Steps per millimetre is the number that tells the firmware how far the machine actually moves. Get it wrong and every dimension comes out proportionally wrong — a part that should be 100 mm arrives at 97, and no amount of slicer tweaking fixes it.
Belt axes
One turn of the motor pulls a length of belt equal to the pulley's tooth count times the belt pitch. A 20-tooth pulley on GT2 belt, which has a 2 mm pitch, moves 40 mm per turn.
Divide the steps per turn by that travel. A 1.8 degree motor is 200 full steps, at 16× microstepping that is 3,200, and 3,200 divided by 40 is exactly 80 steps per mm.
That is where the familiar 80 comes from, and why 20-tooth GT2 pulleys became the default — the arithmetic lands on a whole number.
Leadscrew axes
A screw is simpler: the travel per turn is the lead, full stop. Divide steps per turn by the lead.
The trap is confusing lead with pitch. A common T8 leadscrew has a 2 mm thread pitch but four separate thread starts, so one turn advances the nut 8 mm. Its lead is 8, not 2.
Using the pitch instead of the lead gives a figure four times too high, and the Z axis then moves a quarter of what it should — which looks like a catastrophic failure rather than a settings error.
Microstepping does not add accuracy
Microstepping smooths motion and reduces noise. It does not make the machine more accurate, because the torque holding the motor at a microstep position is a fraction of the full-step torque.
At 16× microstepping, the holding torque at a single microstep is a few percent of full. Any load at all pulls the rotor back toward the nearest full step.
So the number to think of as the machine's real resolution is the full-step distance, not the microstep distance. The calculator reports both.
Calibrating rather than calculating
The calculated figure is the starting point. Belt stretch, pulley tolerance and screw error mean the real figure differs slightly.
The standard method is to command a long move — 100 mm rather than 10, because the error scales and is easier to measure — mark the start, measure the actual travel, and correct: new steps per mm equals old times commanded over actual.
Do it over the longest move the axis allows. Measuring a 10 mm move with a rule introduces more error than it finds.
Extruder steps are the same problem
The extruder is a rotary axis measured in millimetres of filament, and it follows the same formula with the hobbed gear's effective diameter in place of the pulley.
Effective diameter is hard to measure because the teeth bite into the filament, so extruder steps are almost always found by calibration: mark the filament, extrude 100 mm, measure what actually went in.
Get the extruder right before touching flow rate. Flow compensates for filament diameter and material; steps per mm is the machine. Mixing the two makes both wrong.