The 6WD Robot chassis is complete and I couldn't be happier about the results. But first, some lessons learned.
Don't permanently weld acrylic on components you know you will have to adjust you stupid stupid idiot.
There were a few problems with my original design. I opted to build frame rails that would hold the three wheels on each side. The frame rails were to be welded to the main platform with acrylic glue for more permanence than the universe itself. It looked like this:
There are a few problems with this. First, because I planned to use roller chain to transmit power across the three axles, I would need to drill those god damned holes exactly the right distance from each other.
This is stupid. Even if you laser cut the holes with an electron microscope laser cutting device it would be stupid. At the length of chain I was working with, even 0.25mm would be the difference between a slack chain and a taught chain. I drilled these holes as exactly as I could; I created a template, temp glued the template to my working surface, lined up the drill bit in a drill press, clamped everything down, and repeated the process for each hole.
It did not work; the minute differences in distances resulted in inconsistent chain tension.
Secondly, chains always need a way to be adjusted. Anyone who rides a motorcycle knows this. I know this because I ride motorcycles. But instead I decided to ignore the compendium of roller chain maintenance and just try and wing it. Over time chains will become more and more slack.
Finally, by constructing an immovable frame configuration, I was severely limiting my options. I had permanently acrylic-welded the frame rails to the main platform. Acrylic welding cannot be undone unless you work at NASA. It becomes one piece. What if I wanted to set the axle height differently for the middle axle (turns out I did!). What if one day I wanted to make an articulating suspension? Now I can do this because I've decided to modularize the axle assemblies.
Modularized Axle Assemblies
I cut a module out of each frame rail to make six modularized axle assemblies. As long as the bottom edge of the rails (seen above) are flat, you can weld them to a roughly cut piece. What I did was insert the bearings, shafts, and fixed them in place by temporarily putting these sprockets on each side. The sprockets are just there to make sure the axle is perpendicular to the frame rails. By doing this, I could be confident that the axle would be aligned perfectly with the frame rails. Then, I just set it on the 1/8" square piece and welded it on. Here are several in various stages of construction:
Next came assembly of the axle assemblies to the main platform. This time, I decided to use screws. Screws can be unscrewed, which is very much unlike the permanence of the universe. It also allows me to adjust the chain tension by filing the holes or just drilling new ones! Screws!
At this point I'm so overly obsessed with alignment and lasers that I decided to use lasers for alignment. Probably not in the way you're thinking, though. I like to draw lines on acrylic by etching into it with a small screwdriver. An etched line is several micrometers thinner than a line drawn with a pen. Problem is, it's very hard to see an micrometer-thick etched line... unless you use lasers! By pointing a laser at any edge of acrylic, the other edges and small etched lines light up brilliantly.
Arguably, this made absolutely no difference in how well I aligned my center axle assemblies, but it looks fucking cool.
Here's my workspace, just for fun:
Another thing I decided to glue in was the bearings into the holes in the acrylic. My thinking was that I was cheap and lazy and superglue is easy enough to break loose when I needed to replace them. This was just yet another mistake in a stream of fuck-ups that drove me into a state of depression and a lack of progress on my robot. Bearings are remarkably intolerant of problems with alignment. In the process of putting superglue into the holes and shoving bearings into said superglued holes, I introduced tiny variations in angle, thus boning the alignment. And, since the holes now had micrometer-thick layers of superglue, future alignment would likewise be boned.
So, I decided to do it the right way and let the bearings "float" in their holes and provide end-stops to prevent them from relinquishing their robot duties via falling off. I've gotta hand it to Lowe's. I managed to find "special application" vinyl washers with specifications so exact that I might as well have been programming Alan's 3-D printer. Lightweight, vinyl, 13mm inner diameter, sufficient outer diameter for gluing surface area, 1/8" thick. And, I managed to find some shims for my shafts that came in handy later.
Drilling the holes between the axle assemblies and the main platform was as anxiety-inducing as a reality television show about making robots. Especially since I was using a wood drill bit instead of something more proper, but driving to Lowe's to buy a $15 "SPECIAL APPLICATION YOU WILL USE THIS ONLY ONCE" pieces of metal was getting quite old. Using the drill press was not an option because of the geometry of the various clamps, laziness, and the fact that it was pretty cold in my garage (amplifies fear-induced trembling). Also, if I managed to break one of my axle assemblies, it would mean hours of table-sawing, universe permanent acrylic welding, and epoxying vinyl end-stops from Lowe's. I've become very weary of stupid mistakes costing me hours.
Drilling acrylic is actually sort of tricky. You have to go slow or your hole will be a mess. Your working surface has to be solidly planted or it's going to a) spin wildly destroying all within the radius of your target surface or b) immediately bind on the bit upon breaking through, rocketing your target surface upward into your face/the sun/into a million fucking pieces. Make sure and clamp your surface down, wear gloves and protective eyewear, and drink exactly one beer so your hands don't shake uncontrollably from the terror.
Once I had the axle assemblies attached to the main platform I assembled the drive components and attached the wheels.
Beautiful! The chassis glides across the floor showing very little drive-train friction. It's at least tolerant to being rammed into a potted plant at a decent speed without damage. Overall, I am very happy with the result and I'm glad I spent all this time on it. I did notice something right away; turning in place requires quite a bit of force. The corner wheels grip substantially and this level of force is not going to be acceptable. The solution is easy. The center wheels need to be mounted about 1/4" lower than the corner wheels. This reduces contact patch on the corner wheels, reducing drag in a turn. The trade off in traction is negligible; I made this fucker 6WD so it could drive over small mountains, not putter around in office buildings. In any kind of off road condition, all six wheels will make contact. With any luck, this robot will be able to drive up a curb.
Next up, I've been researching motors. I had tested a cheap handheld drill gear-motor from Harbor Freight, but there's just not enough gear reduction. It's a powerful motor, draws far more amps than I want to deal with, and would propel my robot at warp speed. Too bad.
I've decided I want the robot to go around 2 feet per second and carry a 10 pound load straight up if need be. You know, for good measure. There are two motors; one for each side. I think this ought to do the trick: http://www.pololu.com/catalog/product/1445. Feel free to do the calculations if you want. The wheels are 4.24" and I'm hoping to do a 1:1 transmission.