A hydraulic servo is a closed-loop fluid-power system that uses a servo valve, a hydraulic actuator, and a position or force feedback transducer to drive a load to a commanded position with high precision. It solves the problem of moving heavy loads — tonnes, not kilograms — with the response time and accuracy of an electric servo. The valve modulates oil flow in proportion to an electrical command signal, and the feedback element trims out error in real time. Outcome: aircraft control surfaces, steel mill rolling stands, and 4000 tonne presses all hold position to within ±0.05 mm under load.
Hydraulic Servo Interactive Calculator
Vary load, pressure, safety factor, and precision to size the servo cylinder bore and see the closed-loop positioning band.
Equation Used
This calculator sizes the ideal piston area and bore needed for a hydraulic servo cylinder to support a specified load at the available supply pressure. It also shows the total closed-loop positioning band from the stated +/- precision.
- Ideal hydraulic conversion with full supply pressure available at the piston.
- Single effective piston area is used for sizing.
- Friction, leakage, valve pressure drop, acceleration, and structural deflection are ignored.
- Metric tonne uses g = 9.80665 m/s^2.
- Precision input is the +/- half-band; total band is twice this value.
How the Hydraulic Servo Works
A hydraulic servo takes a low-power electrical command — usually a few milliamps to a torque motor — and turns it into a large, precisely controlled mechanical motion. The command pushes a flapper or jet pipe inside a servo valve, which biases pressure across the ends of a spool. The spool shifts a fraction of a millimetre and meters high-pressure oil (typically 200 bar, sometimes 350 bar on aerospace systems) into one side of a cylinder while venting the other side to tank. The cylinder moves, a feedback transducer — usually an LVDT for position or a load cell for force — reports the actual position back to the controller, and the controller continuously updates the command to drive error to zero. That's the closed-loop position control part, and it's what separates a hydraulic servo from a simple proportional valve.
The geometry inside the servo valve is brutal. The spool lands and the sleeve ports are matched to within 1 to 2 µm of radial clearance. Oil filtration must hold at ISO 4406 16/14/11 or cleaner — let a 15 µm particle through and it lodges in the spool clearance, the valve sticks, and the actuator either drifts or buzzes. That's the single most common failure mode in the field. The second is null bias drift: as the torque motor warms, its magnetic gap shifts and the valve no longer centres at zero command. You see this as the cylinder slowly creeping when the controller thinks it's commanding hold. Good systems trim null bias electrically every startup.
The feedback transducer choice drives accuracy. An LVDT mounted on the rod gives you ±0.05 mm repeatability easily; a magnetostrictive sensor like a Temposonics R-series gives ±0.01 mm over a 1 m stroke. Skip the feedback or run open-loop with just a flapper nozzle servo valve and you get a fast, powerful, but blind actuator — fine for a rough-acting press, useless for a flight surface.
Key Components
- Servo Valve (two-stage flapper-nozzle or jet-pipe): Converts a 4 to 50 mA electrical command into precise spool displacement. The first stage is a torque motor driving a flapper between two nozzles; the resulting pressure differential drives the second-stage spool. Spool-to-sleeve radial clearance is held to 1-2 µm and matched lap fits are common. Moog Type 30 and Parker BD15 are typical examples.
- Hydraulic Cylinder or Rotary Actuator: Converts metered oil flow into linear or rotary motion. Bore sizes from 25 mm up to 500 mm cover loads from 1 kN to over 4 MN. Rod surface finish must be Ra 0.2 µm or better — coarser finishes shred the rod seals within hours under servo duty cycles.
- Position Feedback Transducer: Reports actuator position back to the controller. LVDTs cover short strokes to ±0.05 mm; magnetostrictive sensors handle long strokes to ±0.01 mm. Mounting must be rigid — any lash between the rod and the sensor body shows up as oscillation in the closed loop.
- Hydraulic Power Pack: Provides regulated supply pressure, usually 200 bar with ±2% ripple. A pressure-compensated piston pump and accumulator together hold pressure stiff during load transients. Filtration at 3 µm absolute (β3 ≥ 200) on the pressure line is non-negotiable for servo-valve life.
