HX711 Torque sensor

Hello, i am trying to make a torque sensor using an HX711 load cell amp, Arduino Nano 33 IoT, and a Wheatstone bridge (2 resistors and 2 strain gauges). I am confused as to how i can get this torque value when in the calibration sketch, it seems as if the load cell amp can only measure force. I did manage to find this tutorial online: High Tech - Low Buck Digital Torque Meter : 4 Steps (with Pictures) - Instructables but they don't explain the changes to their code. I need to do exactly as they have done, but i am not using a load cell. I know i can calculate Torque by change in resistance (already have that formulated) but that wouldn't work as the load cell amp doesn't return resistance. Any help would be much appreciated, Thanks!

Hi, @nav651
Welcome to the forum.

Please read the post at the start of any forum , entitled "How to use this Forum".

The 711 will output a voltage proportional the the voltage difference across the bridge.
This is no matter what sort/source the bridge is measuring.
You will need to calibrate this value against known torques.

Your strain gauges measure change in stress, which is proportional to force, which is proportional to torque if you are using it in a torque measuring jig.

Can you please tell us your electronics, programming, arduino, hardware experience?

The Instructables is a typical Instructables project, a lot of stuff, but not a lot of information explaining the intricacies of the code.

Have you written any code yet to check your bridge?
Can I suggest you use one of the example codes that come with the HX711 library.

If you have got a circuit, can you please post it?
If you have code can you please post it?

Thanks.. Tom... :grinning: :+1: :coffee: :australia:

Second that. How do you propose to measure torque with strain gages? Post a diagram of your setup.

Strain gages can be built into load cells (as in the posted link) and used to measure force (such as weight). The link shows they'll also respond to force resulting from being twisted. I believe torque is the cross product of the Force vector and the Radius vector at which it's applied. If the Radius magnitude and angle it forms with the force vector are constant, then torque will be proportional to the force.

Firstly that project is bogus I'm afraid - you can't apply a torque to that sort of load-cell and get meaningful results from it - it looks like they actually measure the bending moment at the sensor, not the torque, so the reading will depend on the geometry of the device applying torque.

Many instructables projects are riddled with gross errors and misunderstandings, there seems to be absolutely no curation goes into the site whatsoever, they just want the advertizing revenue and host anything AFIACT.

A torque measurement system needs strain guages setup with the correct geometry and/or mechanism to measure torque only (not affected by axial loading nor bending moment).

Isn't torque a force applied at a radius? Measure the radius and multiply the measured force by the radius.

Yes, r "cross" f

The last three figures show how to apply strain gauges to measure torque Wheatstone Bridge Circuit | Strain Gauge | HBM Note there are no passives.

Regarding the HX711 output and how to get torque...

It would be possible to estimate the torque by calculation only, using fundamentals of mechanics of materials, the properties of the strain gauges, and the HX711 behavior, but this the best approach:

The calculation method should be done, regardless, to ensure that advantage of the HX711 full range is taken but not exceeded.

Sorry for the lack of info guys but I've encountered a problem unrelated to the end result. I'm taking over this project from someone else, and they had setup the Arduino Nano, HX711, and a Wheatstone bridge setup on a socket wrench (2 strain gauges, two resistors). using one of the standard HX711 libraries example code (to get basic print out) I only get an output of -1 The wiring I'm following is the previous persons (they had an actual good output) so I don't believe that to be the issue. I've also swapped out the Nano for another one as well as the breakout board, the only thing I cannot change is the wrench-bridge setup. The bridges connections seem to be intact so I'm not sure if its really at fault or not. I think I am going to buy a standard load cell and try that so see if the wrench is the part that actually does not work, then focus on getting a torque value. Sorry for the incomplete question/setup :frowning:

This was the method I was going to follow. I have a digital torque meter that I can calibrate against.

Load cells do not measure torque or force. They measure strain (deflection) of the beam, and the force or torque inducing that deflection can then be inferred from the strain. The beam will deflect as a result of almost any applied force, be it bending or twisting. Load cell beams are typically constructed to optimize measurement of bending strain, but they WILL measure torque strain as well.

Calibration is a simple matter of applying a known force or torque, and taking a measurement. That measurement gives you a scaling factor that you use to convert other measurements into the corresponding torque. More calibration measurements would be required to determine the linearity and accuracy of the full range of applied forces/torques.

Yes, my original plan was to use the resistance values to calculate torque such as the attached picture but the software wouldn’t work in that way. So right now since I can’t get this homemade load cell to work, I’m gonna order one, drill a nut into it and use my digital torque measurement to calibrate the software.

Are you getting any readings from the bridge.
You may be better off starting with one of the examples in the HX711 library.

You do not need to calculate resistance.
Just correlate the raw digital data output of the 711 with the applied calibration torques.

Tom... :grinning: :+1: :coffee: :australia:

Beware. There are Nanos and Nanos.
A Nano 33 IOT (3.3volt logic) is NOT a 'classic' Nano (5volt logic). The Nano 33 IOT can only work reliable with a HX711 board from sparkfun.com, with FIVE wires between Nano and HX711.

I doubt that a shear beam load cell (as used in the Instructable) can be calibrated with a torque wrench...and then used to measure torque from, say, a drill, with much accuracy.

During calibration, the load cell will respond to both the shear and the torque produced by a torque wrench, whereas a drill produces no (or little) shear and only (or mostly) torque.

If the shear during calibration is large enough, it will confound subsequent attempts to determine torque in the absence of shear.

As usual, it would be helpful to know the end goal...

Yes I’m using the spark fun HX711 but I only have 4 wires coming into it from the Wheatstone bridge

The end goal is to just measure the torque for any application that’s all. I will be ordering a load cell to experiment with

Can you post a picture of the strain gauge setup you have?
Can you please post a circuit diagram?
Can you please post a link to data/specs of the strain gauges?

Do you have a DMM?

Thanks.. Tom... :grinning: :+1: :coffee: :australia:

Not what I asked.

The Wheatstone bridge has four connections,
but are there five wires to the Nano 33 IOT.

Hint: 5volt to VCC, and 3.3volt to VDD.

1 Like

Its the tangential component of a force applied, multiplied by the closest radial distance of the line-of-action of the force from the axis of rotation - you have to be careful if the geometry isn't orthogonal.

Not how I'd phrase it - they might respond in an ill-conditioned manner to torque, but not in any way that you'd call a measurement - its an erroneous mode of action for the geometry of the sensor which is designed to measure shear force in a particular direction only. When you set up a load-cell its important that only forces of the correct geometry are applied (this sort of sensor is not designed to measure bending moment either, just shear due to a force acting at the correct point in the correct axis. Basically read the manual for the sensor and use it as its intended otherwise you aren't assured of any repeatability, accuracy, linearity, and the sensor may even be damaged.