Torque Measurement Question

Looks like unequal force on the bearing during spring loading...If the spring was mounted inside the bearings it shouldn't be a problem...or you could use larger more precise bearings and a non-flexible shaft...but I'm just guessing...

Hi, Solar Eagle. Good to see you here.

Yes, the loading is going to be asymmetric and I am thinking that is the leading cause, but this isn't exactly my field of expertise. It just seems like an awful lot of parasitic friction, if that is the cause.

Well how have you been? I hope you have been doing well...we don't get many visitors here...anymore...

It's easy enough to test by moving the spring next to the force lever, even if a temporary setup, to see if this removes the binding....

I have been good, just busy with work, house, and life.

When you say you don't get many visitors here anymore, do you mean that the forum is stagnating?

...do you mean that the forum is stagnating?..

No not really...still have a lot of the original posters here... just fewer...

Politically, the posting has dropped off due to some members because they got what they wanted,… deserved.

Is the purpose here to balance torque on both strokes?

Are you able to extend the blue shaft a bit past the spring arm and add a bearing to the other side?

A paradigm might be an internal combustion engine: crankshaft with a bearing on only one side of the end connecting rod.

Yes, the return force should be the same as the extension force. At least that is the idea, much like a simple spring.

Nice to have you back…

Doesn’t the spring as it’s stretched, it’s not constant, but it loses tension?

I agree with SE. You may need to move the lever to a mid point between the bearings to equalize the force difference. I would also suggest a roller or sleeve bearing instead of a ball bearing. If you are making a lot of these, sleeve bearing will save some cash, but do have different maintenance issues.

maybe check to make sure the bearings are square with the rod. one might of gotten tweaked a little causing some binding.

Unfortunately, the design does not permit moving the spring, at least not permanently, to the center.

However, I did try that as a test and sure enough, the delta between the two strokes changed considerably.

Thanks for the reply. What would be the advantages of the roller bearing over a ball bearing?

Roller bearing have spread out contact points reducing point pressure. With a ball bearing is this usage you have two points of contact on 2 or 3 balls under pressure. That area of contact is many time greater in a roller bearing.

I know nothing about this sort of thing, but here's my two penn'orth.

If you could use a torsion spring instead of the lever and tension spring, it would improve things.

Or you could add an equal extension to the lever and use two half strength springs.

You could use the same trick with the measurement system.

But, in the real world how will torque be applied to the horizontal shaft to oppose the spring and lever?

Another point of non linear friction would be the spring attachment point.Is the end of the spring and measurement at the attachment points a closed circle,or open ended?If open ended,it could cause the end loop to open slightly when pressure is decreasing and opposite when increasing.

The spring attachment loop could be "winding up""(closing the loop) in one direction,and "unwinding" in the other.

An alternate design could also be used,such as:

A known calibrated weight attached to a rigid lever on the end of the measured shaft might give a more predictable result by having the exact amount of resistance in both directions.Gravity is very reliable.It would only be required to lift the weight to a precise 90 degree angle to get a maximum measurement.Rotating further would give a decreasing value,and so would lowering it.This change could determine the exact level and angle for maximum torque.That way you could have a full 180 degree measurement without reversing the direction of rotation.The measurements should be equal at equal negative and positive angles.If not,something else is causing a non linear measurement.

You could make the resistance variable by making the weight precisely adjustable(sliding or threaded,such as an all thread rod) on the length the arm holding it,using a micrometer or caliper.This would give the possibility of testing the yield point of the shaft material if you wanted to go that far.

Please give feedback.

This would eliminate some of the variables.

Similar to this lever setup,except with a variable length lever,not fixed points.

You could just put adjustable torsion springs on either side of the shaft...

The difference in the curves winding up and winding down is the sum of the frictional force. If the friction winding up and winding down are equal, the true curve should be midway between the two curves.

Moving the spring to the midpoint of the shaft puts both forces (the spring force and the applied force) in a plane perpendicular to the shaft, and there will be no counter forces applied by the two bearings. Bearings are usually designed for radial loads, axial loads, or both.

I'm guessing that your bearings are not designed for radial loads (perpendicular to the axle).

https://www.lily-bearing.com/resources/blog/understanding-bearing-loads/

I am thinking you are right. I'm finding I know less about bearings than I imagined and thank you for the link. I need the education.

The bearings we are using are supposed to be radial load bearings, but may be undersized, despite not exceeding their static load limits. We did find that changing from unspecified rated bearings to ABEC-7 bearings helped.

This reply should contain pictures, but, I've tried all the tricks I know to include them, and, failed miserably......I'm going to try to post the pics separately.

