May 22,2026
Stepper Motor Engineering Guide for Precision Motion Control
A technical guide to stepper motor operation, covering torque‑speed behavior, resonance damping, microstepping accuracy, and comparisons with servo and BLDC systems.
Executive Summary
Stepper motors remain a cornerstone of precision motion control in industrial automation, CNC machinery, 3D printing, and robotics. This guide covers the electromechanical principles, drive modes, resonance mechanisms, and technology trade-offs that determine whether a stepper system performs reliably — with every formula referenced to its published source and every number reproducible from stated assumptions.
One clarification up front, because it is the most common misconception in this field: microstepping increases resolution and dramatically reduces noise and vibration — it does not proportionally increase absolute positioning accuracy. Accuracy remains bounded by the motor's step tolerance, detent torque, magnetic hysteresis, and load angle. Sections 4 and 6 quantify what microstepping can and cannot deliver.
1. Operating Principle & Torque-Speed Characteristics
A stepper motor is a synchronous machine that converts digital pulses into discrete angular increments. A standard two-phase hybrid stepper with 50 rotor teeth advances one quarter of a tooth pitch per full step, giving 1.8° per step and 200 steps per revolution.
1.1 Torque development
Developed torque varies sinusoidally with the load angle — the electrical displacement between stator flux and rotor position:
T = TH · sin(φs − φr)
TH = holding torque (Nm); φs = stator flux angle and φr = rotor angle, both in electrical degrees. The difference (φs − φr) is the load angle: the rotor always lags the flux by whatever angle generates the torque the load demands. At 90 electrical degrees the motor develops TH; beyond that it loses synchronism.
φel = (n / 4) · φmech
n = full steps per revolution. For n = 200 the factor is 50 (the rotor tooth count): one full step of 1.8 mechanical degrees equals 90 electrical degrees, and half-stepping equals 45 electrical degrees. This conversion is the basis of every ripple and resonance calculation that follows.
With ideal sine/cosine current pairs the theoretical holding torque is independent of flux direction:
TH = k · √(IA2 + IB2)
Real motors deviate from this ideal — see §4 for the measured magnitude of that deviation.
1.2 Reading the speed-torque curve

Figure 1: Torque-speed characteristic — illustrative NEMA 34, bipolar drive. Curves are drive-voltage and current dependent; always use the datasheet curve for your exact motor-drive combination.
| Curve / Region | Definition | Engineering Significance |
|---|---|---|
| Pull-out torque | Maximum torque sustainable at a given speed while staying in synchronism (motor already running) | The absolute operating ceiling. Exceeding it causes step loss or stall. |
| Pull-in torque | Maximum torque against which the motor can start, stop, or reverse instantly at a given pulse rate without a ramp — and it is inertia-dependent | Defines the start/stop region. The published curve is valid only for the stated load inertia; adding inertia shrinks this region. |
| Slew range | Between pull-in and pull-out curves | Reachable only with acceleration/deceleration ramping. Most practical motion operates here. |
| Load torque demand | Friction + acceleration + gravity/process torque across the speed range | Must stay below pull-out at every speed, with margin (commonly ≥30–50%). |
| Stall risk zone | Demand exceeds available torque | Redesign required: larger motor, higher drive voltage, gearing, or reduced load. |
Why stepper torque collapses with speed: winding inductance limits how fast current can be established in each phase. As stepping frequency rises, the drive has less time per step to reach rated current, so torque falls. This is why drive bus voltage matters as much as motor selection — raising the supply voltage (within the drive's rating) is often the cheapest way to extend usable speed.
In Figure 1 the load line intersects the pull-out curve at about 1,343 full-steps/s. For a 1.8° motor that is 1,343 × 1.8 ÷ 360 × 60 ≈ 403 rpm — the maximum speed at which this load can be sustained regardless of acceleration strategy.
2. Resonance: Two Different Mechanisms, Two Different Fixes
A stepper system is a mass-spring oscillator: rotor plus load inertia is the mass, and the magnetic restoring torque is the spring. It is inherently underdamped. Two distinct instability phenomena arise, and confusing them leads to the wrong remedy.
