Recommendation for a torque cell

[robnub]

Hi all,

Just found out about this informative forum. Searched everywhere for torque cell however, theres no topic on this

Can anyone recommend me a torque cell which can measure up to 5 Nm and preferably a small one (to fit onto a medium-sized RC helicopter)?

Thank you!

[newInRobotics]

Have a look at futek.com selection, they have what You need, however, them sensors are very expensive.

[robnub]

any torque cells in the price range of below $200?

[jonagik]

You could approximate it as follows:

Mechanical power = speed (rad/s) x torque (Nm)
Mechanical power = Electrical power x efficiency = Voltage x current x efficiency

Torque = (Voltage x current x efficiency)/speed

Speed (rad/s) = (Speed(rpm) x 2 x Pi) / 60

This should be fairly accurate depending on your application.

[robnub]

Thanks jonagik.

I need a torque sensor to determine to torque output by a RC helicopter. The mathematical calculation would be quite tedious as I am building a control system for an automated heli.

[jonagik]

Why do you want to know torque?

If you attach a microcontroller (to perform the calculations) to the helicopter as well as a current sensor (these are little ics that cost a couple of dollars I believe) and a device for measuring RPM (perhaps a switch which is activated by the rotor turning - maybe a Hall effect sensor and a small magnet?) you can approximate torque using my calculations in real time.

The information can then be received by the microcontroller in a number of ways. You could send it in real time to an external device (a computer, an LCD screen, etc). You could save it to the device's or external memory for recovery later.

I don't see what's tedious about it

[Admin]

I need a torque sensor to determine to torque output by a RC helicopter. The mathematical calculation would be quite tedious as I am building a control system for an automated heli.

What is a 'torque output by a RC helicopter'?

Do you mean lifting force? Or perhaps the aerodynamic torque resistance on the rotating blades?

[robnub]

More like the motor torque used to generate the lifting force of the heli.

[Admin]

Torque is force multiplied by distance, where as lift is just a force.

When the blades spin, air resists the blades from spinning. This resistance, which is a counter-torque to the motor, is only semi-proportional to lift force. You can measure that torque, which is what I think you want, but it won't tell you much about the lift force being generated (if that's your goal).

5Nm is quite a bit of torque . . . to get any quality sensor it'll cost several hundred, and even then I'm not sure how you can mount it on a helicopter (with all that wiring) without it destabilizing flight.

Here is another idea . . . remove the motor from the helicopter and characterize it. That basically means determine the torque vs rpm curve at your required voltage. Basically the higher the torque, the lower the rpm.

Then put the motor back on the heli, and use an encoder to measure the rpm in real time. You can then refer to the rpm vs torque curve to deduce the torque.

But I'm really not sure what your goals are, or what torque frequency you need to measure, or at what accuracy you need . . .

(hoping I was somewhat helpful . . .)

[robnub]

Torque is force multiplied by distance, where as lift is just a force.

When the blades spin, air resists the blades from spinning. This resistance, which is a counter-torque to the motor, is only semi-proportional to lift force. You can measure that torque, which is what I think you want, but it won't tell you much about the lift force being generated (if that's your goal).

5Nm is quite a bit of torque . . . to get any quality sensor it'll cost several hundred, and even then I'm not sure how you can mount it on a helicopter (with all that wiring) without it destabilizing flight.

Here is another idea . . . remove the motor from the helicopter and characterize it. That basically means determine the torque vs rpm curve at your required voltage. Basically the higher the torque, the lower the rpm.

Then put the motor back on the heli, and use an encoder to measure the rpm in real time. You can then refer to the rpm vs torque curve to deduce the torque.

But I'm really not sure what your goals are, or what torque frequency you need to measure, or at what accuracy you need . . .

(hoping I was somewhat helpful . . .)

Thanks! That was very helpful!

Maybe I should explain myself more clearly:
I am trying to build an Unmanned RC Heli by implementing control systems using external circuitry. However, this system is only unmanned about the yaw axis. I just want to find out how to measure the tail rotor motor's torque as it moves about the  yaw axis (the main rotor is a constant here as I do not need to know the force required to lift it up since it will be placed onto a "lazy susan" a.k.a turntable and will be tied down to it)

Hopefully this provides a better perspective of what I am doing

[Admin]

Just so I fully understand, one more question . . .

