|Posted by Klein on October 17, 2011 at 11:20 PM|
Well, we didn't get much done before the school year started in early August, so progress ground to a halt while Ben worked on (non-robotic) school work. However, this past weekend was between academic quarters and part of Fall Break, so Ben got lots of time with Minecraft and I got him to help a little with Octowobble.
The result - we worked out a motor driver bug and finished up the edge sensors.
Driver bug - The motor drivers (LMD18200) weren't consistently driving the motors. They would whine and the wheels would only turn if pushed. Once started they ran well with good speed and power. I reviewed the data sheet and noticed that the example schematics used large caps on the supply lines. A small note in the docs suggested 100 uF per amp of running current. Adding a 330 uF did the trick (overkill, but that's what I had on hand).
Reluctant to disassemble the drivers that I had already built, and inspired by Dale's implementation (Polymax Improvements 2009), I used the extra LMD18200s that I ordered last year and built new freeform driver assemblies, attached directly to the body. Putting the drivers next to the motors reduces line loss, and the large body mass makes for a great heat sink. (I'll post pictures of this soon. It'll make more sense then.)
Finally, we copied the single working edge sensor and glued those in under the shell. I've yet to use them in working code *on* the robot, but off-robot testing and voltmeter readings suggest they will work ok.
Categories: Atlanta Robot Rally 2011