too low and the extruder is depositing extra material when acceleration ramps up. Raise PA by 0.01 per test run.

Extruder clicking or gear grinding mid-print: PA is at or above the material’s ceiling. Drop by 0.02 and slow the print.

Inconsistent extrusion on long walls: PA is fine but the filament itself is compressing unevenly through the tube or gears. This is a symptom of overly old, damp, or contaminated filament, not a tuning problem. Dry the filament and retest.

Layer shifts hours into a print: the extruder briefly skipped a step because PA was too high. Drop PA immediately and re-tune with validation runs.

Print speed and TPU flow: the other two variables

Max pressure advance interacts with print speed in ways that do not apply to rigid materials. At 20 mm/s any PA value 0.00-0.20 produces similar prints because the extruder has time to recover from any lag. At 100 mm/s the effective PA window narrows to maybe 0.04-0.08 because faster moves amplify any PA-related error.

The practical implication is that TPU tuning should always be done at the speed you actually print at. Tuning at 40 mm/s and then printing production parts at 80 mm/s is the most common reason people re-tune their PA three times and get three different numbers. Lock the speed first, tune PA second.

Flow rate is a separate variable that users often confuse with PA. Low flow produces thin, under-extruded walls. Wrong PA produces correctly-extruded walls with bad corners. If the walls themselves look wrong, fix flow first by dialing in the extrusion multiplier; then come back to PA.

When to stop chasing the perfect number

Rigid filament PA tuning converges. TPU PA tuning does not, quite. After two or three runs you will have a range — say 0.05-0.08 on 95A direct drive — within which real parts print fine and outside which they fail in obvious ways. Pick the middle of that range and stop. Chasing a single exact number is not productive for TPU because the material itself introduces variability that exceeds the precision of the tuning method.

This is the hardest lesson for users who came from PETG or PLA, where tuning converges on a single value you can lock in and forget. TPU demands a range and a willingness to accept that any given print might land anywhere inside that range. Once you accept this, TPU prints become reliable. Until you accept this, you will keep re-tuning.

What a good TPU print looks like when tuning is correct

The signs of well-tuned TPU at the ceiling of its pressure advance range are specific. Corners are crisp, not rounded or blobby. Thin walls extrude consistently without visible flow variation. Retraction movements leave no visible strings between towers. The extruder motor runs smoothly without audible clicking on small moves. Layer lines look uniform across the part, especially on the outer wall. These five signs together indicate the tuning is in the right range. The absence of one or more indicates a specific problem — and knowing which sign is missing tells you which variable to adjust next rather than restarting the whole tuning cycle from zero.