"Moving the spring to the midpoint of the shaft puts both forces (the spring force and the applied force) in a plane perpendicular to the shaft, and there will be no counter forces applied by the two bearings."

I'm not sure I understand: if you move the spring to the mid point of shaft:

Pic 1.

The force of the spring and the measuring force are both in the same direction, and, the bearings will need to supply radial force in opposition.

Even if the measuring lever is moved to the other side of the beam:

Pic 2.

the forces of the spring and measurement, form a turning force which also put radial force on the bearings.

The only way to eliminate any radial forces on the bearings is to place the measuring device directly on the end of the spring:

Pic 3.

But this is equivalent to simply measuring the spring.

The only real way to assess the performance of the "system" (spring, lever beam and bearings), is to use a measurement system which exactly mimics the way in which the real life system applies a counter force to the spring and lever.

The force vectors also changes on the spring as the pivot arm rotates. It may be nominal, but it changes, especially if you want it to consistently cancel out.

Kinematics is not my field but I trust the telemetry. The difference in torque curves tells me that the analysis has not accounted for an additional, directional force. Have you tried changing the orientation of the mechanism? I suspect gravity assists motion in one direction as a mass is raised and lowered. Friction opposes motion in either direction. The gravity of Earth points in only one direction. Might the partial raising of the spring account for this difference?

Great to see you back.

Thanks! Great to see my old friends here, too!

Yes, the orientation is not the issue. These springs have a maximum torque of over 40 in•lbs and the vertical shaft is 3.25" tall.

There definitely is some form of parasitic resistance and I am thinking of trying to build a couple of simple test jigs to prove that, perhaps a DOE if I can determine what factors to play with. The issue is that the actual product does not have a lot of latitude for changes, but I need to understand the cause first. Then I can look into what changes are possible.

From another guy who joined in 2006: I think SE's mention of using one or more torsion springs is a good idea. A torsion spring with an ID just a bit larger than the shaft OD will take up very little space, and should be more linear than the straight spring and lever.

Looks like my WAG is different than the rest of you! There is hysteresis coming from somewhere! I suspect the difference may be caused by bending and twisting of the blue shaft, but they should act as very stiff springs and return the same energy that went in. Really WAGing it now! So, perhaps the energy is consumed in the bearings by the slight bending of the blue shaft? Thus, would spherical bearings help? Can you minimize the distance between the 2 bearings on the blue shaft? (Even this amount of thinking is more than my brain likes when it's bedtime!!)

The whole system is more complicated than the simple drawing I posted, but the actual design is proprietary to a customer, so I am trying not to cross lines here.

However, your WAG is most likely on the mark when you said the energy must be going somewhere. I need to break the problem down into smaller testable chunks.

In our learning/schooling we usually make "simplifying assumptions." Things like frictionless pulleys. To determine where the energy is going, we have to consider anyplace friction can occur. Such as where the springs attach to the levers; put bearings here? Do the spring coils touch each other--thus sliding friction.

You did not say what the value of the delta was. If within an order of magnitude, I would say to forget it--except for academic curiosity.

The original graph I posted has the forces applied on the vertical axis axis.

Have you tried swapping the return spring and the force positions, then testing again? That is to say, leave the bellcrank mechanism as it is, move the spring to the (now) force bar, and pull on the (now) spring bar.

Not sure just what the new test result might reveal, but will possibly give a clue.

I thought hysteresis losses were an unavoidable part of virtually every mechanical system.

I don't subscribe to the notion that torsion springs would be more linear. In my experience with winding torsion springs on garage doors, there seems to be a lot of friction from the coils rubbing on each other.

What kind of lubrication are you using in the bearings? To minimize the effects, I would use a light oil rather than grease.

Assuming the center of gravity is below the axis of rotation, you are stretching the spring plus lifting the center of gravity. On return you get a gravity assist.

Rotate the entire mechanism 90 degrees until the lever is vertical,and use a weight and gravity to apply torque instead of a spring.Gravity is very reliable and predictable.On a very small scale,gravity will decrease as the height of the lever from the center of the Earth changes,and even smaller changes due to our orbital distance from the moon and sun changes,and our galactic position in our local galactic cluster and super cluster changes and as possible changes on influences outside of our visible universe.Just how accurate does it have to be without getting into quantum mechanics?

Since you did not imply that cost was a limiting factor,consider this method:

Use active radial magnetic bearings,which have no,or nearly no friction.

Remember,you will never get out the exact amount of energy you put into any system,that would imply perpetual motion.There will always be losses.

Here is a link to magnetic bearings.

https://www.engineeringclicks.com/magnetic-bearing/