2.1 Low-speed (mechanical) resonance
The natural frequency of the motor-load system, expressed in full steps per second, is:
F0 = √(n · TH / JT) / (4π)
n = full steps/rev; TH = holding torque (Nm) for the selected drive mode and current; JT = rotor + load inertia (kg·m²). Note the units: F0 is a stepping rate, not an electrical frequency — quoting "120 Hz" without specifying which is a common source of confusion.
| Illustrative System | TH (Nm) | JT (kg·m²) | F0 (steps/s) | ≈ rpm |
|---|---|---|---|---|
| NEMA 17, load inertia = rotor | 0.4 | 1.08 × 10⁻⁵ | ≈217 | ≈65 |
| NEMA 23, load = 2× rotor | 1.2 | 8.4 × 10⁻⁵ | ≈135 | ≈40 |
| NEMA 34, load = 3× rotor | 2.5 | 5.6 × 10⁻⁴ | ≈75 | ≈23 |
Assumptions stated: 1.8° motors (n=200), inertias are representative values — substitute your actual rotor and reflected load inertia. Trouble can also appear near integer multiples and fractions of F0 (F0/2, F0/3, 2F0, 3F0…), with the frequencies closest to F0 usually the worst.
For a second-order system driven at resonance, the amplitude magnification factor is:
Q = 1 / [2ζ · √(1 − ζ²)]
| Damping Ratio ζ | Classification | Peak Magnification Q | Practical Consequence |
|---|---|---|---|
| 0.02 | Very lightly damped | ≈25× | Step loss and severe noise near F0; unacceptable without mitigation |
| 0.05 | Lightly damped | ≈10× | Requires active mitigation |
| 0.10 | Moderate | ≈5× | Often workable with microstepping |
| 0.20 | Well damped | ≈2.6× | Good for general automation |
| 0.50 | Heavily damped | ≈1.15× | Small peak still present; slower response |
| ≥0.707 | No resonant peak | None | Monotonic frequency response; sluggish settling |
Values computed from Q = 1/[2ζ√(1−ζ²)]. Note that a peak exists for all ζ < 0.707 — ζ = 0.5 is not "overdamped" and does still exhibit a small peak. Critical damping is ζ = 1.
Why microstepping helps: the energy injected into the rotor per step is (4TH/n)·[1 − cos(φe)], where φe is the electrical step angle. Shrinking φe collapses this excitation energy:
| Step Mode | φe (electrical deg) | Excitation Energy vs Full Step |
|---|---|---|
| Full step | 90° | 100% |
| Half step | 45° | 29.3% |
| 1/4 | 22.5° | 7.6% |
| 1/8 | 11.25° | 1.9% |
| 1/16 | 5.625° | 0.48% |
| 1/32 | 2.8125° | 0.12% |
At 1/32 step the per-step energy is small enough to be absorbed by internal motor friction, which is why noiseless stepping down toward 0 Hz becomes possible. In practice, residual excitation from motor imperfections (§4) means resonance is reduced, not eliminated.
2.2 Mid-band instability — a different problem entirely
Do not confuse this with low-speed resonance. In the constant-torque region the drive behaves as a current source and adds no phase lag; velocity already lags torque by 90° in the mass-spring system. In the constant-power region the drive becomes a voltage source, adding another 90° — total 180°, which sustains a growing oscillation. The motor may accelerate through this band successfully but cannot run continuously in it without compensation: the oscillation typically takes 0.5–10 seconds to build to stall amplitude. The fix is electronic damping in the drive (anti-resonance algorithms), not merely microstepping or a mechanical damper.
Practical implication: if a machine runs fine during indexing but stalls after several seconds at one particular mid-range speed, suspect mid-band instability and specify a drive with active anti-resonance damping.

Figure 2: Second-order step response and resonance behaviour. Overshoot and ringing follow the classical underdamped response θ(t) = E·[1 − e−ζωnt/√(1−ζ²) · sin(ωdt + φ)], with ωd = ωn√(1−ζ²).
3. Drive Modes: Full-Step, Half-Step, and Microstepping

Figure 3: Phase A and B current waveforms across drive modes.
