A servo brake is a small spring-applied, electromagnetically-released holding brake mounted on the back of a servo motor that locks the shaft whenever power is removed. Unlike a dynamic friction brake sized to dissipate kinetic energy, a servo brake is sized purely for static holding torque — it clamps a stationary shaft, it does not stop a moving one. Its purpose is to hold a servo-driven axis in place when the drive is disabled or power fails, preventing a vertical robot arm or CNC Z-axis from dropping under gravity. Typical units hold 0.32 to 50 Nm and release in 30 to 80 ms once the 24V coil energises.
Servo Brake Interactive Calculator
Vary the vertical load, pulley radius, gearbox ratio, efficiency, and safety factor to size the required static holding torque for a servo brake.
Equation Used
This calculator sizes the servo brake for static holding torque. The load torque at the brake is the gravity torque divided by gear ratio and efficiency; the safety factor then gives the required rated holding torque. Servo brakes should not be sized as dynamic stopping brakes.
- Brake is used for static holding only, not dynamic stopping.
- Load torque is due to gravity on a vertical axis.
- Gear ratio is motor speed reduction to the load.
- Efficiency accounts for gearbox and drivetrain losses.
Operating Principle of the Servo Brake
A servo brake bolts onto the rear of a servo motor, sharing the same shaft as the encoder. Inside the housing you have three things — a steel armature plate, a friction disc keyed to the motor shaft, and a stack of compression springs pressing the armature against the friction disc. With no current flowing, the springs clamp the disc between the armature and a fixed mounting flange, and the rotor cannot turn. Energise the 24V coil and the magnetic field pulls the armature back against the springs by 0.2 to 0.4 mm, releasing the disc. The shaft now spins freely and the servo loop takes over.
Why spring-applied and not spring-released? Because a servo brake is a fail-safe brake. If the power supply drops, the cable breaks, or the drive faults, you want the brake to engage automatically — not stay open. That is the entire reason it exists. The cost is roughly 2 to 4 W of continuous coil power during normal operation, which is trivial next to a servo running 200 W or more.
Get the air gap wrong and you get one of two failures. If the gap creeps above 0.5 mm because the friction lining has worn, the coil cannot pull the armature in cleanly and you get release chatter or the brake never opens — the motor stalls against a clamped shaft and trips on overcurrent. If the gap is too tight, residual magnetism in the armature drags on the disc after de-energisation, and the brake never fully engages, letting a vertical axis sag a few millimetres per minute. Manufacturers like Kendrion, Mayr ROBA-stop, and Miki Pulley specify the air gap to ±0.05 mm and supply shim kits to re-set it as the lining wears.
Key Components
- Friction disc (rotor): A thin steel hub with bonded friction lining on both faces, keyed or splined to the motor shaft. Typical thickness is 3 to 6 mm with a lining wear allowance of 0.5 to 1.0 mm before the brake must be reshimmed or replaced. The lining material is a non-asbestos resin-bonded composite rated for dry running.
- Armature plate: A ferromagnetic steel disc that floats axially on guide pins. The compression springs push it against the rotor; the coil pulls it away. Flatness must hold within 0.05 mm — any warping causes uneven clamping force and rotor squeal at release.
- Compression springs: Six to twelve helical springs distributed around the armature, sized to deliver the rated static holding torque with a safety factor of 1.5 to 2.0. Spring force degrades roughly 5 to 10% over 10 million cycles, which is the reason rated holding torque is conservative.
- Electromagnetic coil: A copper coil potted in epoxy, typically wound for 24 VDC at 0.4 to 1.5 A depending on size. Release time is 30 to 80 ms; engage time after de-energisation is 15 to 40 ms. A flyback diode across the coil extends release time, so high-speed servo applications use an external suppression network instead.
- Mounting flange and air gap shims: The fixed reaction surface bolted to the motor end-bell. Shim stack sets the nominal air gap, usually 0.2 to 0.3 mm. Service technicians remove shims as the friction lining wears to keep release current consistent.
Real-World Applications of the Servo Brake
Anywhere a servo motor drives a load that wants to move under gravity or stored energy when power goes off, a servo brake is the standard answer. Six-axis industrial robots, vertical-axis machine tools, electric forklifts, theatre flying systems, and surgical robots all rely on small power-off holding brakes integrated into the servo motor itself. The common thread is that the brake never sees high-speed engagement under load — the servo decelerates the axis to a stop first, then the brake clamps the stationary shaft. If the brake ever has to stop a moving load, that is an emergency event and the lining wears noticeably each time it happens.
- Industrial robotics: Joints 2, 3, and 5 of a FANUC R-2000iC or KUKA KR QUANTEC six-axis arm — gravity-loaded axes that must hold the arm payload static when the controller is powered off.
- CNC machine tools: Z-axis servo on a Haas VF-2 or DMG MORI NHX vertical machining centre, holding the spindle head against gravity during tool changes and power-off conditions.
- Material handling: Hoist motors on Demag KBK light-crane systems and electric stacker trucks like the Crown SP series, where a dropped load is a safety incident.
