Results of the Design:

Out initial designs called for a conveyor belt to take the cup from the user, deliver it to the drink mixer and then return the cup to the user with the appropriate drink. The conveyor belt design did not change from the original but we did have to change the whole structure of the drink mixer design. Our original intentions were to mount a carousel holding all of the bottles on top of a motor and then have the motor rotate the entire carousel. After trying to implement this idea, we determined that mounting a carousel with full bottles on top of a motor shaft was much more difficult than we originally thought. The motor that we had was strong enough to turn the carousel but we could not figure out a way to securely support all of the weight on top of the shaft. This was a mechanical issue in which we did not possess the means to deal with. We switched to a simpler setup. Instead of having the bottles rotate we decided to keep them stationary and have the motor rotate an array of hoses that were attached to the bottles. This drastically reduced the amount of stress on the motor shaft and we had no problem implementing this new setup.

Our initial plans also called for a control box that would contain the keypad used for drink selection and the LCD display. The idea was that this would keep the controls for the project self contained with only a single ribbon cable coming out of the box. After numerous hours of testing we determined that the ribbon cable was much too noisy for the keypad we were using and we had to mount the keypad directly on the expansion board. After we did this it functioned like it should.

Another problem that we had, which was easily fixed, was that we were doing A to D conversion on the distance sensors all the time to see if there was a cup present. Every time a conversion was finished an interrupt was thrown. We made the mistake of giving these interrupts precedence over the interrupts that we used for the keypad and Nintendo controller when making drink selections. There were so many A to D interrupts that there would be times when the interrupts from the keypad and Nintendo controller were very delayed or skipped. This was fixed by swapping the precedence of interrupts.

These were our three main design modifications. Once solutions to these problems were implemented, the project came together and integration was straight forward.