3.1 Wave drive (one phase on)
Only one winding is energized at a time. Power consumption is lowest, but available torque is 1/√2 ≈ 70.7% of the two-phase-on value, and the flux jumps 90 electrical degrees per step — the highest excitation and vibration of any mode. Rarely used in modern precision systems.
3.2 Full step (two phases on)
Both windings carry equal current, giving the rated holding torque. The flux still advances 90 electrical degrees per step, so torque pulsates strongly between commanded positions — the primary excitation source for low-speed resonance. Two-phase-on positions are also normally the most accurate stop positions, which is why datasheet step accuracy is specified there.
3.3 Half step
Alternating one-phase-on and two-phase-on states gives 45 electrical degrees per step and doubles resolution to 400 steps/rev. Excitation energy drops to about 29% of full step. However, torque alternates between 70.7% and 100% of TH unless the drive compensates by boosting current in the one-phase-on states — most modern half-step drives do exactly this.
3.4 Microstepping (sine/cosine drive)
IA = Ipeak · sin(φs)
IB = Ipeak · cos(φs)
The flux vector rotates in small, controlled increments. Common divisions are 1/4 through 1/256. In the limit of infinite microsteps the machine simply becomes a two-phase synchronous PM AC motor whose rotor lags the rotating field by the load angle.
| Drive Mode | Steps/rev (1.8°) | Relative Excitation Energy | Noise / Vibration | Best Use |
|---|---|---|---|---|
| Wave (1-phase-on) | 200 | High; torque only 70.7% | Highest | Legacy / low-power only |
| Full step (2-phase-on) | 200 | 100% (reference) | High | Maximum torque, coarse moves |
| Half step | 400 | 29.3% | Moderate | Simple upgrade over full step |
| Microstep 1/16 | 3,200 | 0.48% | Very low | Best all-round choice for most machines |
| Microstep 1/32 | 6,400 | 0.12% | Minimal | Quiet operation; usually sufficient |
| Microstep 1/256 | 51,200 | Negligible further gain | Minimal | Smoothness at very low speed; not higher accuracy |
The hidden ceiling on high microstep ratios: pulse rate. Shaft speed = pulse rate ÷ steps per revolution. To reach 400 rpm at 1/16 stepping requires ≈21,300 pulses/s; at 1/256 it requires ≈341,000 pulses/s — beyond most PLC pulse outputs and many controllers. High division ratios are therefore a low-speed smoothness tool, not a way to get both high resolution and high speed. Many modern drives resolve this by interpolating internally while accepting a coarser command rate.
4. What Microstepping Really Delivers: Resolution vs. Accuracy

Figure 4: Measured microstepping position ripple and holding torque ripple.
4.1 Incremental torque — the real limit on usable resolution
Each microstep commands only a tiny change in load angle, so it generates only a tiny torque. The correct expression is:
TINC = THFS · sin(90° / SDR)
TN = THFS · sin(90° · N / SDR)
SDR = microsteps per full step; N = number of microsteps taken (N ≤ SDR); THFS = full-step holding torque. Note the argument is 90°, not 180° — using 180° doubles every result and is a frequently copied error.
| SDR | Steps/rev (1.8°) | Step Angle | Incremental Torque (% of THFS) |
|---|---|---|---|
| 1 (full) | 200 | 1.8° | 100.00% |
| 2 (half) | 400 | 0.9° | 70.71% |
| 4 | 800 | 0.45° | 38.27% |
| 8 | 1,600 | 0.225° | 19.51% |
| 16 | 3,200 | 0.1125° | 9.80% |
| 32 | 6,400 | 0.05625° | 4.91% |
| 64 | 12,800 | 0.028125° | 2.45% |
| 128 | 25,600 | 0.0140625° | 1.23% |
| 256 | 51,200 | 0.00703125° | 0.61% |
Worked example — "empty resolution": a motor with THFS = 0.35 Nm at SDR = 256 produces TINC = 0.35 × sin(0.3516°) ≈ 0.00215 Nm per microstep. If bearing friction plus detent plus load torque totals 0.06 Nm, the shaft does not move until the accumulated torque exceeds that value — requiring 29 consecutive microsteps (≈0.20° of commanded motion) before any actual movement occurs.