- Servo Controller: Closes the loop in software or analogue hardware, usually at 1 to 5 kHz update rate. Implements PID with feed-forward, dither injection (50-400 Hz at ~5% command) to prevent spool stiction, and electrical null trim to compensate torque-motor drift.
- Filtration and Conditioning: Pressure-line filter at 3 µm, return-line filter at 10 µm, off-line kidney loop running continuously when possible. Oil must hold ISO 4406 16/14/11 or cleaner. Skip this and servo-valve life drops from 20,000 hours to under 2,000.
Who Uses the Hydraulic Servo
Hydraulic servos show up wherever you need to move a heavy load fast and land it accurately. Electric servos win below roughly 10 kN; above that, hydraulics dominate because power density per kilogram of actuator is unmatched. The mechanism is designed this way because oil is nearly incompressible and pumps can sustain high pressure indefinitely — you get stiffness and continuous force in a package an electric motor of equivalent rating cannot match. When tolerances drift or contamination creeps in, the system loses bandwidth first, then precision, then finally the valve sticks. That progression is the diagnostic signature.
- Aerospace: Primary flight control actuators on the Boeing 777°— the elevator, rudder, and aileron actuators are electrohydraulic servoactuators running at 3000 psi (later 5000 psi on the 787).
- Steel and Metals: Automatic gauge control on hot strip mills — the SMS Group and Danieli rolling stands use hydraulic servos to hold strip thickness to ±10 µm at 20 m/s line speeds.
- Plastics: Injection moulding clamp and injection units on Engel Duo and KraussMaffei MX presses — servo valves control injection velocity profiles to ±1% across the shot.
- Materials Testing: MTS 810 and Instron 8800 servohydraulic test frames running fatigue cycles at 30 Hz on automotive suspension components and aerospace coupons.
- Heavy Industry: Continuous casting mould oscillation drives at ArcelorMittal slab casters — hydraulic servos generate sinusoidal mould motion at 100-300 cycles per minute with ±0.1 mm amplitude accuracy.
- Earthquake Simulation: Six-degree-of-freedom shake tables at the E-Defense facility in Miki, Japan — 24 hydraulic servoactuators reproduce real seismic ground motion on full-scale building specimens.
- Robotics: Boston Dynamics Atlas (early hydraulic versions) used custom hydraulic servoactuators at the hips and knees for 1500 W peak power per joint.
The Formula Behind the Hydraulic Servo
The core sizing equation for a hydraulic servo is the no-load flow gain — how fast the cylinder moves for a given valve flow rating and supply pressure. At the low end of typical operating supply pressures (around 100 bar) you get half the flow you'd see at 200 bar, so the actuator runs slower and stiffer. At nominal 200 bar most industrial servo valves are rated. Push to 350 bar (aerospace territory) and flow goes up by √1.75 ≈ 1.32× over the 200 bar figure — but spool erosion accelerates and you need harder valve materials. The sweet spot for industrial work sits at 210 bar where standard Moog and Parker valves are characterised on the datasheet.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| QL | Load flow through the valve to the actuator | L/min | gpm |
| QR | Rated flow at rated pressure drop and rated current (datasheet value) | L/min | gpm |
| ΔPL | Actual pressure drop across the valve at the operating point | bar | psi |
| ΔPR | Rated valve pressure drop (usually 70 bar / 1000 psi total, 35 bar per land) | bar | psi |
| i | Actual command current to torque motor | mA | mA |
| iR | Rated command current (typically ±10, ±15, or ±40 mA) | mA | mA |
Worked Example: Hydraulic Servo in a fatigue test rig actuator
A wind-turbine blade test lab in Aalborg is sizing the cylinder velocity for a hydraulic servo on a 5 m stroke fatigue rig that flexes a 60 m blade root at 0.5 Hz. The valve is a Moog G761-3005 rated at 38 L/min at 70 bar drop and ±10 mA. Supply pressure is 210 bar. Cylinder bore is 100 mm, rod 70 mm. The lab needs to know peak rod velocity at full command and how it changes if they ever drop supply pressure to save pump load.
