TPU print settings:
nozzle, bed, speed, retraction and drying

For a 95A TPU, start at 220-230 °C on the nozzle and a bed between 30 and 60 °C, print slowly at 20-30 mm/s unless the spool is sold as a high flow grade, and keep retraction short or switch it off at first. Dry the filament before printing, at 60-70 °C for 4 to 8 hours depending on the maker, because TPU takes up water from the air and prints badly when wet. Softer grades such as 85A need a wider nozzle and more patience. Prusa, Polymaker and Bambu Lab agree on going slowly and keeping the spool dry. They part ways on fan speed, bed temperature and retraction length.

We read the documents for this page on 26 October 2026: Prusa's knowledge base article on flexible materials, the data sheet for Polymaker PolyFlex TPU95, Bambu Lab's data sheets for TPU 85A, TPU 90A, TPU 95A HF and TPU for AMS, and Bambu Lab's TPU printing guide. Prusa's article covers flexible filament in general and names no single product. The filament types hub places TPU among the other plastics, and the PETG settings page uses the same layout for a rigid filament. Nanopack has no commercial tie to any of these companies.

TPU settings by manufacturer

Bambu Lab's "under 200 mm/s" is a ceiling for its own printers and profiles. Volumetric flow limits what TPU can actually do. On its store page Bambu Lab compares TPU 95A HF with its standard TPU 95A at the same 230 °C nozzle: 12 mm³/s against 3.6 mm³/s of maximum volumetric speed, which in its test model meant 147 mm/s against 44 mm/s. A TPU that is not sold as high flow belongs much closer to Prusa's 20 mm/s.

Shore hardness: 85A, 90A and 95A

TPU grades are named by Shore hardness, measured with a durometer. Bambu Lab's guide orders the common grades from hardest to softest as 77D, 70D, 68D, 55D, 95A, 90A, 85A, 83A, 80A, 75A and 70A. The D scale is used for the harder end. A higher number on the same scale means a stiffer, less stretchy filament, so 90A is softer than 95A, and 85A is softer again.

Softness is what makes printing hard. Prusa says that the softer the filament is, the harder it is to print. Bambu Lab recommends 85A or harder for stable printing, says softer TPU such as 83A or 80A bends during feeding and can buckle or tangle, and states that its printers do not support TPU below 85A. Its TPU 85A also cannot run through a 0.4 mm nozzle: the data sheet lists only 0.6 and 0.8 mm, and the guide marks 0.2 mm and 0.4 mm as not compatible.

At the other end, Bambu's TPU for AMS is rated 68D. That makes it far stiffer than any A grade. Its data sheet gives a Young's modulus of 1190 MPa along the layers, where TPU 95A HF has 9.8 MPa. It is the only TPU Bambu Lab lets the AMS feed automatically. For the other grades, the AMS HT can only hold the spool as a sealed box, and the filament has to leave through a dedicated TPU outlet at the back.

Nozzle temperature

Polymaker's and Bambu Lab's ranges share 220-230 °C. Prusa's starts at 230 °C, so 230 °C is the only figure all of them allow. Prusa's article gives two adjustments that pull in opposite directions. Raising the nozzle by 5 °C lowers the resistance of the filament, which helps when the extruder struggles. Bambu Lab, in its stringing advice, suggests lowering the nozzle by about 5 °C because TPU flows easily and oozes when too hot, and then cutting the maximum volumetric flow by 0.4 mm³/s to match.

The test bars tell you where each company actually prints. Polymaker printed its TPU95 specimens at 230 °C on a 50 °C bed with the fan on. Bambu Lab printed TPU 95A HF at 230 °C on a 35 °C bed at 140 mm/s. Polymaker's bars came from the top of its own range and Bambu Lab's from the middle of its range.

Print speed and why TPU is hard to feed

Most TPU failures happen before the plastic reaches the nozzle. A soft strand pushed by the drive gears can bend and tangle in them, which Prusa lists among its warnings next to clogging and oozing. Prusa recommends TPU only to experienced users and suggests loosening the idler so the gears push the filament gently. It also warns that automatic filament change may not work properly with flexibles.

Bambu Lab's guide spends more words on the path than on the hotend. It recommends a sealed dry box or the AMS HT with low feeding resistance, raising the container so the slack filament does not drag over the bed, and switching off flow dynamics calibration before a TPU print. It also advises a new nozzle or a cold pull first, since leftover carbon or glass fibre from earlier prints adds resistance. The nozzle clog guide explains the cold pull.