You want to know the rotation torque about the z-axis of the main body as created by the tail rotor multiplied by the tail length, right?

[robnub]

Just so I fully understand, one more question . . .

You want to know the rotation torque about the z-axis of the main body as created by the tail rotor multiplied by the tail length, right?

Spot on!

[Admin]

Ok here is another idea . . .

You have the heli on a lazy susan, meaning it'll rotate freely about the z axis.

Get a force sensor, and mount it firmly. It should be located opposite the tail rotor.

So when the tail rotor turns on, it pushes the tail against the force sensor. (you may need to tape the rotor motor onto the force sensor so it doesn't bounce around)

Take that force and multiply it by the distance between the tail rotor and the z axis of the heli. That'll give you torque.

The advantage of the force sensor is that you do not need to modify the heli to make the measurement, and force sensors are a heck lot cheaper. (have a look and see if that works for you or not)

[Soeren]

Hi,

Wouldn't it be just as good mounting the force cell to the turn table so it could be used for other choppers as well?
As I see it, there should be no difference between the forces acting on the heli and the forces acting on the turn table (except perhaps distance, but that needs to be factored in anyway).

[robnub]

My idea of this project is to get a photo-reflective sensor and place it on the tail rotor to measure the RPM of the tail rotor. With this, I am able to work out the amount of torque by measuring the power of the motor as well. Since I would require the speed of the tail motor as an input to my control system as well. Any suggestions?

[Admin]

Out of curiosity, why do you need to know torque from the tail rotor?

[robnub]

Out of curiosity, why do you need to know torque from the tail rotor?

Its part of my project specification :X

[Admin]

Out of curiosity, why do you need to know torque from the tail rotor?

Its part of my project specification :X

geez, one of those things . . .

My idea of this project is to get a photo-reflective sensor and place it on the tail rotor to measure the RPM of the tail rotor. With this, I am able to work out the amount of torque by measuring the power of the motor as well. Since I would require the speed of the tail motor as an input to my control system as well.

The thing with RPMs is that it's only loosely related to torque. If you only need to measure the torque at steady state, it'll work. But if RPM's are constantly changing, such as with a control algorithm in a changing environment, you'll never reach steady state.

[robnub]

Out of curiosity, why do you need to know torque from the tail rotor?

Its part of my project specification :X

geez, one of those things . . .

My idea of this project is to get a photo-reflective sensor and place it on the tail rotor to measure the RPM of the tail rotor. With this, I am able to work out the amount of torque by measuring the power of the motor as well. Since I would require the speed of the tail motor as an input to my control system as well.

The thing with RPMs is that it's only loosely related to torque. If you only need to measure the torque at steady state, it'll work. But if RPM's are constantly changing, such as with a control algorithm in a changing environment, you'll never reach steady state.

Yep, you made a very good point there. Thanks for the heads up. You recommended fixing a force sensor at the opposite end of the tail rotor. May I know how that works?

[Admin]

Ok picture this . . . the helicopter is on a lazy susan, so it rotates when torque is applied by the tail rotor.

Now, take your hand and put it behind the tail rotor. When the rotor spins, it applies a force to your hand (pushing your hand back). Now instead of your hand, mount a force sensor there. Multiply the force it reads by the distance between the tail rotor and z-axis of the lazy susan.

Soeren also recommended, in so many words, to mount a torque sensor between the table and the z-axis of the helicopter. My issue with this is that if the z-axis location is not the same as the center of mass (is it?), you might damage the torque sensor by applying torque in the wrong direction. But it's still an option if it's cheaper than the force sensor using the above technique.

Now, I didn't do the math to see if these are within spec, but here are some cheap force sensors to get you started:
http://www.trossenrobotics.com/p/phidgets-force-sensor.aspx
http://www.newark.com/jsp/search/productdetail.jsp?SKU=25R1227&CMP=AFC-GB100000001

[robnub]

Ok picture this . . . the helicopter is on a lazy susan, so it rotates when torque is applied by the tail rotor.

