Picking things back up

[webby2]

My input is via a long "T" handle Allen wrench.  When I start spinning the system I am putting in a lot of work and using one motion or path of input.  When the system is spinning up at some point I shift from making a circular motion with my hand at an almost perpendicular relationship to the test part to a centered but angled motion.

The Allen wrench is not fixed to the test part but is pushing against one surface of the test-part so as to cause a rotation of that test part, which it is free to rotate on the shaft it is on.

When I have it spinning up at some value and the test-part starts to lift up the shaft and I am using the angled motion my input feels like it drops way off and as I try to spin faster I need to add a larger resistance to the output shaft to stop it from spinning faster which would lift the test-part up and off of the shaft.  Done that a few times now.

The whole test-bed and mount table shake a lot since it is out of balance due to the weight on the test-part.

Since all this is being done with Mr. Hand It does not mean much of anything, but it is only an interesting observation.

[webby2]

I think I understand what Mr. Hand is doing.

Initially Mr. Hand is spinning up the weight, then CP is building and creating its own input force, this new force quickly outperforms the weight.  After this point Mr. Hand is no longer primarily supplying a force of acceleration to the weight but rather keeping the CP and the system in sync.

[webby2]

I do not like using a velocity related force as a drive force, that is I do not want to use CP as the output force source because that means the faster I spin the system the more force the system puts out.  I would rather have the output of the system dependent on the amount of force I put in so that it all can be controlled for the application that it is designed for independent of rate of rotation.

I have built test-beds in the past that I could use to replace Mr. Hand with the current test-part and test-bed, the basic need I understand.  While I was going over design ideas and I was going back to an old test-bed I built, I realized that the issue I ran into with that test-bed is exactly what is happening within this current setup.

It all has to do with the difference between torque and leverage and the work transfer function that each uses.

A simple solution I came up with for that old test-bed, I used, but I did not like so much because it limited my access to some of the interactions I wanted to play with.  Looking at the setup on my screen I see that there are other solutions, one of which I used but for a completely different thing, and one that is almost easier than the one I used.

I made this other test-bed that allowed me to take a non-circular motion and convert it into a circular motion, just not the way it is usually done.

I am going to combine these things from the past into a single device for the input section of my current test-bed and hopefully I will be able to control the input force independent of rotational velocity, but at the least it will allow for a synchronized input to output timing without interfering.

[webby2]

I opted for a chain and sprocket for part of this, I thought it would be the most economical way.  Well when you make some silly mistakes it gets to be frustrating when what should be a nice smooth movement ends up being hard and messy.

When using a chain and sprocket it is important to chamfer the edges of the sprocket, not just make the sprocket 0.5mm thinner than the chain, then I was off by just a little bit on shaft spacing so I had to add a 1\2 link into the chain, and I am using plastic printed sprockets on printed shafts going through metal bearings and using a metal chain,, guess what bends a little bit,, guess what has a little run-out,, silly mistakes.

[webby2]

Getting frustrated with run-out and such.

I do not want to spend a lot of money buying sprockets and stuff so I have to come up with a way to print my parts and get them close.

My chain goes to tight in one section of motion and then too loose but when you are trying to keep things lined up that does not work very well.

[webby2]

I changed the way I was printing the sprockets and it looks nice, however I am going to be transferring a lot of force so I thought it would be a good idea to thread some steel rod through the shafts.  That sounds like a good idea but the issue is that it allows the shafts to warp a little bit and then the run-out is huge,, so I guess I will need to just plan on less force for now.

[webby2]

I need to design in a point of failure that will be easy to replace.  I need this due to the limited force that my plastic parts will be able to handle and I would not want to break a part in the middle of everything that would require me to disassemble the whole thing to replace.

I also need to be able to amplify my input work enough to more than compensate for my frictional losses and show a gain.

[webby2]

This is getting somewhat intriguing.

I have the system almost complete, the new part, and I made 2 crude alignment type of test parts to see about the angles and stuff.  I should have a change in the force function depending upon the angles.

