Servo motor

A servo motor is a motor operated inside a closed feedback loop: a controller sends a target position, speed, or torque, a sensor, typically a rotary encoder, reports what the motor actually did, and the controller continuously corrects the difference between the two.^[1] The name comes from the Latin servus ("slave" or "servant"), carried into French as servo-moteur by the engineer Joseph Farcot, who used the term in an 1868 patent and a later book to describe the hydraulic steering engines he built so a single sailor could turn a warship's multi-ton rudder with a small wheel.^[2][3] A servo motor "serves" a command rather than merely spinning when power is applied, and that distinction, feedback versus none, is what separates it from an ordinary motor and from an open-loop stepper motor. Servo motors occupy nearly every powered joint of an industrial robot arm and most joints of a humanoid robot, because holding a commanded angle and resisting an external push both require the motor's controller to know, continuously, exactly where the joint is.

In brief: an ordinary motor just spins when power is applied. A servo motor also carries a built-in sensor that reports its exact position, so a controller can tell it to go to a specific angle and stay there, check whether it arrived, and correct course if something pushes it off target, often thousands of times per second.

How a servo motor works

A complete servo system has three parts: the motor, which produces rotation; a feedback sensor, which measures the result; and a controller, usually called a servo drive or servo amplifier, which compares the two and adjusts the electrical power sent to the motor.^[1][4] Strictly, "servo motor" refers only to the motor and its attached sensor, while the drive electronics are a separate component, though the two are rarely sold or used apart.^[4]

The control cycle repeats continuously. First, a higher-level controller, such as a robot's motion planner, sends the servo drive a target, for instance an instruction to move a joint to a specific angle. Second, the encoder reports the joint's actual position, often many thousands of times per second. Third, the drive calculates the error, the gap between target and actual position, and adjusts the current flowing into the motor's windings to close that gap, then repeats the cycle.^[5]

In practice this runs as three nested control loops rather than one. An inner current, or torque, loop regulates how much current flows into the motor windings, which sets the motor's output torque directly. A velocity loop wraps around the current loop, comparing commanded speed with measured speed and adjusting the torque command to correct any difference. An outer position loop wraps around both, comparing the commanded position with the encoder's measured position and issuing the velocity commands needed to close that gap.^[5][6] Running all three loops inside a single drive, rather than splitting them across separate devices, shortens the delay between measuring an error and correcting it, part of why modern integrated servo drives behave more stiffly and precisely than older, slower control architectures.^[6]

PID control and tuning

Each of these loops is usually closed with a proportional-integral-derivative (PID) controller, a standard three-term feedback algorithm.^[7] The proportional term pushes back on the error in rough proportion to its size, acting somewhat like a spring: a larger gap produces a stronger correction. The integral term accumulates error over time, which lets the loop eliminate the small steady-state offset that a proportional term alone cannot close, for example the sag caused by gravity pulling on a loaded arm. The derivative term reacts to how quickly the error is changing, which damps overshoot and oscillation.^[8]

Tuning a servo loop means setting the gain of each term, commonly written Kp, Ki, and Kd, high enough for a fast and accurate response without causing the joint to oscillate or squeal. A common starting method raises the proportional gain alone until the motor just begins to oscillate, then backs it off and adds derivative gain to damp the remaining overshoot.^[8] Badly tuned gains produce recognizable symptoms: too little gain and the joint feels sluggish and visibly lags behind its command, a condition called following error; too much gain and the joint vibrates or hunts around the target.