- Medical robotics: Joint brakes on the Intuitive da Vinci surgical system arms, ensuring instruments cannot drift when the surgeon disengages the master controls.
- Entertainment and rigging: ETC Prodigy and Stagemaker SR motorised flying winches in theatre fly-towers, holding scenery and lighting trusses suspended over performers.
- Semiconductor and lab automation: Vertical Z-stages on wafer handlers and pick-and-place machines like the ASM AD8312, where a 5 kg payload sitting 300 mm above a $20,000 wafer cannot be allowed to drop.
The Formula Behind the Servo Brake
The core sizing equation tells you the minimum static holding torque the servo brake must deliver to hold a gravity-loaded axis without slip. At the low end of the typical operating range, you have a small horizontal robot wrist where gravity torque is negligible and you only size against process-disturbance torque — a 0.32 Nm brake on a 200 W servo is plenty. Mid-range is the sweet spot for most installations: a vertical Z-axis carrying 20 to 50 kg through a ballscrew reduction, needing 2 to 8 Nm of holding torque at the motor shaft. At the high end, you have direct-drive vertical axes on heavy gantries where the brake sees 30 to 50 Nm and any under-sizing leads to slow gravity sag while the machine sits idle overnight.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| Tbrake | Required static holding torque at the motor shaft | Nm | lb-ft |
| SF | Safety factor, typically 1.5 to 2.0 for vertical axes | dimensionless | dimensionless |
| m | Mass of the gravity-loaded payload plus moving carriage | kg | lb |
| g | Gravitational acceleration | 9.81 m/s² | 32.2 ft/s² |
| r | Effective lever arm — ballscrew lead / 2π or pulley pitch radius | m | ft |
| i | Gear or belt reduction ratio between motor and load | dimensionless | dimensionless |
| η | Drive train back-driving efficiency (ballscrew typically 0.6 to 0.9) | dimensionless | dimensionless |
Worked Example: Servo Brake in a vertical pick-and-place Z-axis
You are specifying the servo brake on the Z-axis of a benchtop pick-and-place machine — think Juki RS-1R class — where a 12 kg gantry head plus tooling rides on a 10 mm-lead ballscrew driven directly by a 400 W AC servo. The machine must hold position indefinitely with the drive disabled, including overnight power-down, without the head drifting more than 0.1 mm.
Given
- m = 12 kg
- g = 9.81 m/s²
- ballscrew lead = 10 mm/rev
- i = 1 (direct drive)
- η = 0.85 (precision ground ballscrew)
- SF = 1.8 (vertical axis)
Solution
Step 1 — convert the ballscrew lead to an effective lever arm. A 10 mm lead means one full motor revolution lifts the load 10 mm, so the equivalent radius is the lead divided by 2π:
Step 2 — compute the nominal required holding torque at the motor shaft, using SF = 1.8:
So you need a brake rated for at least 0.4 Nm. The next standard size up is typically 0.5 or 1.0 Nm — pick the 1.0 Nm Kendrion 86 611 or equivalent Mayr ROBA-stop micro. That gives you margin without oversizing the rotor inertia, which would slow the servo's settling time.
Step 3 — check the low end of the typical operating range. If the same machine ran a lighter 6 kg head, the required torque drops to roughly 0.20 Nm — well within a 0.32 Nm brake, the smallest stock size most servo OEMs offer:
At this end of the range the brake is barely working — gravity sag is undetectable and you can leave the machine powered off for a week without measurable drift.
Step 4 — high-end check. Same screw, but loaded with a 30 kg head (a heavier through-hole odd-form placement head):
You are now right at the limit of a 1.0 Nm brake. In practice you would step up to a 2.0 Nm unit, because back-driving efficiency η climbs to 0.9 when the screw is freshly greased and warm — meaning the load reflects more torque to the motor shaft, not less, and a marginal brake will slip a few microns per hour. That shows up as the head sitting 50 to 100 µm low after an overnight shutdown, which is enough to crash a 0402 nozzle into a board fiducial on power-up.
Result
The Z-axis needs a servo brake rated for at least 0. 4 Nm static holding torque, and you should specify the next stock size up — a 1.0 Nm power-off holding brake. At the low end of the load range (6 kg head) the brake is barely working at 0.2 Nm and overnight drift is unmeasurable; at nominal 12 kg you sit comfortably at 0.4 Nm with margin; at 30 kg you are pushing a 1.0 Nm brake and should step up to 2.0 Nm to absorb efficiency variation. If you measure overnight drift greater than 0.1 mm with a brake that should hold the load, the three most likely causes are: (1) coil keep-alive voltage leaking from a poorly-wired drive enable circuit so the brake never fully engages, (2) friction-lining contamination from ballscrew grease mist reducing the effective coefficient by 30 to 50%, or (3) air gap drifted above 0.4 mm from lining wear, leaving residual magnetism that holds the armature partially open after power-off.