On reversal it is worse: torque must first be unwound from its positive value through zero to a sufficient negative value. This is why ultra-high division ratios deliver smoothness, not usable incremental positioning, under load.
4.2 Why accuracy does not scale with resolution
Four motor-intrinsic error sources set the accuracy floor, and none of them shrink when you increase SDR:
- Full-step (two-phase-on) step accuracy — typically specified as ±5% of one full step, or as an absolute angle. This is the only accuracy figure most datasheets guarantee.
- Microstepping position ripple — deviation of actual microstep stop positions from theoretical, caused by non-ideal sine/cosine behaviour (varying air-gap geometry, tooth shape, magnetic hysteresis). It is a property of the motor, cyclic over each full step, and must be added to the two-phase-on accuracy to obtain total microstepping accuracy.
- Detent torque — typically 5–20% of holding torque, sometimes aiding and sometimes opposing the commanded torque.
- Hysteresis — CW and CCW approaches settle at different positions; in high-resolution systems this can be several times the nominal microstep length.
Published measurement (state the motor type when quoting it): for a 57 mm, 7.5° permanent-magnet stepper operated at 1/32 step, the measured CW position ripple was 1.65° — 22% of a full step. Individual microstep lengths ranged from about half to three times nominal. After applying a compensated sine/cosine profile, ripple fell to 0.53° ≈ 7%.
These figures belong to that specific PM motor. A 1.8° hybrid stepper has different geometry and different numbers — do not transfer percentages between motor types. Ask the supplier for measured data on the exact model if microstep accuracy matters.
Separately, microstepping holding torque ripple — variation of available torque across microstep positions — is normally 10–30% of nominal, driven by tooth geometry. This is distinct from the ±10% torque variation between different two-phase-on positions, which comes from manufacturing tolerances. Motors deliberately optimized for maximum holding torque at two-phase-on positions generally have worse sine/cosine conformity and are therefore poorer choices for microstepping.
4.3 Compensation — and its limits
Measured ripple can be corrected by shaping the current profile. For holding-torque compensation:
Inew = Iold × (TH,nominal / TH,measured)
Position-ripple compensation instead re-maps the commanded flux directions so the measured stop positions land on theory. Two cautions apply: the compensation is calibrated at a specific load and speed, so a changed load torque can make a calibrated table perform worse than no calibration at all; and at higher speeds rotor EMF alters the flux history, so low-speed calibration data no longer applies. Bidirectional applications may need separate CW and CCW profiles.
5. Stepper vs. Servo vs. BLDC

Figure 5: Illustrative torque-speed envelopes. Actual values depend on frame size, drive voltage, and control method — compare specific datasheets, not generic shapes.
| Characteristic | Stepper (open loop) | AC Servo (closed loop) | BLDC |
|---|---|---|---|
| Torque at zero speed | Full holding torque without feedback — the stepper's defining advantage | Rated torque held by the position loop | Requires position feedback and control to hold |
| Torque vs. speed | Falls steadily with speed as inductance limits current rise | Flat to base speed, then constant power | Flat to base speed; wide field-weakening range |
| Practical speed range | Best below roughly 600–1,000 rpm; usable torque above that is small | Commonly 3,000 rpm rated, 5,000–6,000 rpm peak | Very high speeds achievable |
| Position verification | None — step loss is silent unless detected externally | Encoder; following error monitored and corrected | Hall sensors give commutation, not precise position |
| Current at standstill | Full rated current unless the drive has idle-current reduction → heat | Only what the load demands | Only what the load demands |
| System cost & complexity | Lowest; simple pulse/direction interface | Highest; encoder, tuning, more commissioning effort | Moderate |
| Typical failure mode | Silent step loss on overload or resonance | Following-error fault trip (detected) | Loss of synchronism on abrupt overload |

Figure 6: Qualitative selection radar. Scores are editorial judgements for orientation only — they are not measured ratings and should not be used as specification values.