Because manual tuning takes time and an experienced engineer, several manufacturers have added automatic tuning to their servo drives. Panasonic, for example, says the AI-assisted autotuning feature on its Minas A7 servo system, launched in Japan in January 2024, cuts manual tuning effort by more than 90 percent and lowers positioning settling time by 45 percent compared with expert manual tuning, though those figures are the company's own and have not been independently audited.^[9]

The role of the encoder

The feedback sensor is what makes a servo a servo. Most servo motors use a rotary encoder mounted on or built into the motor shaft, an optical, magnetic, or, less commonly today, capacitive device that converts shaft rotation into a digital signal the drive can read.^[10] Encoders split into two memory types. Incremental encoders only count movement while powered and lose their position reference the moment power is cut, so the system needs a homing routine after every restart. Absolute encoders report a unique code for every shaft position and therefore know their exact position immediately at power-up, a property that matters for any joint that must know its own angle before it moves, such as a leg recovering from a stumble.^[11]

Some servo motors use a resolver instead of an encoder: an older, fully analog, electromagnetic sensor with no onboard electronics that tolerates heat, shock, and contamination better than most encoders, at the cost of lower resolution. That ruggedness is why resolvers persist in harsh environments, such as steel mills and some aerospace and military drives, even as encoders have displaced them almost everywhere else.^[12][13] Low-cost hobby servos typically substitute a simple potentiometer, a variable resistor whose output voltage tracks shaft angle, for a true encoder.

Servo motor types

Servo motor terminology names both the underlying motor hardware and the electronics that drive it, and manufacturers do not always use the terms consistently, so it helps to separate the three labels most often used in industrial catalogs.

DC servo motors

The oldest and simplest type pairs a small brushed motor with a potentiometer or encoder for feedback. A pair of carbon brushes physically contacts a rotating commutator to reverse current in the windings and keep the rotor turning; the brushes wear down, spark, and generate electrical noise as they do, so brushed DC servos are now mostly confined to low-cost, low-cycle applications such as hobby electronics and older industrial equipment.^[14]

Brushless DC (BLDC) servo motors

A brushless DC (BLDC) servo motor replaces the mechanical brushes with electronic commutation: Hall-effect sensors or the encoder itself report rotor position, and the drive electronically switches current through the stator windings in the correct sequence, typically a six-step pattern known as trapezoidal commutation.^[14][15] Removing the brushes eliminates their wear and electrical noise, so BLDC servos last longer, run quieter, and need less maintenance, a major reason they dominate robot joints and dexterous hand actuators.^[15]

AC servo motors

An AC servo motor typically uses the same permanent-magnet, brushless hardware as a BLDC servo motor, but the drive commutates it with a smoothly varying sinusoidal current computed through field-oriented control, rather than switching current in six discrete steps.^[14][16] That smoother commutation produces less torque ripple and quieter, more precise motion, which is why AC servo motors are the default choice in CNC machine tools and industrial robot arms. In practice, "AC servo" and "brushless DC servo" often describe the same permanent-magnet synchronous motor hardware driven by different commutation software rather than two fundamentally different kinds of motor.^[14][16]

Industrial versus hobby RC servos

Hobby and radio-control (RC) servos package a small motor, gearbox, potentiometer, and control circuit inside one compact housing with a three-wire connector, controlled by a pulse-width-modulated (PWM) signal repeated at 50 Hz, with pulse width between roughly 1 and 2 milliseconds; a 1.5-millisecond pulse centers the shaft near its middle position.^[17] Cheaper analog hobby servos correct position around 50 times a second, while digital hobby servos use an onboard microprocessor to correct roughly 300 times a second, producing faster, smoother response at a higher price.^[18] Removing a hobby servo's internal mechanical stop and disconnecting its potentiometer converts it into an open-loop, continuously rotating gear motor, a common hobbyist modification that illustrates the underlying point: the closed feedback loop, not the gearbox, is what makes a servo a servo.^[19]

Industrial servo motors are typically sold as a bare motor paired with a separate, fully configurable servo drive that communicates over an industrial network such as EtherCAT or CANopen, and run from a higher DC bus voltage derived from three-phase AC power. They deliver far higher continuous torque and duty cycle than hobby servos, support fully adjustable PID tuning rather than a fixed factory response, and are commonly rated for tens of thousands of operating hours.