Choosing the Servo Brake: Pros and Cons
Servo brakes compete with two other approaches for holding a stationary servo axis: a separate dynamic disc brake sized to actually decelerate the load, and a self-locking drive train (worm gear or low-lead ballscrew) that uses friction in the transmission instead of a dedicated brake. Each fits a different problem.
| Property | Servo brake (power-off holding) | Dynamic disc brake | Self-locking drive train |
|---|---|---|---|
| Primary function | Static holding only | Dynamic stopping under motion | Static holding via drive friction |
| Typical holding torque range | 0.32 to 50 Nm at motor shaft | 5 to 5000 Nm at output shaft | Limited by worm or screw self-locking, typically up to 100 Nm equivalent |
| Engage / release time | 30 to 80 ms release, 15 to 40 ms engage | 50 to 200 ms full clamp | Instantaneous (passive) |
| Fail-safe behaviour on power loss | Engages automatically (spring-applied) | Depends on actuator type | Always engaged |
| Drive efficiency cost | 2 to 4 W coil power during running | None during running | 10 to 50% efficiency loss in worm gearing |
| Typical lifespan | 10 million cycles dry, 5 million under load slips | 1 to 5 million dynamic stops | Lifetime of the drive train |
| Cost (added per axis) | $80 to $400 for an integrated unit | $300 to $3000 for a sized caliper or disc brake | Built into drive cost |
| Best application fit | Vertical servo axes, robot joints, CNC Z-axes | Hoists, large cranes, emergency-stop loads | Slow positioning stages, lift platforms |
Frequently Asked Questions About Servo Brake
This is almost always a thermal contraction issue combined with marginal sizing. As the machine cools down, the ballscrew shrinks axially and the friction lining cools and contracts slightly, which reduces normal force on the rotor. If you sized the brake right at the calculated minimum torque without safety factor, you have no margin to absorb that 5 to 10% reduction.
The fix is to step up one size in holding torque, or pre-tension the ballscrew bearings so axial shrinkage doesn't transfer to the brake stack. Check that the coil is fully de-energised at rest too — some drives leave a holding voltage on the brake line for fast re-release, which can keep the armature 50% pulled-in.
You can, but you'll wear it out fast. Servo brakes are sized for static holding torque, not energy dissipation. Each emergency stop on a loaded vertical axis can wear 0.05 to 0.1 mm of lining — meaning 10 to 20 hard stops will consume the entire wear allowance. The lining also glazes when overheated, which drops the coefficient of friction by half and turns your safety brake into a sliding clutch.
The correct architecture is: drive performs controlled decel under E-stop using regenerative braking through the drive resistor, then the holding brake clamps the already-stopped shaft. Reserve the brake-only stop for the case when the drive itself has failed.
Put the brake on the motor shaft whenever you can — that's where servo brakes are sized to live. Required torque scales down by the gear ratio, so a 100:1 reducer needs 1/100th the brake torque if the brake sits before it. A motor-side 1 Nm brake replaces a load-side 100 Nm brake, which costs five times more and weighs twenty times as much.
The exception is when you cannot trust the gearbox to back-drive safely — for instance a planetary with a broken carrier. If the gearbox can fail in a way that disconnects motor from load, you must put the brake on the load side. This is why some heavy industrial robot wrists use load-side brakes despite the cost penalty.
Doubled current draw means the magnetic circuit is fighting an enlarged air gap. The two common causes are friction lining wear (the rotor has thinned, so the armature has to travel further to clamp) and bolt-up distortion of the mounting flange (uneven torquing warps the reaction face by 0.1 to 0.2 mm).
Measure the air gap with a feeler gauge through the inspection slots — if it exceeds the spec value (usually 0.3 mm) by more than 0.1 mm, pull a shim from the stack. If shimming doesn't bring current back down, the lining is at end-of-life and the rotor needs replacement. Running at doubled current also overheats the coil potting and shortens insulation life by roughly 50% per 10 °C above rated temperature.
The clack is the armature slamming into its magnetic stop when the coil energises. It's normal up to a point — every spring-applied brake makes some release noise. It becomes a problem when the impact peaks above roughly 5 g of acceleration on the armature, because the friction lining can micro-fracture at the bond line.
If the noise has gotten louder over time, your air gap has grown. The armature now travels further before hitting the stop, so it arrives with more kinetic energy. Re-shim to nominal gap and the clack should quiet down. If you need silent release for a medical or theatre application, look at units with built-in damping rings or a controlled-release driver that ramps the coil voltage over 20 to 30 ms instead of switching it on hard.
Most servo brakes tolerate ±10% on coil voltage, so 21.6 to 26.4 V is fine for a 24V unit. Below that, magnetic pull drops with the square of voltage and you risk incomplete release — the armature lifts partially, the rotor still drags, and you cook the lining. Above +10%, coil heating rises with voltage squared and the insulation degrades faster.
The trap is voltage drop on long cable runs. A 20 m cable to a robot wrist with 0.5 mm² conductors can drop 2 to 3 V at 1 A coil current. Measure voltage at the brake terminals, not at the supply, and upsize the cable if you see less than 22 V at the brake during release.
References & Further Reading
- Wikipedia contributors. Electromagnetic brake. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
