Reflections

 


There were a number of problems we stumbled upon while working on this project, but controlling the servos was by far the most frustrating one of them all. The Handyboard can control one servo without any difficulty, but to control two it requires a few special add-on programs that must be downloaded from the www.handyboard.com website. The Handy Board would not function properly with both servos attached. Sometimes the Handy Board would just shut down or the servos would not work when sent a command from Interactive C. The solution for this, and many of our other problems, was just to shut everything down and leave it alone for a few hours. Usually the components would work properly at a later time, they just had to cool down. Over the course of our project, however, we finally concluded that it was the servos causing the problems and not the Handy Board.
We purchased the cheapest servos we could find, and this proved to be a mistake. Two of the three we purchased burnt out before the demonstration day, and the third was acting up. There was an Airtronics servo left from a past years project which worked significantly better for us.

We had problems with all the mechanical systems except the turntable. The primary problem with the candy dispenser was that we couldn't get the servos that operated it to work. But, aside from that, it also tended to jam as the candy settled in the hoppers. The sliders worked fine when the candy was not present and with the servos detached, it required a very small force (maybe half an ounce) to push the slider either way. The problem was that the levers used to extend the sliders were connected to the outside edge of the sliders rather than the centers. This put a slight twisting moment on the slider, causing greater friction between it and the walls of the candy dispenser. This friction was not enough to stop it from moving when unloaded, but just enough to cause it to stop when candy was present.

As we mentioned earlier, the solenoid didn't work for the chocolate dispenser, and we were forced to use a gear motor and levers to operate the pump. This ended up working much better than the solenoid, and was much more dependable. It applied a much more consistent force than the solenoid did as well, so it pumped more efficiently. One quirk of soap pumps is that the amount of liquid they pump is proportional to how quickly they're depressed. So, if it's pushed too slowly, it displaces virtually nothing. The motor applied a very constant force and pumped as well or better than expected.

The whipped cream dispenser worked flawlessly with the empty can that we prototyped with, but when we used a full, new can it no longer pulled the tip far enough to squirt whipped cream. Because our prototype can was empty, and had been so frequently used, the spring had worn on the can, reducing the force needed to pull it. After assessing the situation, we found the problem could be remedied if we shortened one of the brackets, putting the can at an angle.

 

 

Problem: Advice:
Servo Control Design for no more than one servo. Or, if you require more, use a demultiplexing circuit to control multiple servos from the same port. *But* beware of the problem below...
Outputs The Handyboard is limited in the number of outputs it has. It may simplify your life to have more sensing than actuation in your project design.
Candy Dispenser Jamming We should have attached the levers to the sliders at the center rather than the outside edge. Better yet, we should have used something like the chocolate dispenser to actuate them rather than servos. It was much more robust.
Solenoids Solenoids only work well for short quick motions, and don't provide much force. Substitute gearmotors where possible.
Driving Table While our system worked quite well for positioning the table, other groups had problems with this issue. The design we chose solved these problesm well. A gear motor or stepper motorcould also have been substituted to provide faster stopping, and implemented with a simple feedback loop for more precise positioning (neither of which were needed for our application).
Whipped Cream Dispensing Testing, testing, testing. Unexpected problems come from every angle, so to ensure the project will work, make sure time is taken to test with each component exactly as it will be when it is used as intended. Then, try to account for all things that could go wrong, because they probably will (and late at night at the last minute, even though they worked before).
Handyboard Resetting If things aren't working, just shut it off and let it be for an hour or two. We spent countless hours trying to find problems that were actually in the handyboard.