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

I won the bid for the encoders on ebay.  Including shipping, the total cost for two encoders is about $35.  Much better than any of the other options I was looking at.  I just hope they work as good as I expect.  Now that I have spent some money on it, I guess I am committed to building this contraption.

I found a better solution for encoders – ebay!!  I’ve never used ebay before, but somebody is selling a pair of Allen Bradley encoders with 1000 pulses/rev that look perfect for my project.  I placed a bid and I will see what happens.

Holy cow!!  Encoders are much more expensive than I ever expected.  The cheapest encoders I can buy are about $90 each from Automation Direct, and I need two.  Need to find a cheaper way to do it.

After extensive searching and calling a few companies for prices, I found another encoder solution.  Digikey has CUI modular encoders that attach to an existing shaft for $30.  They look kind of cheap, but they would probably work fine for this project.  I would have to figure out how to mount a shaft for them to attach to, though.  The $30 encoders are capacitive, however that works.  They also have some that are standard optical encoders for about $37.  When cost is a major factor, these look pretty good.

Inverted Pendulum Design v1 (pdf)

Drew the plans for the system up in CorelDraw.  Each view is drawn full scale on a single page.

The system will consist of a cart that is supported in the back by a linear bearing.  The front of the cart will be supported by a rack and pinion gear system.  A motor on the cart will turn the gear to move the cart.  An encoder will be next to the motor with a smaller spur gear that meshes with the large one on the motor.  This encoder will measure the position of the cart.  An encoder on the other side of the cart will support the pendulum and allow it to swing freely while it measures the position of the pendulum.  The whole system will be mounted on a wooden shelf, which will have space in the front for the cart system and space in the back to mount a power supply and the control system.  It would be nice to have some limit switches to prevent the cart from hitting the ends of the shaft, but those will probably be added later.

Right now my estimate is that the cart will be about 1.5 kg, and the pendulum will be about 0.25 kg.  Picking the right motor is the critical item right now.  I would like something that can travel the entire length of the track in one to two seconds, and has enough torque to rapidly accelerate the cart.

Hopefully, the whole system will be controlled using an NI cRIO unit with a 9505 motor controller unit and 9411 digital input unit to read the pendulum encoder.  I will have to talk to NI to find out if I can somehow borrow the equipment for a few months.  It is a little too expensive to buy for a hobby.  If I can’t borrow the equipment, I have a few other ideas of how the controls could be implemented.

I have been wanting to work with the hardware for an inverted pendulum for a while.  A year or two ago I developed a coding challenge involving an inverted pendulum that I want to run using NI software.  The ideal situation would be to use the NI Control Design and Simulation module to simulate a physical system.  The competition would be run using the simulation, then the winning entry could be transferred to the real hardware.  I was working with some folks at NI on the idea, but we never really got anywhere. 

I started thinking about the challenge again, and decided I needed to have a system that I could use here to run the challenge properly.  At first I thought about borrowing a system from NI, but that didn’t look feasible.  I started looking at prices and figured I could build my own for a few hundred dollars.