Sunflower Version 3 Build Report

Version 2 of my 30 lb bot Sunflower competed at Steel Conflict 2 and NERC's Motorama event in February 2003. At those events, Sunflower's overall record was 7-4, coming in fifth at SC2, and 3rd at Motorama.

While I was happy with the performance, I wanted to address the shortcomings which I thought I'd identified in the Version 2 design, and since the bot was pretty beat up anyway, decided to rebuild it from scratch for BotBash.

Analysis of losses:

SC2, vs VD3: After a tremendous collision, Sunflower was flipped upside down; in that position, despite being designed to be functional inverted, it had almost no mobility. VD3 took more damage than Sunflower and had I been able to move inverted I would not have lost that fight.

SC2, vs Omega Sword: Sunflower hit OS and again was flipped upside down. This time we were near one of the pits, and Sunflower skidded right into the pit. Because of the poor performance inverted, there was nothing I could do to avoid it. I should not have driven so recklessly near the pit, but may have been able to recover if the bot was truly invertable.

Motorama, vs Helios: Well, Helios is definitely a tough opponent, and I'm not sure that I'll ever beat it in one-on-one combat, but again invertability played a part as Helios flipped me over early and then pretty much had its way with me.

Motorama, vs JB Johnson: Several things went wrong in this fight. First, SF's drumette lost one of its teeth, making the weapon too unbalanced to be effective. Second, and more devastating, two of the FETS on a victor shorted out from a heavy impact, causing a short circuit that not only fried the victor, but pretty much shut the bot down as well.

The goals of a rebuild, then, are:
  1: make the bot truly invertable
  2: make the drumette sturdier
  3: prevent the victors from shorting out like that (not hard)
  4: do these things without harming its effectiveness or introducing new problems

Why is Sunflower 2 so flippy? There are three reasons. First, its center of gravity is not as low as, for example, Helios, because of the height of its body. Second, it has a relatively small wheel base (the points where its wheels touch the ground make about a 9x10 inch rectangle, which is quite small for a 30 lber). Finally, the drumette spins very fast, and the gyroscopic forces generated tend to make the bot want to flip over counterclockwise.

Nothing can really be done about these to make the bot less flippy while still maintaining the good features, so the solution lies in a truly invertable design; that way, if it flips over I don't really care.

After some thought, I cadded out the following design:

Three wheels per side, with only two on the ground. Because this increased the height of the front of the robot, I was able to shrink the overall dimensions further to 12x12.5 (from 13x13).

Having a pretty complete CAD design means that I know how big and what shape everything should be, so I decided to start out with the pieces of the drive train. The extra complication of the drive train meant that I had to switch from #35 chain to #25 for weight reasons.
After chopping some shafting and going through the tedious process of boring out the sprockets to accomodate the shafts, here's a pile of parts:

Those are 1.25" thick Colson wheels; the fronts are 2.5" diameter, and the rears are 3.5" diameter. The shafts are mostly 0.5", but a few are 0.375". This is a problem, though: The holes in the wheels are 0.906".
Fortunately, it's not too hard to make shaft adapters if you have some small machine tools. If you don't you can buy such things pre-made from various online sellers, which I have done in the past. This time around, I decided to make them...
Starting with some 1" diameter aluminum rod chopped into pieces, I turned them down to size on my lathe and drilled the appropriate size holes through them. Then I popped them onto the mill to cut the outside keyways.
The inner keyway of the spacers was made with a standard broach kit. But here's a puzzle: how to key the inside of a colson? The only thing I could think of was to make a custom bushing to do the job: take a blank piece of aluminum rod turned down to 0.9 inches and cut a deep groove in it for the broach to pass through.
With this done, cutting keyways in colsons is a breeze.
With those parts cut, the various drive shafts could be assembled, finishing that phase of the project.
The top and bottom plates for this version of Sunflower are essentially identical -- the first time that has been true for one of my robots.

Each plate is made of .08 aluminum.  After cutting the outline and wheel holes, I used a scribe to draw the locations of the frame components then drilled mounting holes.  The location of the holes is not that critical, which saves a lot of time.
My friend Pete (Team Cosmos) has a mill that is considerably larger and nicer than mine. He graciously allowed me to borrow it to work on the frame pieces, which are cut from 0.375 aluminum and some 0.45 black polycarbonate that I got at a local plastics place from their scrap bin.

A couple of evenings yielded this harvest.
After cutting the inner rails to make room for stuff, I made the chain segments.  Here is a test assembly of one side to make sure everything fits okay.
It's starting to look like a bot!  Some drilling, tapping and countersinking and the frame is pretty much ready to go.
Yet more drilling and tapping.  Yikes, there are a lot of screws in this thing!
Took a break from drilling and tapping to build a gizmo to secure the drive motors in place and hold the main power switch.
Finally, after lots more drilling, tapping, and countersinking, the bottom and top plates are attached to the frame.

Time to get to work on the inside of this shell.
The assembly process can now begin.  First, the drive motors and some of the electrical system.
Next, the batteries are mounted.  There is enough clearance between the batteries and the inner rails for the chain drive to pass through, but just barely.

Note that the battery pack on the left is split into two pieces because of space constraints.
With the beat-up wedges from Sunflower 2 attached, the drive train components got installed.  Everything seems to fit okay!

After the wiring is finished, it'll be time for a test drive.
After a successful drive test, the weapon motor and weapon get mounted.  Getting close now!
A quick paint job and a beauty shot!