Torque sensor capacity should be based on the maximum motor torque rather than nominal or maximum continuous torque, because the latter do not account for the motor's potential peak torque.
The calculation follows the motor-to-load chain: motor speed equals load speed multiplied by gear ratio, motor torque equals load torque divided by ratio and efficiency, and the selected sensor must cover the resulting peak [S1][S3].
Breakdown Torque Sets the First Capacity Limit
Breakdown torque, also called pull-out torque, is the brief maximum torque a motor can produce before its speed collapses and the motor stalls [S2]. It is a better basis for preliminary torque sensor capacity than rated or continuous torque, because a transient stall can impose far more load than the normal operating cycle [S4].
Use the motor data at the intended operating point and distinguish 3 torque conditions: starting or locked-rotor torque, full-load continuous torque, and breakdown torque [S2]. For an unprotected installation where the sensor itself must survive a stall, select a working range that covers breakdown torque with the required overload margin. For a protected installation with a documented trip or overload control, the engineering basis may be the maximum credible operating torque, including the permitted overload interval.
A worked chain can make the selection checkable.
Apply Gear Ratio, Efficiency, and Location
The reduction ratio scales ideal torque from motor to load, while efficiency reduces the torque actually delivered [S1][S3]. For ratio i and efficiency η, the output torque is motor torque multiplied by i × η. Conversely, the equivalent motor torque required at the load is load torque divided by i × η [S1].
Do not apply the same ratio and efficiency twice when calculating the location of the sensor. A motor-side sensor sees motor-shaft torque before the gearbox; a gearbox-output sensor sees load-side torque after multiplication by the ideal ratio and adjustment for losses. A load-side reaction sensor also measures the reaction path rather than the rotating output shaft, so its connection, shaft, and bending loads must be compatible with that arrangement [S4].
Compare the common mounting options against four criteria:
The gearbox must also be checked independently. A sensor can survive its selected torque while the gearbox, coupling, key, or shaft reaches its own lower limit. The device with the lowest rated capacity, including the sensor, defines the weakest point in the mechanical chain [S4].
Separate Rated Capacity from Measurement Resolution

Maximum capacity answers how much torque the sensor can survive, while measurement resolution answers how small a useful signal can be detected reliably [S4]. These are different requirements: a high-capacity sensor can still be unsuitable if it cannot resolve the small torque produced during normal operation.
For a servo with a known torque gradient, multiply the gradient in N·m/RPM by the motor speed in RPM to estimate torque. In general, for torque gradient g in N·m/RPM and speed n in RPM:
T = g × n
That minimum measurable torque establishes the low end of the working range, while breakdown torque establishes the high end. Check that the proposed sensor can resolve the minimum useful torque while retaining overload protection for the maximum credible event [S4].
Accuracy requirements must therefore be stated as more than a percentage. The engineering package should define minimum operating torque, nominal continuous torque, starting torque, breakdown torque, peak duration, allowable overload, and whether reversal torque is present [S2][S4]. A torque sensor without those operating limits is only partially specified.
Account for Duty Cycle, Shock, and Inertia
Sensor sizing must cover the entire duty cycle, not only steady-state running torque [S2]. Startup torque must overcome inertia and static friction, while temporary overloads can produce torque above continuous output [S2]. If the sensor measures the motor shaft, reflected inertia and startup behavior remain in that measurement; if it measures the output shaft, the calculation must use the actual gear ratio and efficiency.
Three timing questions determine whether a rated range is realistic. First, estimate how long breakdown torque can persist before the motor stalls or the protection system removes power. Second, identify whether the event is repeatable or a one-time commissioning anomaly. Third, establish whether a coupling, brake, or clutch removes torque before the sensor sees it [S2][S4].
Also separate torque from bending and axial loads. An inline sensor that is rated for high torsion but mounted on a shaft with a large moment arm may still fail from bending stress. Record shaft diameter, keyway geometry, coupling mass, runout, and radial or axial load where applicable. Torque is defined about a specific axis, so a valid measurement requires the force, shaft axis, and moment arm to be properly defined [S3].
For startup-limited applications, the selection sequence should also be reviewed in Size Torque Sensors for Motor Startup Spikes, where transient peaks are treated separately from continuous operating load.
Verify Shaft, Speed, and Interface Constraints

Electrical compatibility must be selected after the mechanical range is defined. Possible sensor outputs include ±10 VDC, 4–20 mA, mV/V, and SPI; a DAQ or PLC that accepts one interface may require an amplifier for a full-bridge mV/V sensor [S4]. The complete signal specification should include supply voltage, output type, bandwidth, cable requirements, and whether bidirectional torque must be distinguished.
Speed matters because a sensor designed only for static calibration may not resolve torque from a motor operating at 1,000 RPM [S4]. Confirm rated shaft speed, torsional response, allowable operating temperature, and the effect of the sensor, coupling, and adapter on system inertia. The dimensional checks are equally specific: shaft diameter, keyway presence or keyless clamping method, flange bolt pattern, thread specification, and available axial length.
A useful acceptance table should contain at least 8 fields: sensor location, nominal continuous torque, starting torque, breakdown torque, peak duration, service factor, maximum shaft speed, and output interface. Add minimum measurable torque and permissible bending moment where the installation creates them. This turns a catalog-capacity choice into an auditable engineering decision.
Build a Traceable Selection Record
The final sensor rating should remain above every credible mechanical demand while the measurement range remains useful at the opposite end of the duty cycle. Record the source for each input: motor datasheet, gearbox ratio, rated efficiency, duty cycle, sensor datasheet, and system interface specification. Motor and gearbox efficiency must not be represented as a universal constant because efficiency varies with load, speed, temperature, and operating point [S1][S2].
Use the following verification sequence:
The cleanest selection record states the motor breakdown torque, gearbox ratio, efficiency, calculated peak at the sensor, selected service factor, minimum operating torque, chosen sensor range, and interface. If any field lacks a value, the sensor specification remains incomplete even when the catalog lists a nominal torque rating.