Skip to content
PrintReckonSupport Us

TPU 85A: genuinely soft, genuinely slow, and the labour is the biggest line on the bill

People who find 95A too stiff assume 85A will be a small step down. It is not. The Shore A scale compresses hard at the top end, so the perceptual gap between 95A and 85A is far wider than the ten points suggest — the difference between a stiff shoe sole and a pencil eraser. If you have been printing 95A and want something you can genuinely squash, this is the material, and it is a different printing problem.

The filament path must be closed, everywhere

Soft filament under compression buckles at very low force. Between the drive gear and the melt entrance there can be no gap at all, because the filament will find it and coil up inside the extruder rather than going forward. A machine that prints 95A acceptably may simply refuse 85A, and the difference will be a two-millimetre unsupported span you cannot see without dismantling the hotend.

  • Direct drive only. A Bowden tube is a long unsupported column; this is not a tuning problem.
  • Speeds under about 15 mm/s. Feed force rises with speed, and the buckling limit is where the material gives up. Patience is the setting.
  • Retraction close to zero. Retracting soft TPU stretches it like a rubber band. The extruder then has no idea where the material is, and the next extrusion arrives late.
  • Reduce the drive gear tension. Over-tightened knurled gears chew a groove into soft filament and stop feeding entirely.

Compensation is meaningless — print, measure, print again

Stored shrinkage runs 0.8% to 2.0%, which looks like a figure you could scale for. It is not, because the part also recovers elastically after printing, keeps recovering for hours, and deforms under the light pressure of the caliper jaws while you measure it.

There is no honest scale factor for a material this soft. The workflow is to print a small test article of the actual feature that matters, let it sit overnight, and adjust the model. On every other material this site covers, the shrinkage compensation calculator does this work for you; on 85A the calculator would give you a confident number about a part that has not finished moving.

The labour is the line item people forget

Support material welds itself to a soft surface far more tenaciously than to a rigid one, and you cannot snap it off — the part flexes away from the pliers instead of letting the support break. Removal is slow, fiddly hand work, and the same softness means any trimming has to be done carefully or you cut into the part.

Cost it properly and the shape of the bill changes completely. Take a 60 g part, 7 hours on a machine drawing 90 W, a $45 spool, $0.30 an hour of machine wear, a 12% failure rate — and twenty minutes of hand finishing charged at $25 an hour:

  • Labour — $8.33.
  • Machine time — $2.10
  • Filament — $2.70
  • Failure allowance — $0.67
  • Electricity — $0.11, for 0.63 kWh

$13.91 in total.

The plastic is under a fifth of the cost. That is a genuinely different shape of bill from anything else on this site, and it means the useful optimisation is designing supports out of the model rather than shopping for cheaper filament. Twenty minutes of hand finishing is the assumption carrying that total, and it is the line most people never write down at all. Take it out of the print cost calculator and 85A stops looking expensive — and that is how the job comes to be quoted at a loss.

Designing for it

Softness is not only a material property here — the printed lattice does much of the work. Two walls and sparse infill in 85A gives something close to foam; four walls and dense infill gives a part that feels a whole hardness grade stiffer. Before ordering an even softer spool, change the wall count.

Small holes come out undersized and often close entirely, bridges sag badly, and fine text disappears. Design with generous radii, avoid unsupported spans, and expect the first attempt to be a learning exercise.

Nozzle 215 to 235 °C, bed 30 to 50 °C, no enclosure. Drying at 55 °C for four to six hours, and 85A is the least tolerant of moisture in the family — a damp spool prints visibly furry within the first layer.

What TPU 85A is genuinely bad at

  • Any extruder without a fully constrained path.
  • Speed, comprehensively.
  • Retraction-heavy geometry — a model with many small separated features is the worst case.
  • Dimensional work of any kind.
  • Load-bearing structure. Tensile of 15 to 30 MPa is the lowest in the flexible family, and it stretches enormously before it breaks.
  • Cost. $35 to $65 per kilogram, plus the labour above and a higher failure rate than you are used to.

Where the difficulty is repaid

Seals and gaskets that must conform to an imperfect surface. Prosthetic and orthotic liners. Soft-touch grips and overmoulded handles. Vibration isolation feet that actually isolate. Squeezable toys and stress objects. Anatomical models where the feel is the point.

For anything that must hold its shape while yielding a little — wheels, bumpers, semi-rigid mounts — 98A does the job at a fraction of the difficulty, and the comparison between the two grades covers where the boundary really sits.

More in Flexible filaments