Building an Inverted Pendulum System The Ups and Downs of Defying Gravity…

Today I finished cutting and drilling all the parts of the cart.  I started with a new piece for the arm that connects the front section to the bearing shaft.  The old one was pretty rough on one side since we cut it from an angled piece.  Since I had to attach stuff to the top and bottom of the arm, I decided to cut a new arm from the pieces I bought from Online Metals, which gives me nice flush mounting on both sides.  Drilled a bunch more holes with the drill press and discovered a few things:

• It is much easier to see the spot punched in the metal if you remove the paper template you used for punching.
• A small drill bit will usually “pop” into the punched spot, making things line up very nicely.
• A large drill bit has a mind of its own and tends to ignore the punched spots, giving me terrible hole alignment.
• Next time, for any hole about 1/4” or larger, I will drill a smaller hole first, then make it larger.
• When drilling a really large hole, start small and step up slowly.  Large steps make the drill chatter.

After drilling several holes that didn’t quite line up, I had to fix them.  My first effort is to just drill a slightly larger hole.  If that doesn’t work, I use a rotary file to enlarge a hole in a selected direction, making it a little more like a slot.

The final cart assembly looks pretty good.  I mounted all the different components onto the cart.  I put the broken encoder in the position for the pendulum so I can test the pendulum without breaking another encoder.  After I am comfortable with the pendulum behavior I will switch to the working encoder.  I used a couple of cable ties to keep the wires from bending and breaking.  Eventually they will be replaced with a cable carrier.  Funny detail – every set of screws in the cart is a different size and type.  It has metric (M3, M6), english (8-32, 10-32, 4-40), pan head, button head, flat head, slotted and phillips.

Completed cart - top view

Complete cart - bottom view

The next step is to attach the bearing shaft to the shelf at the right height so that the cart is horizontal and the pendulum will swing vertically.  Getting the height right could be tricky.  I also have to figure out how to make the pendulum so the hole is centered and straight.  Then I will be all set for a control system.  I hope that sbRIO board gets repaired soon.

Drilled holes in shelf and attached rack gear.  It looks pretty good.  Now I just need to find time to finish building the cart and get it all mounted on the shelf.

The first picture is the entire shelf with the rack mounted flush to the front edge.  The second picture shows a closeup where the rack gear is attached to the aluminum bar using bolts, and the aluminum bar is attached to the shelf using wood screws.

Shelf with rack attached

Rack Closeup

This picture gives you an idea of what the whole system will look like once the cart is completely assembled and the linear bearing is mounted to the shelf.  Of course, the second encoder will be moved to the new bracket on top of the cart once I make it.

Whole system layout

I did a few calculations, and the bearing loads on the encoder from the swinging pendulum are going to exceed the maximum loads specified for the encoder.  I don’t want to destroy the encoder, so I decided to redesign the cart.  I added a bracket on top of the cart that extends out beyond the gear.  The encoder will be mounted on this bracket, and the pendulum will be almost flush to the bracket.  This will minimize the loads on the encoder shaft.  Time for some more cutting and drilling, I guess.

Inverted Pendulum Design v2 (pdf)

Replacement encoder arrived today.  Checked it out to make sure everything works.  A and B signals are fine.  Found out there is no Z signal, but I really don’t need it anyway.  It is interesting to look at the guts of the encoder – obviously older technology than the other encoders.  Replaced the short cable with the last piece of my ethernet cable.  This time I was a little smarter and left about two inches of the old wires and just soldered new wires to old wires, then put a heat shrink tubes over the joints.  I coiled up the extra wire inside the encoder case – it was a tight fit.  One amazing thing – the mounting holes in the new encoder are the exact same size and location as the old encoder.  I don’t have to drill any new holes in the brackets to mount it.

Here’s a couple of pictures of what the cart looks like so far.  Haven’t drilled all the holes for the bracket yet, but you can get a pretty good idea of what the final version will look like.

Cart - Top View

Cart - Bottom View

A friend with the proper equipment drilled all the holes in the aluminum bar and attached it to the rack gear.  Now I just need to pick it up.

Won the bid on the new encoder on ebay.  It was a bargain at $18 including shipping.  At this point, I am going to quit trying to get the broken encoder working.  It is still a possibility but not really worth pursuing unless I need a replacement encoder.

The LEDs came in today.  I soldered a resistor onto the positive lead of each LED so they would work properly with a 5 V power supply.  I connected each LED and tried waving it in front of the encoder sensor.  All of the LEDs managed to make the digital signal flicker, so it looks promising.  The problem is that none of them was consistent.  I couldn’t get a good steady signal on both A and B channels at the same time.  I suspect if I drilled a hole in the side of the encoder where the old LED is, I could put a new LED in the same spot and it might work.  The light would have space to spread out and go through the lens and be focused on the sensor.  Drilling through the side of the encoder would be a little risky, though.  I am leaning towards just getting a new encoder if I can get one at a reasonable price.  Most of the ones on ebay are priced higher than I am willing to pay.

Sigh.  Just found out that the sbRIO board from NI is not working and needs to be repaired.  That means it will be a while before I can start controlling the system.