Given
- QR = 38 L/min
- ΔPR = 70 bar (total, 35 bar per land)
- PS = 210 bar
- Load pressure drop assumed = 70 bar (matched-load, max power transfer)
- Cylinder bore = 100 mm
- Cylinder rod = 70 mm
- i / iR = 1.0 (full command)
Solution
Step 1 — at nominal 210 bar supply with matched-load ΔPL = 70 bar across the valve, the load flow at full command equals the rated flow:
Step 2 — convert that to rod velocity using the cap-end annulus area on extension. Cap-end area Acap = π × (0.050)2 = 7.85 × 10-3 m²:
That's 81 mm/s peak rod velocity. For a 0.5 Hz sinusoid at ±200 mm amplitude that's exactly what you need — peak velocity of a sine is 2π × f × A = 2π × 0.5 × 0.2 = 0.628 m/s... wait, that exceeds 81 mm/s. The rig will be valve-flow limited, not amplitude limited. The lab will either need a larger valve (Moog G761-3007 at 63 L/min) or accept reduced amplitude.
Step 3 — at the low end of supply pressure, drop PS to 100 bar to save pump energy. ΔPL available drops to roughly 33 bar at matched load:
You lose 32% of velocity. The blade fatigue test would slow noticeably and waveform distortion at the velocity peaks becomes visible on the LVDT trace. At the high end, push supply to 280 bar and ΔPL climbs to ~93 bar:
That's a 16% velocity gain over nominal — but you are now running the G761 outside its characterised pressure band. Spool erosion accelerates, and the valve datasheet linearity figure of ±3% no longer applies.
Result
Peak rod velocity at nominal 210 bar supply with full ±10 mA command is 0. 081 m/s (81 mm/s). At a glance that feels slow — slower than a hand-cranked vice — but on a 60 m blade it produces visible whip at the tip and substantial inertial reaction at the root fixture. Across the operating range the velocity scales from 0.055 m/s at 100 bar through 0.081 m/s at 210 bar to 0.094 m/s at 280 bar; the sweet spot is firmly at 210 bar where the valve datasheet was characterised and linearity holds. If the lab measures peak velocity 20% below the 81 mm/s prediction, the most likely causes are: (1) a worn rod seal letting cap-end pressure bleed past the piston, which you confirm by capping the rod-end port and checking for crossport leak above 0.5 L/min, (2) supply-pressure droop during the velocity peak because the accumulator is undersized or pre-charge has dropped from 0.6 × PS to below 0.4 × PS, or (3) command-signal cable noise causing the controller to back off the demand — check the torque-motor current with a clamp meter, not just the controller's reported command.
Choosing the Hydraulic Servo: Pros and Cons
Hydraulic servos compete with electric servos and proportional hydraulic valves. The choice comes down to force, bandwidth, accuracy, and how much infrastructure you can justify. Below 10 kN of continuous force, electric usually wins on simplicity. Above 50 kN, hydraulic dominates. Proportional valves split the difference but give up bandwidth and precision.
| Property | Hydraulic Servo | Electric Servo (ball-screw) | Proportional Hydraulic Valve |
|---|---|---|---|
| Force capacity (typical) | 1 kN to 4 MN | 0.1 kN to 100 kN | 1 kN to 4 MN |
| Position accuracy | ±0.01 to ±0.05 mm | ±0.005 to ±0.02 mm | ±0.5 to ±2 mm |
| Closed-loop bandwidth | 50 to 200 Hz | 20 to 100 Hz | 5 to 30 Hz |
| Power density (kW/kg of actuator) | 5 to 10 | 0.5 to 1.5 | 5 to 10 |
| Filtration requirement | ISO 4406 16/14/11, 3 µm absolute | None | ISO 4406 19/17/14, 10 µm |
| Capital cost (10 kN class system) | $15,000-40,000 | $5,000-12,000 | $3,000-8,000 |
| Service life (typical) | 10,000-20,000 hr valve, cylinder >50,000 hr | 20,000-30,000 hr screw | 15,000-25,000 hr |
| Best application fit | High force + high bandwidth + accuracy | Moderate force + clean environment | High force + loose accuracy |
Frequently Asked Questions About Hydraulic Servo
You're hitting the oil-column resonance of the cylinder. The trapped oil between the piston and the valve forms a hydraulic spring; combined with the load mass it gives a natural frequency typically between 30 and 150 Hz. Once your controller bandwidth approaches that frequency, the loop goes unstable.
