Stepper Motor Sizing and Torque Curves for CNC Router Builds: NEMA 17 vs 23, Microstepping, and Resonance

Picking a stepper motor for a CNC router build by holding torque alone is one of the most common sizing mistakes hobbyists make, and it's understandable — holding torque is the one number every motor listing prominently advertises. But holding torque is measured at zero RPM, and a CNC router's axes rarely sit still while cutting; what actually determines whether a motor can keep up with your feed rate under load is its torque curve — how much torque it can still deliver as RPM climbs. This guide covers how to actually size steppers for a router build, when NEMA 17 is enough and when you need NEMA 23, and how microstepping and mechanical resonance interact with your torque budget in ways the holding-torque spec sheet won't tell you.

Why Holding Torque Isn't the Number That Matters

Every stepper motor's usable torque falls off as step rate (and therefore RPM) increases, because the driver has less time per step to fully energize the coil against its own inductance. A motor rated for 0.8Nm of holding torque might only deliver 0.3–0.4Nm at the RPM your router actually runs at during a rapid move or an aggressive feed rate. This is why two motors with identical holding torque specs can perform very differently on the same machine — the shape of the torque-speed curve, which is a function of coil inductance and the driver's supply voltage, matters more than the peak number. Higher supply voltage (48V vs 24V drivers, for example) pushes current into the coil faster and extends useful torque further into the RPM range, which is part of why many CNC-focused stepper driver setups run higher voltage than typical 3D printer stepper drivers.

NEMA 17 vs NEMA 23: When Each Makes Sense

NEMA 17 motors (42mm faceplate) are standard on most desktop CNC routers in the Wolfpawn 4040 Pro class and similar lightweight gantry machines, where axis mass is modest and cutting loads are moderate — wood, plastics, and light aluminum passes with appropriately conservative feeds and speeds. They're a reasonable default because the machines they ship on are built around that torque budget from the frame rigidity down. NEMA 23 motors (57mm faceplate) deliver substantially more torque and are the right call when you're driving a heavier gantry, running ballscrews instead of belts (which generally need more torque to overcome mechanical friction and preload than a belt-driven axis), cutting harder materials like aluminum or steel routinely, or running a machine large enough that axis mass alone eats into your torque budget before the cutting load is even factored in. Swapping NEMA 17 for NEMA 23 on a machine designed around 17s is rarely as simple as bolting on a bigger motor — mounting plates, coupler bore sizes, and driver current ratings all need to match, and the added motor mass on a moving axis (as opposed to a fixed one) can itself become a problem.

Microstepping and the Torque You're Actually Losing

Microstepping (running a driver at 1/8, 1/16, or 1/32 steps instead of full or half steps) smooths motion and reduces resonance at low speeds, but it comes with a real, often-overlooked cost: torque per microstep is not linear with the sine/cosine current profile drivers use, so at very fine microstepping settings the motor delivers meaningfully less torque at the same current setting than it would at full step. This usually isn't a problem for the smooth, moderate-speed motion typical of CNC routing, but it matters if you're pushing a motor close to its torque limit already — dropping from 1/16 to 1/8 microstepping can recover usable torque headroom at the cost of slightly coarser motion resolution, which is a trade worth knowing about rather than discovering as an unexplained stall.

Resonance: The Torque Killer Nobody Budgets For

Every stepper motor has a mechanical resonance frequency (typically in the low hundreds of pulses per second on unloaded NEMA 17/23 motors) where torque output drops sharply due to the rotor's own oscillation working against the driving field — this is often called the "mid-band resonance" zone. If your machine's typical rapid-move step rate happens to land in that resonance band, you'll see missed steps or audible motor "singing" even though your calculated torque budget looks fine on paper. This is a large part of why TMC-family drivers with StealthChop/SpreadCycle (covered in this site's guide to tuning TMC2209 drivers) have become popular even on machines that were originally fine with generic step/direction drivers — better current waveform shaping meaningfully reduces resonance-band torque loss, not just noise.

Motor ClassTypical Holding TorqueGood Fit For NEMA 17 (42mm)0.4–0.9NmBelt-driven light gantry routers, wood/plastic cutting NEMA 23 (57mm)1.0–3.0NmBallscrew drives, heavier gantries, routine aluminum work NEMA 23 High-Torque / NEMA 343.0Nm+Larger format machines, steel/hard material work

Sizing Your Own Build

Work backward from your actual axis: calculate the linear force needed to overcome your worst-case cutting load plus gantry mass acceleration, convert that to required torque at your leadscrew or belt pulley's mechanical advantage, and then check that required torque against the motor's torque curve at the RPM your target rapid speed actually corresponds to — not against the holding torque spec. Most stepper manufacturers publish torque-speed curve graphs for exactly this reason; if a listing only gives holding torque with no curve, treat that as a yellow flag rather than assuming it performs like a similarly-rated motor that does publish one. Build in headroom — sizing a motor to exactly your calculated worst case with no margin leaves no room for a dull bit, a slightly aggressive feed override, or a colder shop day thickening your cutting oil and adding friction.

Safety Notes

An undersized or resonance-stalled motor doesn't fail safely — it silently loses position while the controller keeps commanding moves as if nothing happened, which means your next cut can plunge the bit into material (or the spoilboard, or a clamp) at the wrong depth with no warning. If you notice a job finishing with visible steps skipped or dimensional error, stop and address the torque budget rather than just re-running the job and hoping it holds up next time. When increasing motor current to chase more torque, verify your driver and motor are both rated for the higher current and that the driver has adequate heatsinking — overcurrent on an undersized driver is a real fire risk, not just a reliability concern.

Getting stepper sizing right up front saves a lot of downstream troubleshooting that otherwise gets misdiagnosed as a firmware or backlash problem. When in doubt, oversize modestly rather than sizing to the edge of your calculated torque budget — the cost difference between a well-matched motor and a slightly oversized one is small compared to the time lost chasing intermittent skipped steps on a machine that's running at its limit.