Scroll sideways for more →

Servo motors versus stepper motors

A stepper motor is the open-loop alternative most often compared with a servo motor, and the comparison is useful for understanding why the closed loop matters. A stepper's rotor has many magnetic poles, often 50 to 100, that act like detents; each electrical pulse advances the shaft by one fixed increment, so a controller can infer position simply by counting the pulses it has sent, without ever measuring the actual shaft position.^[20] That approach is cheap (no encoder, no tuning) and provides inherent holding torque whenever power stays on, and it works well at low speed under a light, predictable load. It breaks down under a sudden load spike or at high speed, when the rotor can silently "lose steps," meaning the controller's internal count no longer matches the shaft's real position; unlike a servo, a stepper has no way to detect or correct that drift on its own.^[21]

A servo motor closes that gap with its encoder: the drive always knows the true shaft position and can correct for a missed step, an external push, or a changing load in real time. That is why open-loop steppers remain common in 3D printers and low-cost CNC machines but are almost never used at a load-bearing robot joint, where an undetected position error could mean a collapsed leg or a crushed workpiece.^[22] Hybrid "closed-loop stepper" motors add an encoder to stepper hardware to recover some of this safety margin, but they still fall short of a true servo's dynamic performance.^[21] The tradeoff for a servo's precision is cost and complexity: a servo system needs an encoder, a more capable drive, and, usually, tuning effort that a stepper does not.

Scroll sideways for more →

Servo motors in robot joints and humanoid robots

A robot joint has to do two things a plain motor cannot: hold a commanded angle against an external force, and know precisely what angle it is holding. Both require closed-loop feedback, which is why a servo motor, rather than an open-loop motor, sits inside nearly every robot actuator, the combined assembly of motor, feedback sensor, and reduction stage (and often a torque or force sensor) that makes up a single powered joint. A humanoid robot, which may have several dozen degrees of freedom across its body, uses a comparable number of individually tuned servo motors, one per powered joint.

A servo drive can run a joint in one of three control modes. In position control, the drive moves the joint to a commanded angle and holds it there, correcting for any disturbance, the mode used whenever a robot needs to reach and hold a precise pose. In velocity control, the drive regulates rotational speed regardless of load, useful for continuously driven axes such as wheels. In torque control, the drive regulates output torque directly through motor current, letting the joint push with a commanded force while its resulting speed is left free, closer to how a compliant, force-aware arm or hand needs to behave around people or fragile objects.^[23] The proportional gain of the position loop also sets how stiff a joint feels: a high gain snaps the joint back hard against any push, while a lower gain lets it flex, a tuning tradeoff between precision and compliance.^[23]

Most humanoid robot joints today use a brushless DC or AC permanent-magnet servo motor, often built in an outrunner configuration, paired with a reduction stage such as a harmonic drive or a planetary gear train to trade the motor's naturally high speed and low torque for the low speed and high torque a joint needs. A minority of designs instead use a low-ratio, highly backdrivable quasi-direct drive actuator, which sacrifices some torque density so the motor's own current can serve as a rough torque sensor and so the joint can be pushed or pulled by an external force without damage, properties useful for a robot that learns by trial and error or works in close contact with people.^[24]

Humanoid designers frequently add a second encoder to important joints: one at the motor shaft, mainly to commutate the motor and close the fast current loop, and a second at the joint's output side, after the gearbox, to measure the limb's actual angle.^[25] The two readings normally agree, but they diverge under gearbox backlash, cable stretch, or structural flex, and that divergence is itself useful: it approximates how heavily the joint is loaded, and it corrects for the sim-to-real gap that appears when a joint trained in simulation behaves slightly differently on physical hardware.^[25][26]

Dexterous hand actuators partly break from the servo-plus-gearbox pattern. Because fingers need to be light, fast, and gentle rather than strong, some hand and finger joints use small servo motors built around a coreless motor instead of an iron-core design, trading peak torque for lower rotor inertia and smoother, quieter, cogging-free motion, properties that matter more for a fingertip than raw strength does.^[24][27]

Manufacturers

The servo motor supply chain splits roughly into two groups: companies that primarily sell branded servo motor and drive product lines to machine builders, and companies that primarily build finished robots or machine tools and either manufacture their own servo motors or integrate motors developed with an outside specialist. The eight companies most often named as servo motor sources for industrial robotics fall across both groups.