Prusa gives no figure for Bowden printers. A reader comment under its article, about the Bowden-fed MINI+, suggests careful idler pressure, a harder TPU and minimal retraction, and accepts some stringing as the price.

Retraction and stringing

Retraction advice varies more than anything else here. Polymaker's 3-6 mm at 40-60 mm/s is several times Bambu Lab's 0.8-1.4 mm for TPU 95A HF and 0.4-0.8 mm for the 85A and 90A grades. Prusa gives no number at all. It suggests starting with retraction low or off and raising it gradually while checking the result. A long retraction pulls soft filament back through the extruder, where it can compress instead of moving, so starting short is the safer test.

Bambu Lab's stringing list goes beyond retraction. It suggests printing one object at a time to cut travel moves, lowering the nozzle temperature by about 5 °C, adding 0.2 mm to the retraction length when oozing is severe, and drying the spool fully. Damp TPU strings even with good settings.

Bed, build surface and fan

The bed figures do not overlap across all three sources. Bambu Lab says 30-35 °C with glue, Polymaker 25-60 °C and Prusa 60-75 °C, with larger objects needing the higher end. Adhesion is rarely the problem with TPU. Bambu Lab says it usually sticks strongly, and Prusa warns that very soft flex can bond to smooth PEI hard enough to damage it.

Prusa's textured powder-coated sheet needs no separation layer. On smooth or satin sheets Prusa asks for glue stick or Kapton tape. Bambu Lab's guide says every plate it sells except the Cool Plate SuperTack works with TPU, and its data sheets list glue. To remove a large flat part, Bambu Lab suggests lifting one corner, spraying isopropyl alcohol into the gap and peeling slowly.

On cooling the sources split. Prusa wants the print fan off. Polymaker and Bambu Lab both say the fan should be on, and Polymaker's test bars were printed with it on.

How strong TPU is, and how it compares with PETG

The A grades keep most of their strength across the layers. TPU 95A HF holds 22.3 of its 27.3 MPa in the Z direction, about 82%. The 68D grade behaves more like a rigid plastic: half its tensile strength and almost all of its stretch disappear across the layers. Polymaker reports its numbers differently, as stress at set strains, and measures 551% elongation at break for TPU95 in the XY direction.

Against PETG, TPU loses on tensile strength and wins on impact. Bambu Lab's PETG Basic sheet gives 51 MPa of tensile strength and 34.2 kJ/m² of impact strength along the layers. Its TPU 95A HF sheet gives 27.3 MPa and 123.2 kJ/m² in the same tests. For a part that has to hold a load without stretching, PETG is the stronger choice. For a part that gets hit, bent or squashed, TPU survives better.

Moisture and drying

Prusa's 60 °C is the lowest figure here. For comparison, Bambu Lab gives 0.45% water uptake for its PETG Basic. Prusa's drying article names TPU, with polyamide and PVA, among the filaments that need drying more often than PLA. Wet TPU strings, bubbles and bonds poorly between layers, according to Bambu Lab, which tells users to store it below 20% relative humidity with desiccant and to dry it again after long storage. Bambu Lab ships these grades on spools rated to 90 °C. The general filament drying guide covers how to tell when a spool needs it, and the AMS 2 Pro and AMS HT drying guide covers Bambu Lab's drying units.

Where the documents disagree

The fan is the clearest split. Prusa keeps it off, while Polymaker and Bambu Lab switch it on, and Polymaker printed its test bars with it running. The bed figures barely touch: Bambu Lab's 30-35 °C and Prusa's 60-75 °C have nothing in common, and Polymaker's 25-60 °C spans most of the gap. Retraction runs from Prusa's "low or off" to Polymaker's 6 mm.

Part of this is hardware. Bambu Lab writes for its own printers and hotends, Prusa writes for any flexible on its machines, and Polymaker gives one set of numbers for a 0.4 mm nozzle. Bambu Lab also notes that its test bars were annealed and dried at 70 °C for 12 hours before testing, yet advises against annealing TPU prints, because parts that are not very simple in shape can deform. On a printer none of them wrote for, start from the filament maker's sheet, keep the speed low, and change one setting at a time.