Selection rule of thumb: below roughly 1,000 rpm with substantial low-speed torque demand and a tolerance for open-loop operation, a microstepped stepper is usually the most cost-effective solution. Above that speed, or where a silent step loss would scrap product or cause a safety issue, choose closed-loop — either a closed-loop stepper (§6.3) or a servo. Decide on your torque-speed points, not on a generic crossover RPM.
6. Selection Guide & Application Recommendations
6.1 Selection checklist
| Parameter | Method | Guidance |
|---|---|---|
| Torque budget | Tload = friction + J·α (acceleration) + gravity/process | Compare against the pull-out curve at each speed, not against holding torque. Keep 30–50% margin. |
| Drive voltage | Check current-rise time vs. step period | Higher bus voltage extends usable speed; confirm drive and insulation ratings. |
| Inertia ratio | Jload / Jrotor (reflected through gearing) | Keep modest (commonly ≤10:1) for open-loop stability; also recheck the pull-in curve, which is inertia-specific. |
| Resonance check | F0 = √(n·TH/JT)/(4π); also identify the mid-band region | Avoid continuous operation near F0; specify anti-resonance damping for mid-band. |
| Microstep ratio | Balance smoothness, incremental torque, and pulse-rate ceiling | 1/16–1/32 suits most machines. Verify controller pulse capability at max speed. |
| Accuracy budget | Two-phase-on step accuracy + microstep position ripple + hysteresis + load-angle deflection + transmission error | Do not assume microstep angle equals accuracy. If the budget fails, add feedback or reduce transmission error. |
| Thermal check | Duty cycle at rated current; ambient; mounting heat-sinking | Use idle-current reduction where available; verify case temperature at thermal equilibrium. |
6.2 Application guidance
| Application | Typical Choice | Key Considerations |
|---|---|---|
| CNC axis (light milling / routing) | NEMA 23/34 stepper, 1/16 microstep | Cutting-force torque margin; surface finish benefits from smoothness; consider closed-loop to prevent silent scrap |
| 3D printer | NEMA 17 stepper, 1/16–1/32 | Low acoustic noise; driver current and idle reduction; belt/pulley error often dominates motor error |
| Robot joint / indexing axis | Closed-loop stepper or AC servo | Feedback essential; gravity load and reversal hysteresis; verify holding behaviour at power-off (brake may be needed) |
| High-speed packaging | AC servo | High rpm and rapid accel/decel; stepper torque collapse rules it out |
| Precision inspection stage | Stepper + linear encoder, or linear servo | Measure at the load, not the motor; compensate leadscrew error; unidirectional approach to defeat hysteresis |
| Valve / dosing actuator | Stepper with gearbox, 1/8–1/16 | Gearing multiplies incremental torque, mitigating empty resolution; check backlash on reversal |
A note on pump drives: positive-displacement high-pressure pumps are constant-torque, continuous-duty loads, normally driven by AC induction motors, not steppers — the requirement is sustained shaft power at fixed speed, not indexed positioning. TITECHO's hollow-shaft range addresses exactly this: a TEFC squirrel-cage induction motor with a through-shaft design that mounts directly to Hawk / Interpump / AR style pump heads, eliminating the coupling and bracket. Published range: centre heights 100/112/132/160 mm, 2.2–18.5 kW, 380 V (customizable), 50/60 Hz, Class F insulation at 80 K rise, IP54, IEC mounting dimensions. Removing the coupling also removes coupling compliance and misalignment — a real reliability benefit, but one that belongs to induction-motor engineering rather than to stepper microstepping.
6.3 Closed-loop stepper systems — measured benefits
Closing the position loop around an encoder on a stepper motor changes its behaviour fundamentally: the drive supplies only the current the load actually needs, monitors following error, and can apply active damping. Published laboratory comparisons of equally sized open-loop and closed-loop stepper systems performing identical move profiles with identical inertial loads report:
| Metric | Open Loop | Closed Loop | Change |
|---|---|---|---|
| Average power (same work) | 43.8 W | 14.2 W | ≈68% lower |
| Case temperature at equilibrium | 76.0°C | 36.9°C | Less than half |
| Max acceleration | 1,000 rev/s² | 2,000 rev/s² | 2× |
| Top speed before position error | 600 rpm | 1,200 rpm | 2× |
| Move time (same profile target) | 110 ms | 60 ms | ≈45% shorter |
Source: Applied Motion Products StepSERVO laboratory comparison, reported in Motion Control Tips. These are results for one specific motor pair and duty profile — treat them as demonstrated potential, not a universal guarantee. Your savings depend on how much torque margin your open-loop system currently wastes.