FANUC, the Japanese CNC and robotics group that became independent from Fujitsu in 1972, is among the most vertically integrated companies in the industry. It began producing DC servo motors under license from Gettys Manufacturing in 1974, developed its own AC servo motor in 1982, and today designs and builds the servo motors used in both its CNC controls and its industrial robot arms in-house, alongside the controllers themselves.^[28]

Yaskawa Electric, founded in Japan in 1915 to build electric motors and generators, has manufactured its own servo motors since its small-volume Sigma AC servo line launched in 1992. The current Sigma-7 and newer Sigma-X families are sold as standalone industrial servo products and also used inside Yaskawa's own Motoman-brand industrial robots, a lineage that traces to the Motoman-L10, Japan's first fully electric industrial robot, in 1977.^[29][30]

Panasonic, founded in Osaka in 1918 as a maker of lamp sockets and renamed from Matsushita Electric to its present name in 2008, sells its own Minas brand of AC servo motors, including the A6 family, which carries a 23-bit encoder, and the newer A7 family, which Panasonic says adds AI-assisted automatic tuning.^[9][31]

Siemens, founded in Berlin in 1847 as a telegraph-equipment maker, sells its own Simotics brand of synchronous servo motors, including the S-1FK7 and S-1FT7 families, which integrate with Siemens's broader factory-automation drives and software. Siemens is not itself a robot-arm manufacturer; its servo motors more commonly appear inside machine tools and packaging equipment built by other companies.^[32]

Bosch Rexroth, the drive-and-automation subsidiary of the German Bosch Group, traces its roots to an 18th-century forge before entering hydraulics in the 1950s and later electric drives; it took its current form in 2001 when Bosch's own automation-technology unit merged with Mannesmann Rexroth. It sells its own IndraDyn line of synchronous AC servo motors alongside IndraDrive servo drives, used across general factory automation and some robotics applications.^[33][34]

ABB traces to the 1988 merger of Sweden's ASEA, founded in 1883, and Switzerland's BBC Brown Boveri, founded in 1891. ASEA had already built the IRB 6, the first all-electric, microprocessor-controlled commercial industrial robot, in 1974. ABB now sells its own eSM, DSM, and HDS branded servo motors through its Motion division to machine-building customers.^[35][36]

KUKA, the German robot maker founded in Augsburg in 1898, whose name is an acronym for Keller und Knappich Augsburg, built the FAMULUS, the world's first six-axis, electrically driven industrial robot, in 1973. Unlike the other companies here, KUKA functions primarily as an integrator rather than an in-house servo motor manufacturer: it works with outside motor specialists, for instance co-engineering the AKM synchronous servo motors used in its compact KR Agilus robots together with Kollmorgen, rather than designing that motor line itself.^[37][38]

Robotis, founded in South Korea in 1999, built its business around the Dynamixel line of "smart" servo motors, which package a motor, gearhead, controller, and network interface into a single addressable module rather than a bare motor sold for a separate drive. Dynamixel servos are widely used in robotics research and education; in 2026 Robotis released the AI Sapiens K0, a fully open-source humanoid robot with 23 degrees of freedom that uses a new Dynamixel-Q quasi-direct-drive actuator, illustrating the industry's broader shift toward lower-reduction, more backdrivable servo actuators.^[39][40]

Scroll sideways for more →

Market-size estimates for the global servo motor and drive industry vary by scope and methodology: Mordor Intelligence puts the market at roughly $14.6 billion in 2025, growing to about $15.4 billion in 2026, while SNS Insider estimates a larger $19.1 billion for 2025. Both firms forecast continued growth into the early 2030s, driven mainly by factory automation and industrial robotics demand, with the Asia-Pacific region accounting for the largest regional share.^[41][42]

