17-03-2010 23:06

Chris is me

Re: pic: 2791's Brand New Arm

So this arm didn't work more than once. The metal support failed under load combined with falling a few feet. The largest triangle's outer edge (on the left) bowed out on both sides, I believe.

Does anyone have any suggestions for what to change in a redesign? I'm wondering if thicker material (the metal might have been 1/4" AL, so 3/8" would help), smaller triangles, and more standoffs throughout the arm (but not in the way of the piston) would help enough. I'm also looking into flanges.

Disclaimer: I didn't CAD this, I'm not that smart yet.

17-03-2010 23:31

s_forbes

Re: pic: 2791's Brand New Arm

Flanges will likely prevent the buckling problem you ran into (provided you have a way to bend it... 1/4" is pretty thick). Also, does it need such large speed holes in the supports? How close are you guys to the weight limit?

A picture of it attached to the robot would be helpful, too!

17-03-2010 23:35

Akash Rastogi

Re: pic: 2791's Brand New Arm

As I suggested before (might as well write it down this time):

Smaller truss pattern
More space along the edges of the material and in between triangles
Flanges are friends, not food
1/2" radius (or w/e you prefer) around standoffs
More standoffs (a pair of standoffs after each set of two triangles)

17-03-2010 23:57

James Tonthat

Re: pic: 2791's Brand New Arm

Flanges and more material, not more thickness.

18-03-2010 00:23

Ian Curtis

Re: pic: 2791's Brand New Arm

Flanges are where you'll get the most bang for your buck. Once it starts bending with no flange, there's not much keeping it from continuing to bend. With a flange there is more material that needs to fail before that bending can continue.

Mathematically, bending depends on the moment of inertia, all other things being held equal. The moment of inertia of a rectangular section varies linearly with the length of the edge of rectangle parallel to the direction of the moment, and with the *cube* of the length of the edge of the rectangle perpendicular to the moment. That probably doesn't make a lot of sense*, but it gives you a *huge* increase in strength for a small flange that runs along the edge of you aluminum arm.

*Because I can't explain it well, not because it isn't true.

18-03-2010 00:28

Andrew Schreiber

Re: pic: 2791's Brand New Arm

Just a personal off topic request, could someone who CAN explain this well (or at least point me in the direction of the relevant equations) elaborate on this. I am an ignorant CS major but I am REALLY curious.

Sorry for threadjacking Chris.

18-03-2010 01:05

Ian Curtis

Re: pic: 2791's Brand New Arm

Wikipedia does a good job with the math under this heading: "Second moment of area for various cross sections"

I tried writing more in depth descriptions, but I can't get something down that uses the math.

In physical terms, think of it like holding 2791's arm with one end in a vice. If you were to press down on in, it would take a lot of effort to bend it. However, to bend it away/towards you would take significantly less, right? It seems from Chris's post that it was in this away/towards bend that the air failed. It turns out that adding just a little material in the plane that bending is occurring gives you much more strength.

TL;DR 2791 built an arm well prepared for tension/compression, but not strong enough for bending.

P.S. This all assumes I read Chris's original post correctly. By bowing, you meant that head on the arm used to look like || and now looks like ()... right?

18-03-2010 01:21

Tom Line

Re: pic: 2791's Brand New Arm

We are using a billet bar of 3/4" x 1" solid aluminum for the bottom "leg" of ours, about 15 inches long. The upper leg is 1/8 wall 1x2" tube aluminum. Our "wings" are .5" diameter solid aluminum rod bent slightly to get a better lock on the vertical tube.