Mechanically, closed-loop NEMA-frame steppers generally retain the same faceplate, pilot, and bolt pattern, so brackets usually carry over; the higher available torque may mean a larger shaft diameter, requiring a coupling change. The main added cost is the feedback device, which has become steadily cheaper.
Conclusion
Stepper systems earn their place through low-speed torque density, simplicity, and cost — provided the designer respects their real limits. The five decisions that determine success:
- Size against the pull-out curve at speed, with margin — never against holding torque alone, and remember the pull-in curve is inertia-specific.
- Treat the two resonance mechanisms separately: microstepping and inertia management for low-speed mechanical resonance; active drive damping for mid-band instability.
- Use microstepping for smoothness, not accuracy. 1/16–1/32 is the practical sweet spot; check incremental torque against friction and check the controller's pulse-rate ceiling.
- Build an honest accuracy budget from step tolerance, position ripple, hysteresis, load angle, and transmission error — and add feedback if the budget fails.
- Consider closed-loop where step loss is unacceptable or where wasted standstill current is a thermal problem; published tests show large gains in power, heat, and throughput.
Match the technology to the duty: steppers for indexed low-speed positioning, servos for dynamic high-speed motion, and — for continuous constant-torque loads such as pumps — conventional AC induction motors, which remain the correct engineering answer for that class of work.
Technical references: FAULHABER Application Note AN015 "Microstepping" (incremental torque TINC=THFS·sin(90°/µPFS), accuracy-vs-resolution discussion, detent torque 5–20%); Industrial Circuits / Ericsson Application Note "Microstepping" (torque equation, electrical/mechanical angle conversion, natural frequency F0, excitation energy, position and holding-torque ripple measurements, compensation method); GeckoDrive "Mid-Band Instability" (phase-lag mechanism and electronic damping); Applied Motion Products StepSERVO lab comparison via Motion Control Tips (power, temperature, acceleration, speed, move-time data); Oriental Motor speed-torque curve documentation (pull-in / pull-out definitions); Analog Devices and PMD Corp microstepping articles (resolution ≠ accuracy). Worked examples are illustrative calculations from stated assumptions, not guaranteed performance values. Confirm all parameters against the specific motor and drive datasheets.
Need the Right Motor for Your Drive Application?
TITECHO manufactures three-phase and single-phase induction motors from 0.12–375 kW in IE2/IE3/IE4 classes with 100% copper stator windings and VPI insulation under an ISO 9001:2015 quality system — including the hollow-shaft TEFC range (2.2–18.5 kW, centre heights 100–160 mm, Class F / 80 K, IP54) designed for direct mounting to high-pressure pump heads. Tell us the load type, duty cycle, speed, supply, and mounting constraints, and we will confirm the suitable configuration with its data sheet and outline drawing.
✉️ info@cntecho.com • 📞 +86 13305761511 • 24-Hour Response • 1-Year Warranty
Final suitability, documentation scope, and service terms are confirmed in the quotation and order documentation.
TITECHO – TECHO ELECTRICAL & MECHANICAL (TAIZHOU) CO., LTD
Taizhou City, Zhejiang, China | www.cntecho.com
Related News
LATEST
INFORMATION
Get the latest product information of the company
NAVIGATION
PRODUCTS
CONTACT US
Telephone: +86 13305761511
Email: info@cntecho.com
Add: RM. 602, BUILDING B, NO.1551 SHUANGSHUI RD., LUQIAO DIST., TAIZHOU CITY, ZHEJIANG, CHINA
Copyright © 2026 TECHO ELECTRICAL & MECHANICAL (TAIZHOU) CO., LTD. All Rights Reserved.