See also

• Brushless DC motor
• Stepper motor
• Actuator
• Rotary encoder
• Coreless motor
• Harmonic drive
• Quasi-direct drive
• Humanoid robot

References

1. Electrical4U, "Servo Motor: Definition, Working Principle, and Applications." https://www.electrical4u.com/what-is-servo-motor/ ↩
2. Wikipedia, "Joseph Farcot." https://en.wikipedia.org/wiki/Joseph_Farcot ↩
3. Otto Mayr, "Origins of the Servo-Motor" (History), IEEE Xplore. https://ieeexplore.ieee.org/document/485765/ ↩
4. Wikipedia, "Servomotor." https://en.wikipedia.org/wiki/Servomotor ↩
5. Motion Control Tips, "What are servo drive loops?" https://www.motioncontroltips.com/servo-drive-loops/ ↩
6. Motion Control Tips, "FAQ: What are servo motor current, velocity and position loops and bandwidths?" https://www.motioncontroltips.com/faq-servo-motor-current-velocity-position-loops-bandwidths/ ↩
7. PMD Corp, "How To Tune a PID Loop: A PMD Guide for Servo Motor Applications." https://www.pmdcorp.com/resources/type/articles/get/how-to-tune-a-pid-loop ↩
8. Motion Control Tips, "A simple PID loop tuning technique: Part 1 of 2." https://www.motioncontroltips.com/a-simple-pid-loop-tuning-technique-part-1-of-2/ ↩
9. Panasonic Newsroom, "Panasonic Industry Commercializes the Industry's First AI-equipped Servo System, Reducing Human Operation Time by 90%." September 13, 2023. https://news.panasonic.com/global/press/en230913-3 ↩
10. ASPINA, "What is a servo motor?" https://www.aspina-group.com/en/learning-zone/columns/what-is/004/ ↩
11. Celera Motion, "Absolute Encoders vs Incremental Encoders: A Guide." https://www.celeramotion.com/inductive-sensors/support/technical-papers/absolute-encoders-vs-incremental-encoders/ ↩
12. Encoder.com, "Resolvers vs. Encoders." https://www.encoder.com/encoders-vs-resolvers ↩
13. Kollmorgen, "Feedback Devices: Exploring Hall-Effect Sensors and Resolvers." https://www.kollmorgen.com/en-us/blogs/feedback-devices-exploring-hall-effect-sensors-and-resolvers ↩
14. Kollmorgen, "What are the differences between DC, BLDC and AC servo motors?" https://www.kollmorgen.com/en-us/blogs/what-are-differences-between-dc-bldc-and-ac-servo-motors ↩
15. Oriental Motor, "Brushless DC Motors (BLDC Motors) vs. Servo Motors vs. Inverters." https://www.orientalmotor.com/brushless-dc-motors-gear-motors/technology/brushless-dc-motors-servo-motors-inverter.html ↩
16. Electrical Blog, "AC vs DC vs Servo Motors: Which Should You Use?" https://www.electrical-blog.com/ac-vs-dc-vs-servo-motors-which-should-you-use/ ↩
17. SparkFun Learn, "Hobby Servo Tutorial." https://learn.sparkfun.com/tutorials/hobby-servo-tutorial/servo-motor-background ↩
18. Servo Magazine, "Analog Servos for Robotics," July 2015. https://www.servomagazine.com/magazine/article/july2015_Ostendorff ↩
19. Adafruit Learning System, "Modifying Servos for Continuous Rotation." https://learn.adafruit.com/modifying-servos-for-continuous-rotation/overview ↩
20. Wevolver, "Stepper vs Servo Motors: Mastering Motor Selection for Precision Engineering." https://www.wevolver.com/article/stepper-vs-servo-motors-a-comprehensive-comparison-for-your-next-project ↩
21. Motion Control Tips, "What's the difference between servo and closed-loop stepper motors?" https://www.motioncontroltips.com/faq-whats-the-difference-between-servo-and-closed-loop-stepper-motors/ ↩
