• Klajd Lika

How to Select a Force Torque Sensor

Making sure you get the best force torque sensing solution for your application can be complicated. Selecting the best Force Torque Sensor is dependent on the specific application. Understanding the types of features, specifications, and software needed, can be tricky. This comprehensive guide will make the process of selecting the best force torque sensor for your application a little easier.

At Bota Systems, when we pair a sensor with an application, we focus on the most important specifications, these commonly being range, noise free resolution, sampling rate, signal type, and size.

Force and Torque Range

We define the range as how much force or torque a sensor is capable of measuring starting at 0 N.

For example:

A scale may have a limited weight capacity of 180 Kg, and if a 200 Kg person steps on the scale, it will probably still register their weight but will output 180 Kg instead of the accurate weight of 200 Kg because it has a range limitation of the scale is 0-180 Kg.

For force-sensing applications and measuring, the sensor's range needs to be greater than, or equal to, the maximum force to be applied, measured, and controlled.

The FT sensor's specification sheets will usually define a range for single-axis loadings. Some specification sheets will illustrate with a 2D graph all the combinations that can be achieved in multiaxial loadings. All combinations that fall inside the graph should be accurate. Combinations that fall outside of the graph are known as saturated signals. These saturated signals will lead to inaccurate force measurements- similar to our example of a 200 Kg person using a scale with a limited weight capacity of 180 Kg.

Combined Loadings graph for force sensing applications

Selecting a sensor that is capable of measuring the maximum force to be applied is essential.

Noise Free Resolution

The easiest way to select a sensor is by choosing a sensor that can detect the minimum force to be measured from the application.

We define minimum measurable force as the minimum amount of force a sensor can measure and the incremental changes in measurements it can sense. At Bota Systems, we call this noise-free resolution.

Knowing the minimum measurable force needed for your application and directly comparing it with the noise-free resolution of the sensor is the best way to select a sensor with proper sensitivity. The noise-free resolution should be smaller than, or equal to, the minimum force measured.

The noise-free resolution of Bota Systems' sensors is defined as the 6σ of a signal over 1 second of measurements in stable environmental conditions.

Note: Noise-free resolution is equivalent to peak-to-peak noise.

One can use this application note from Analog Devices to calculate the effective resolution or other signal characteristics associated with noise for more sophisticated signal processing.

Minimum Sampling Frequency

The sampling frequency is essentially the amount of data a sensor collects per time unit (seconds). These data packets deliver to the output port of the sensor (Serial or EtherCAT) with the same rate and very small latency, which is discussed later. It is measured in samples per second and sometimes in Hz.

For example:

When we record a video on our smartphones, the video is essentially consecutive images appearing one after another captured from the vision sensor of the smartphone. The amount of images appearing within a second is called the frame rate or frame frequency. Frame rate is expressed in Frames Per Second (fps), and Frame Frequency is expressed in Hertz (Hz). The quality of the video reconstruction on our screens depends on how many images we have per second. The higher the fps, the smoother and more natural the video looks. Below 30 Hz a human eye can detect flickering, and some information is lost. Therefore, it is recommended to record or playback a video with more than 30 Hz, to have smoother image quality for the human eye to detect more information.

The same happens with a force sensor signal. Instead of frames per second, we use samples per second. Instead of the three-dimensional nature of a video signal, the force sensor signal is simpler and one-dimensional and corresponds to force (N).

If you have a sensor with a 1000 Hz sampling rate and excite the sensor by pressing on it, that means that the sensor collects one sample every 1 millisecond.

The sampling rate is important because it correlates to the ability of the sensor to reconstruct a signal with detail. The faster a robot’s computer can have this information, the faster it can take action and correct the applied force.

The identification of the overall system’s bandwidth has to do with the application dynamics. If the application controls impact forces between two low mass metal pieces, then a 1000Hz sampling frequency may not be enough to detect the spikes during the hard contact. Nevertheless, most industrial robotic applications do not need such high requirements, and anything between 200 to 500 Hz is usually sufficient.

Advanced robotics, such as fast quadrupedal locomotion, may require ultra-high sampling frequency because the feet can have contact with the ground for less than 20 milliseconds. Within this small timeframe, robot controllers require high-frequency sensors to analyze the signal, process it, and respond quickly.

Here is a list of sampling frequency ranges for common force-sensing applications:

  • Wrist force-torque sensor- 50 to 200 Hz is usually enough due to slow controllers

  • In hand manipulation- 50 to 200 Hz

  • Aerodynamic force measurements- 5 to 100 Hz

  • Force measurement for surgical robots- 100 to 1000 Hz

  • Bipedal locomotion- 400 to 1000 Hz

  • Monitoring and post-processing- The higher, the better. The advantage of post-processing is that the signal can be denoised after recording

  • Hand guidance- 10 to 200 Hz

  • Polishing- 10 to 500 Hz

  • Grinding- 50 to 500 Hz

  • Assembly of solid high stiffness components like PCBs- 200 to 1000 Hz

Signal Type

To better understand signal types concerning force torque sensor selection, you may want to familiarize yourself with signals. Categorized by their output, the two main types of signals, and force torque sensors, are analog and digital.

Analog Force Torque Sensors

Analog force torque sensors require extensive programming, external electronics, and a connection to a network protocol, making this force sensing solution more expensive, time-consuming, and complex.

However, the advantage of using an analog sensor is that it can operate under extreme temperatures, and why they are still used in applications today.

Applications that need analog sensors are those where the environmental conditions such as temperature, humidity, pressure are extreme.

Digital Force Torque Sensors

Digital force torque sensors have the advantage of EMI immunity, no induced or parasitic load from cables to affect their signal, quicker integration time, lower complexity, and overall less expensive. Also, due to embedded electronics and firmware that takes care of the necessary calculations, the digital sensors come with certain advantages against the analog ones:

  • Low and predictable noise levels

  • Ready to use data in N and Nm

  • Low drift from temperature fluctuations because it is calculated in the device

  • No signal aliasing. A digital sensor either works or not

  • No additional components or external electronics are required when compared to analog sensors

With new technology and innovation, digital sensors can operate in the same extreme temperatures and environments as analog sensors.

If a digital sensor can work with your application, we recommend selecting one because analog sensors are more complex and prone to measurement errors if you are not an expert.

Force Torque Sensor Size

When it comes to sensor size, the most important dimensions are the diameter and the height of the transducer. When selecting a sensor for robotic manipulation, consider a shorter sensor since taller sensors reduce the effective payload of a robot.

Bota Systems sensors are designed to be compact and eliminate external adapters needed to minimize the offset from any EOAT. The picture below is a Universal Robot integrated with a SensONE force torque sensor and EOAT gripper and showcases how simple force sensing can be with Bota Systems.

multi axis sensor on a robot arm with an EOAT gripper


Understanding the specifications your applications requires for range, noise free resolution, sampling rate, signal type, and size will allow you to use the process of elimination and narrow down your search for your best force sensing solution.

If your application is unique or you are still not sure of the best solution for you, we recommended contacting us, and we will help pair you with the right solution.