Now, take your hand and put it behind the tail rotor. When the rotor spins, it applies a force to your hand (pushing your hand back). Now instead of your hand, mount a force sensor there. Multiply the force it reads by the distance between the tail rotor and z-axis of the lazy susan.

Soeren also recommended, in so many words, to mount a torque sensor between the table and the z-axis of the helicopter. My issue with this is that if the z-axis location is not the same as the center of mass (is it?), you might damage the torque sensor by applying torque in the wrong direction. But it's still an option if it's cheaper than the force sensor using the above technique.

Now, I didn't do the math to see if these are within spec, but here are some cheap force sensors to get you started:
http://www.trossenrobotics.com/p/phidgets-force-sensor.aspx
http://www.newark.com/jsp/search/productdetail.jsp?SKU=25R1227&CMP=AFC-GB100000001

This method requires 2 force sensors so I can measure the clockwise and counter-clockwise yaw force right? What if I use a tension/compression load cell and stick it to the lazy susan?
I researched on torque sensors and find that they are way way way too expensive, so I don't think it is an option at all

[Admin]

This method requires 2 force sensors so I can measure the clockwise and counter-clockwise yaw force right?

If you physically attach the sensor to the motor of the rotor, it'll work for both torque directions (ie you only need one force sensor). The sensors I linked to can't be mounted easily as you point out, but there are more expensive ones out there that can. In the end you just got to search around and compare your options.

Oh, and I'm sure you'll realize this at some point but . . . you gotta mount it as to not interfere with airflow. You don't have to mount it at the rotor location, you can mount it at any point on the tail of the heli (force increases but length decreases, so the torque is the same).

What if I use a tension/compression load cell and stick it to the lazy susan?

I don't understand what you mean . . . (ie I don't see how that measures torque )

[robnub]

This method requires 2 force sensors so I can measure the clockwise and counter-clockwise yaw force right?

If you physically attach the sensor to the motor of the rotor, it'll work for both torque directions (ie you only need one force sensor). The sensors I linked to can't be mounted easily as you point out, but there are more expensive ones out there that can. In the end you just got to search around and compare your options.

Oh, and I'm sure you'll realize this at some point but . . . you gotta mount it as to not interfere with airflow. You don't have to mount it at the rotor location, you can mount it at any point on the tail of the heli (force increases but length decreases, so the torque is the same).

What if I use a tension/compression load cell and stick it to the lazy susan?

I don't understand what you mean . . . (ie I don't see how that measures torque )

I don't understand what you mean by physically attaching the sensor to the motor 

The tension/compression load cell can measure the force applied on the lazy susan. I thought this might be better..

[Admin]

I don't understand what you mean by physically attaching the sensor to the motor 

You take the sensor, and using screws, it is fixed to some point of the heli tail (such as the motor of the rotor).
(I'm probably not explaining myself right, but you can't just let the tail bounce up against the sensor, it needs to be firmly attached)

What if I use a tension/compression load cell and stick it to the lazy susan?

I don't understand what you mean . . . (ie I don't see how that measures torque )

The tension/compression load cell can measure the force applied on the lazy susan. I thought this might be better..

ummmm but you want to measure the torque about the z-axis, not the forces applied to the lazy susan . . . or maybe I'm not understanding the question? where on the lazy susan would you mount the sensor?


(I think both of us are too lazy to draw diagrams that would easily clear this all up lol)

[robnub]

I don't understand what you mean by physically attaching the sensor to the motor 

You take the sensor, and using screws, it is fixed to some point of the heli tail (such as the motor of the rotor).
(I'm probably not explaining myself right, but you can't just let the tail bounce up against the sensor, it needs to be firmly attached)

What if I use a tension/compression load cell and stick it to the lazy susan?

I don't understand what you mean . . . (ie I don't see how that measures torque )

The tension/compression load cell can measure the force applied on the lazy susan. I thought this might be better..

ummmm but you want to measure the torque about the z-axis, not the forces applied to the lazy susan . . . or maybe I'm not understanding the question? where on the lazy susan would you mount the sensor?


(I think both of us are too lazy to draw diagrams that would easily clear this all up lol)

Maybe this would help

If i can attach the sensor to the bearing of the lazy susan, I am able to get the torque by multiplying the distance from the center of mass and the force from the sensor.