The fix is either a notch filter at the resonance, lower proportional gain with feed-forward velocity to compensate, or shorter hose runs between valve and cylinder — every metre of hose adds compliance and drops the resonance. Mounting the valve directly on the cylinder (manifold-mounted) typically pushes resonance up by 30-50%.
Almost certainly yes — you're flow-saturating. At standstill the valve only meters leakage flow, so position holds. During velocity peaks the valve is already fully open and physically cannot pass more oil, so the actuator falls behind the command.
Quick check: log the controller's commanded current versus actual flow. If commanded current sits at 100% during the lag, you're saturated. Either step up to the next valve frame size (Moog G761-3005 to 3007 doubles flow capacity) or reduce the commanded velocity profile. Don't try to tune your way out of it — you can't tune past a physical flow limit.
Jet-pipe valves (Moog 30 series, Abex/Parker BD series) tolerate contamination far better — a 25 µm particle that would jam a flapper-nozzle just deflects past the jet pipe. They're the standard for steel mills, mining, and forestry equipment where oil cleanliness will realistically degrade between filter changes.
Flapper-nozzle valves give better null stability and slightly higher bandwidth (300+ Hz versus 200 Hz typical for jet-pipe). Pick them for aerospace, precision test rigs, and lab environments where you can guarantee ISO 4406 14/12/9 oil. The price difference is marginal; the contamination tolerance is the deciding factor.
Thermal expansion of the torque motor frame and oil density change with temperature. As supply oil heats from 30°C to 55°C, the torque motor's magnetic gap shifts micrometres, and the spool finds a new mechanical centre that no longer aligns with zero electrical command.
The proper fix is electrical null trim implemented in the controller — most modern drive cards do this automatically by zeroing command at no-flow conditions during a brief calibration cycle. If you're running an older analogue card without auto-null, add an oil cooler to hold tank temperature within ±5°C, which keeps drift under 0.5%.
You can run dirtier oil, but valve life drops roughly with the cube of particle count above the rated cleanliness. Going from 16/14/11 to 18/16/13 quadruples particle counts and typically cuts servo-valve life from 15,000 hours to under 4,000 hours. The valve doesn't fail catastrophically — it gets sticky, null shifts, hysteresis grows from 3% to 8%, and your closed-loop performance silently degrades.
If the budget says you can't run a 3 µm pressure-line filter plus an offline kidney loop, you should be looking at a proportional valve and accepting the lower performance, not a servo valve in dirty oil.
Yes, but not by the servo valve itself. A spool valve cannot be trusted to block flow with the precision needed to hold a load — internal leakage is always 1-3% of rated flow and the load will drift. The standard solution is a pilot-operated check valve (counterbalance valve) plumbed directly into the cylinder ports, which mechanically locks the cylinder when the servo loop releases.
For aerospace and lift applications, designers also add a separate solenoid-operated dump valve that vents supply pressure to tank on a fault, and bias the cylinder so the load drifts to a safe position rather than continuing under power. Never rely on the servo valve null position alone for load holding.
Two usual culprits. First, the LVDT excitation frequency (typically 2.5 or 5 kHz) is beating against switching noise from the hydraulic pump VFD or the servo amplifier PWM. Move the LVDT carrier frequency away from harmonics of the VFD switching frequency — most signal conditioners let you pick 2.5, 5, or 10 kHz.
Second, ground loops. The LVDT body, the cylinder, the controller chassis, and the power pack are often bonded to different earth points. Float the LVDT shield at the actuator end and ground only at the controller. If noise persists, you have a mechanical issue — check that the LVDT core rod is not contacting the bore, which shows up as random spikes rather than continuous hash.
References & Further Reading
- Wikipedia contributors. Servomechanism. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