22. Oriental Motor, "The Choice Between Servo Motors and Stepper Motors." https://blog.orientalmotor.com/the-choice-between-servo-motors-and-stepper-motors ↩
23. SVRC, "Robot Control Modes Explained: Position, Velocity, Torque, and Impedance." https://www.roboticscenter.ai/learn/robotics-library/robot-control-modes-explained ↩
24. RoboticsTomorrow, "What It Takes to Make Humanoid Robots Move Like Humans: The Engineering Behind Joints, Hands, and Precision Control," February 2026. https://www.roboticstomorrow.com/article/2026/02/what-it-takes-to-make-humanoid-robots-move-like-humans-the-engineering-behind-joints-hands-and-precision-control/26152 ↩
25. Stabilarity Hub, "Proprioception and Internal State Estimation: Joint Encoders, Torque Sensing, and Body Schema for Humanoid Robots." https://hub.stabilarity.com/proprioception-and-internal-state-estimation-joint-encoders-torque-sensing-and-body-schema-for-humanoid-robots/ ↩
26. Heidenhain, "Why dual encoders are the right choice for robotics design." https://www.heidenhain.us/resources-and-news/4-reasons-dual-encoders-are-the-right-choice-for-robotics-design/ ↩
27. Firgelli, "Cored vs Coreless DC Motors: What Actually Changes?" https://www.firgelliauto.com/blogs/robotics/cored-vs-coreless-dc-motors-explained ↩
28. FANUC Corporation, "FANUC's History." https://www.fanuc.co.jp/en/profile/history/ ↩
29. Yaskawa, "History." https://www.yaskawa-global.com/company/profile/history ↩
30. Yaskawa, "Sigma-7 Servo Products." https://www.yaskawa.com/products/motion/sigma-7-servo-products ↩
31. Panasonic Industry, "MINAS A6 Family Servo Motor." https://na.industrial.panasonic.com/products/industrial-automation/motion-control/lineup/ac-servo-motors/series/3456 ↩
32. Siemens, "SIMOTICS S-1FK7 Servomotors." https://www.siemens.com/en-us/products/simotics/s-1fk7/ ↩
33. Bosch, "History of Bosch Rexroth." https://www.bosch.com/stories/the-history-of-bosch-rexroth/ ↩
34. Bosch Rexroth, "Synchronous Servo Motors (Indradyn S)." https://boschrexroth.africa/en/synchronous-servo-motors-indradyn-s ↩
35. ABB, "140 years of ASEA." https://new.abb.com/news/detail/106125/140-years-of-asea ↩
36. ABB, "Servo products." https://www.abb.com/global/en/areas/motion/drives/low-voltage-ac-drives/servo-products ↩
37. Wikipedia, "KUKA." https://en.wikipedia.org/wiki/KUKA ↩
38. Kollmorgen, "Kuka and Kollmorgen co-engineer optimised motors for compact robots." https://www.kollmorgen.com/en-us/service-and-support/knowledge-center/success-stories/even-higher-power-density ↩
39. Robotis, "About Us." https://www.robotis.us/about-us/ ↩
40. Interesting Engineering, "World's first full-stack humanoid robot open-sourced for developers," 2026. https://interestingengineering.com/ai-robotics/worlds-first-full-stack-humanoid-robot-open-sourced ↩
41. Mordor Intelligence, "Servo Motor Market: Size, Share and Growth." https://www.mordorintelligence.com/industry-reports/global-servo-motor-market ↩
42. SNS Insider via GlobeNewswire, "Servo Motors and Drives Market Report 2026-2035," June 30, 2026. https://www.globenewswire.com/news-release/2026/06/30/3319419/0/en/servo-motors-and-drives-market-report-2026-2035-rising-at-6-42-cagr-to-reach-usd-35-65-billion-by-2035-sns-insider.html ↩