As I am playing with this there are a few positions I can set things into that behave a little different, that is in one setup the reaction to the"input" force drops and the system will make a fast change so I am thinking that at that relationship then if I were to keep shifting the reactionary "input" force, which has dropped, the system would keep shifting.  This is close to what I am trying to do.

Another one would be using CP where the CP hold force is converted into another output work function.

[webby2]

I should also mention that I made these parts fairly weak so I could see one of the forces I am trying to use.

[webby2]

My chain and sprockets have already failed in keeping things in sync.  I knew they would not hold up for a long time but I thought that I would get a few weeks of testing before they went bad.

What I was playing with was making the motions and stuff that I wanted, not sure on the force values but some of the methods of interacting did work, again not sure on the force values.

[webby2]

so my primary test-bed has proven many times and in different ways that my larger diameter wheel can drive a smaller diameter wheel at a 1:1 rotation and a 1:1 force for a 1:1 work transfer, it goes the other way as well.

This is all accomplished by some simple but maybe a little clever design where the hidden distance is allowed to exist.  Think about this way, if I have a 100mm radius turning an 80mm radius I would need some kind of gear that has that 5:4 ratio, mine is a virtual gear but it is real enough to perform the function needed.

My new test-part is doing the same kind of thing but using a real gear instead and I am trying to design it so that it is a poor gear design, I am willing to throw away some work to losses.

I am shooting for basically a 100mm diameter turning an 80mm diameter at a 1:1 rotation and with a 1:1 force but minus those losses I am designing into the system.

I am going this rout since my chain and sprockets are not doing so well.

[webby2]

To be a little clearer, what I think I can do is throw away the change in distance so that my input force on the 100mm is the same force out on the 80mm,,, say 1N of force in and 1N of force out at a 1:1 rotation.

[webby2]

Throw away is not really correct but it kind of shares what I am not overly concerned with.

So, my test-bed allows me to transfer the work, via the perimeter of the wheels, through the system at a 1:1 rotation and a 1:1 force as seen by each wheel.  I use the same size drive disc on each wheel and measure the same force and the same change in angle.

None of that should be a surprise since our math requires that this be the case as far as final values go.

I am mentioning this again because the perimeters are not the same and so there is a difference of distance traveled, this difference is made up for with the work transfer function of the designed interaction.  This WTF is another exchange\conversion that happens internally, something I now have access to.

[webby2]

I have a build error that I did not know about.  I have 8 sub-systems with a part that connects to all of them and they all are supposed to be in sync with their motions and stuff, well somehow either a design error or a slicing error or a Freecad error, the systems are sort of indexed so that they all only line up when the parts are put together in one orientation and that still does not work nicely.  The system is supposed to allow for a free rotation between all the parts but as the system rotates it starts to bind and that spring loads the actual parts so there are 4 positions that it wants to stop in.  This is not a good thing for me.

This test-part was setup so that I could use it for testing 2 different thoughts on how to transfer the work.

One way would require that I flip the input and output functions at the appropriate time while maintaining a constant reactionary condition.  I could see Bessler using something like this for his wheel since Gravity is directional it would provide a means to facilitate the flip and then he might be able to use the constant reactionary condition as his actual output.

Another way, which I am preferring to use, has a constant input and constant output and no flipping required.  It needs a little better relational control and I think I should use a couple of limiters.

I have made my parts somewhat weak so that I can see the forces acting upon those parts, they bend and flex and it is this bending that I am using as a visual guide for the build.  Obviously the final build parts will be stronger.

Another reason I chose to use 8 sub-systems is because each one can only pass a smallish amount of torque, 8 working together means I could pass 8 times as much before failure, hence the chain and sprockets to not only pass the output via the chain but the chain was supposed to keep all the parts in sync.

What I am seeing from my flawed test-bed, going for the constant setup, is that the forces and directions are doing what I was expecting.  I have taken what should be an external rotation and stopped it and turned it into a virtual rotation.

I still need to design the feedback system that is needed between to parts, one moves a little relative and the other moves more, this work is the feedback I need to build for.

[webby2]

This should not need to be said, but, no engineer would design one of my systems to work the way I am using it