/sci/ - Science & Math » Thread #12589645

967KiB, 3118x2524, combined.jpg

View Same Google iqdb SauceNAO

Anonymous Sat 16 Jan 22:07:05 2021 No.12589645 View Reply Original Report

Quoted By: >>12589665

Hello /sci/
I want to do some measurements on an element that behaves as a non-linear spring. I want to characterize the torque vs deflection curve, and for this I have made a measurement setup. Now, I myself am an electrical engineer and I do not have much experience with mechanical things besides the very basics. After measuring I have ran into some issues, and I just cant seem to figure out what is wrong, and neither could anyone around me.

Pic related is the measurement setup. The elastic element (think of it as a torsion spring) is connected to the test bench through a torque sensor. The other side of the spring is connected to an arm. The other side of the arm is fixed to the test bench using a bolt (and a washer in between to ensure the arm doesnt touch anywhere else). In the middle of the arm a load-cell is located.

Anonymous

View Same Google iqdb SauceNAO IPM.jpg, 28KiB, 300x300

Anonymous Sat 16 Jan 2021 22:11:06 No.12589665 Report

Quoted By:

>>12589645
The problem is, that changing the arm length seems to change the measured torque. Pic related is what I thought is the IPM. Is that correct?
From this I would think that the applied bending moment is equal to the torque generated by the elastic element, which is just the deflection * spring constant. In my reasoning, the load-cell might measure a lower torque because of the slope in the bending moment, however this can be calibrated out with the torque sensor (which has too small of a range for all the torques).
When using the measurement unit I have used also shorter length arms, and not just the torque measured by the load-cell is lower, but also by the force/torque element. I.e. the torque of the element seems to be lower with a smaller arm length but same element deflection.
Note that the load-cell on the pointer measures a practically identical torque as the force sensor, regardless of arm distance, which seems to imply that the bending moment is constant?
Can anyone explain to me why I am getting the results I am getting? A vastly different measured torque if the arm is fixed closer to the elastic element (i.e. the arm is shorter).

Anonymous

Anonymous Sat 16 Jan 2021 22:25:22 No.12589762 Report

Quoted By: >>12590309 >>12590522

regardless of arm length, the same angular deflection should measure the same torque on your sensor. help me understand a few things, anon.

how are you changing the length of your arm? how are you measuring torque? how is the arm set to different deflection angles (it seems like you use the bolt and wingnut on the right of your image, but is it applying a force in the plane of the arm, or is it pinching the plate)? what's the purpose of the load cell, a redundant way to measure forces?

Anonymous

Anonymous Sat 16 Jan 2021 23:16:58 No.12590309 Report

Quoted By: >>12590439

>>12589762
>the same angular deflection should measure the same torque on your sensor. help me understand a few things, anon.
Right, this is what I thought as well.
The torque sensor would measure the same torque regardless of arm length
The load-cell would measure a 'scaled version' because it essentially measures bending moment.
This however, is not what happens. Both the load-cell and torque sensor measure a torque that is dependent on arm-length.

>how are you changing the length of your arm?
If you look at the first image, you see that there's a 'top plate' with cutouts through which the bolt feeds. So the arm length was with a different arm + different top plate.
>how are you measuring torque?
The 'ground truth' torque is measured with a mini40 calibrated force/torque sensor.
>how is the arm set to different deflection angles (it seems like you use the bolt and wingnut on the right of your image, but is it applying a force in the plane of the arm, or is it pinching the plate)?
It is indeed with a bolt through the arm, then through the top plate. It is pinching the arm.
>what's the purpose of the load cell, a redundant way to measure forces?
The 'ground truth' force/torque sensor is accurate, but can not measure the whole torque range that I expected to measure (which is +-10Nm). The sensor only goes up to +-4Nm.