[Admin]

That won't work.

Attached is what will . . . a clamp is mounted onto the lazy susan (SL) with a rod coming out. The heli is mounted firmly onto the SL. The rod coming out the clamp will push against the force sensor (blue arrow), and the force sensor is mounted onto something firm so it doesn't move (grey blob).

You can either use one rod, where it's screwed into the force sensor. Or if you can't find a force sensor that allows this, you can use two rods parallel with each other, with two push-only force sensors, one for clockwise the other for ccwise. The red line represents the moment arm, from the center of the LS to the clamp. Multiply that distance by the force to get torque applied by the tail rotor. The heli will rotate about the center of mass when it's in flight, so that should be located above the center of the LS.

Hope that helps?

[robnub]

That won't work.

Attached is what will . . . a clamp is mounted onto the lazy susan (SL) with a rod coming out. The heli is mounted firmly onto the SL. The rod coming out the clamp will push against the force sensor (blue arrow), and the force sensor is mounted onto something firm so it doesn't move (grey blob).

You can either use one rod, where it's screwed into the force sensor. Or if you can't find a force sensor that allows this, you can use two rods parallel with each other, with two push-only force sensors, one for clockwise the other for ccwise. The red line represents the moment arm, from the center of the LS to the clamp. Multiply that distance by the force to get torque applied by the tail rotor. The heli will rotate about the center of mass when it's in flight, so that should be located above the center of the LS.

Hope that helps?

Understand what you mean. However, I will want the LS to rotate freely up to 360 degrees. Wouldnt the rod limit the movement of it?

[Admin]

. . . you can't measure force nor torque if it's allowed to move freely . . .

I think you need to sit down with your prof and ask exactly what the heck he needs you to measure and why

I originally thought you wanted torque so that you can spec out motors or know the internal forces within the heli, but it appears you're only really interested in measuring control . . . right?

As you are designing a control algorithm (right?), there are other ways to measure if it's working. A better way to go about it is to measure response rate. Lets say it's controlled to rotate 30 degrees from it's current angle, how fast does it go to the new angle? Does it overshoot or undershoot? In a way it's related to torque, but it's likely not the torque that you were really interested in, right? If that's the case, you just need a quadrature encoder attached to the LS.


edit: What's your engineering background? (so I know how to word my responses . . . I assumed you were a mechanical engineer but now I realize you aren't lol)

[robnub]

. . . you can't measure force nor torque if it's allowed to move freely . . .

I think you need to sit down with your prof and ask exactly what the heck he needs you to measure and why

I originally thought you wanted torque so that you can spec out motors or know the internal forces within the heli, but it appears you're only really interested in measuring control . . . right?

As you are designing a control algorithm (right?), there are other ways to measure if it's working. A better way to go about it is to measure response rate. Lets say it's controlled to rotate 30 degrees from it's current angle, how fast does it go to the new angle? Does it overshoot or undershoot? In a way it's related to torque, but it's likely not the torque that you were really interested in, right? If that's the case, you just need a quadrature encoder attached to the LS.


edit: What's your engineering background? (so I know how to word my responses . . . I assumed you were a mechanical engineer but now I realize you aren't lol)

Alright. Just clarified with my prof and this is what he wants. Basically, I need to measure the displacement angle about the yaw axis. Like what you said, if I want it to rotate 30 degrees from the current angle, I have to design a control algorithm which will allow it to get to that angle (without much oscillation and delay).
Pardon me, but I am not a mechanical engineer :X just an electrical engineer

[Admin]

Alright, in that case, just throw an encoder onto the LS and you're good to go.

Basically you implement a PID loop and play around with the P, I, and D constants until it does what you want. Keep in mind though that a finely tuned PID loop goes whack when there's strong wind coming in from random directions. When it comes to small flying vehicles in wind, it often shouldn't even point in the direction you want it to travel! Look up papers that say stuff like 'control of a micro-auv'.

You'll have to get a big fan to simulate winds and see how your control system responds to major disturbances.

ps - If you use WebbotLib, it has easy to use libraries for PID and quadrature encoders.