Your take on CAN...

[Levansic]

I took an informal poll at the LA regional this weekend, and ~35% of the teams I talked with we're using CAN in some form. Generally, more established teams were more likely to use it, while young or rookie teams were far less likely to try.

There were some interesting usage philosophies between the teams. All of the teams that I talked with that used CAN, were also using PWM on their robots in some capacity. Usually, but not always, the PWM controllers on robots using CAN were Victors.

Specifically, team 294 used CAN control "everywhere that matters", meaning drive, bridge tipper, and shooter. They use PWM with Victors on their ball elevator, to save weight. Conversely, I was told by a student on 330, that they use PWM for their tank drive, and CAN for their bridge tipper and shooter. They use a black jaguar as a serial to CAN bridge, and felt that using CAN for those parts was essential, as they took advantage of the closed loop modes.

I heard some complaints from students about perceived unreliability of CAN, but still saw it employed on their robot. My observation was that the rail at the center of the field, or more properly the impacts with that rail, caused the majority of component and system failures, regardless of which control method was chosen.

One other observation I had was that poor design choices would lead teams on a quixotic search for scapegoat components. I saw quite a few 4-wheel tank drives that could go forward and back just fine, but would stall their CIM motors when attempting to turn. No doubt, it may have worked adequately on linoleum at home, but not on the competition carpet. Low gear ratios coupled with high-friction contact points at the corners colluded to make their drive systems into current sinks, causing system-wide brown-outs. Unfortunately, this is probably how some of the Jaguar naysayers got started.