Anonymous

Anonymous Sat 16 Jan 2021 23:41:25 No.12590439 Report

Quoted By: >>12590520 >>12590522

>>12590309
a few thoughts

first, the bolt/wingnut used to set the deflection shouldn't pinch. it should just stop the arm returning to the zero position by applying a force in the tangential direction. the pinching may load the arm in unexpected ways that throw the torque readings off.

second, you are loading a rotating part at the end of an axle that's unsupported. maybe the axle should extend farther and their should be a journal bearing on the top. otherwise you may be torquing your axle such that the torque doesn't point along the axle.

have you considered making a mass balance to apply a known force?

Anonymous

View Same Google iqdb SauceNAO StaticTestSetupWCAD2.png, 3MiB, 2739x1000

Anonymous Sat 16 Jan 2021 23:56:20 No.12590520 Report

Quoted By: >>12590623 >>12592120

>>12590439
>first, the bolt/wingnut used to set the deflection shouldn't pinch. it should just stop the arm returning to the zero position by applying a force in the tangential direction.
This is indeed something I worried about. The first design I made had a slot arrangement that held the arm in a position that way. (see pic)
The problem with this is, that it needs a 'cascading' arrangement to achieve sufficient resolution (1 degree per step), which results in the arm being effectively shorter and longer for even and odd deflection steps respectively.

How would you suggest the arm can be fixed? Or rather go with the mass balance that you suggested?

>otherwise you may be torquing your axle such that the torque doesn't point along the axle.
I have thought about it, but the F/T sensor shows torques and forces in all 6 DOF, and does not significantly seems to show other forces except the relevant torque.

>have you considered making a mass balance to apply a known force?
I have thought about a system to apply a varying weight to the element. In the end I settled on this because it is easier to apply a series of known deflections and measure the resulting torque, than it is to apply known weights and measure the deflection accurately.

Anonymous

Anonymous Sat 16 Jan 2021 23:57:23 No.12590522 Report

Quoted By:

>>12590439
>>12589762
I completely forgot to thank you for your reply by the way. I do appreciate it very much.

Anonymous

Anonymous Sun 17 Jan 2021 00:24:55 No.12590623 Report

Quoted By: >>12591335 >>12592120

>>12590520
this pic seems like a better way to set the deflection.

have you tried replacing the spring element with a non-spring element and applying a known force a known distance from the axle to predict the torque you should measure?

thinking about it more, a balance is probably overkill, but using weights may still be a good idea for troubleshooting and calibration. it might be as easy as putting the unit on its side and hanging weights off the end (this configuration would almost certainly benefit from a bearing, even if the off axis readings seem low)

Anonymous

Anonymous Sun 17 Jan 2021 03:17:10 No.12591335 Report

Quoted By: >>12592120

>>12590623
once again, upon more thought, putting it on its side would make calibration hard because the arm's weight would torque the sensor away from zero

Anonymous

View Same Google iqdb SauceNAO LessChoppyDataset2.png, 29KiB, 840x630

Anonymous Sun 17 Jan 2021 11:29:24 No.12592120 Report

Quoted By:

>>12590623
>this pic seems like a better way to set the deflection.
I have tried to use the already available holes to do another measurement, and it does help produce something more in-line with what I expect. Many thanks for the suggestion and I feel silly for not realizing this myself.
I am however still unsatisfied with the fact that I dont know why the torque differs with distance. Pic related are some measurements that I have done with the device in >>12590520
You can clearly see a 'stair step' pattern from the change in arm length with each odd/even change.
>have you tried replacing the spring element with a non-spring element and applying a known force a known distance from the axle to predict the torque you should measure?
I am not sure if I follow. A non-spring element would mean it would not deflect with (ideally) any force, so you could not predict the torque that way?
>thinking about it more, a balance is probably overkill, but using weights may still be a good idea for troubleshooting and calibration. it might be as easy as putting the unit on its side and hanging weights off the end (this configuration would almost certainly benefit from a bearing, even if the off axis readings seem low)
>>12591335
>upon more thought, putting it on its side would make calibration hard because the arm's weight would torque the sensor away from zero
This indeed was also why I haven't tried this yet. I have made a kind of roller, which could translate the vertical force vector of a hanging weight into a horizontal force vector. The problem would be though to make sure this horizontal force vector remains perpendicular to the arm.

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches