Anorexia Build Report (Robots)

The genesis of the Anorexia design dates back to BattleBots 5.0. Based on what I learned at the competition, I had the idea to build an ultra-small, ultra-thin robot. There were two reasons why I felt this would be a valid design idea; first, such a tiny robot would be hard to hit, and second, the smaller it is, the thicker the frame and armor can be for the same weight. Also, as this was the first robot I would completely design and construct from scratch, I wanted to make it both simple and a learning experience as I progress towards competence (or at least, the appearance of competence) in machining.

After some initial research, I came to the conclusion that an internal height of about 1.5“ was about the skinniest a robot could be without getting totally insane. The main governing factor is the size of the motors, and the smallest motors available that have enough power to shove around a robot are all in the 1.3-1.5” diameter range (for example, the Astroflight 915).

However, availability problems with the Astroflight, coupled with rumored fragility on the part of their planetary gearboxes, put the plan on hold during the fall. Until I had a drive solution, I was out of luck.

Then, at Robocide, I found out that the gearheads at Team Whyachi had come out with a new gearbox, the T-Box. A scaled-down version of their TWA-40, it was exactly (wait for it…) 1.5“ high, drives a 2.25” wheel, and is designed to use standard “RC Monster Truck” motors; I eventually picked the Trinity Monster Maxx Mild motors, $45 each from Tower Hobbies. The RC car industry is also starting to get into brushless and cobalt motors, so if more torque or speed is needed, I could go to something like a Hacker C40 or a Promax Cobalt Speed 400. But for now, the Trinity motors look to be more than peppy enough, considering there will be 4 in the robot.

After a bit of research, I settled on 1 1/2 x 9/16 x 3/16 Steel “C” Channel as my frame material. 3/16“ of solid steel is plenty beefy, and the 9/16 flanges make it even stronger. Channel is definitely the way to go when balancing strength to weight. Because my wheel diameter is only 2.25”, I could only spare 1/8“ top and bottom for armor, leaving 1/4” wheel clearance both normal and inverted. Should be enough.

Firing up a spreadsheet, I did some weight calculations. Clearly, the smaller the robot, the lighter it is going to be. Assuming a square robot, for every 1“ reduction in the length and width of the robot, I save 4” of perimeter channel, and make a big reduction in the size of the top and bottom plates (going from a 16“ to a 15” square reduces the size of the armor plates from 256 square inches to 225 square inches).

After a bit of fooling around, I figured out that a 14×14“ plate (with channel inset, so 12 7/8 x 12 7/8 x 1 1/2 internal space) was about as small as I could comfortably go, in particular since I wanted a speed controller on each motor, and plenty of batteries (4 custom 7.2V BattlePacks hooked in pairs to provide over 6 AH at 14.4V; overkill for now, but I'm thinking ahead, as you'll see).

Assuming aluminum armor plates (with small steel plates over the motor mounts) as well as aluminum stiffening struts front-rear and left-right, my weight calculations came in at 28.27 pounds. And that was with a bit of a fudge factor against the inevitable weight creep, and before weight-reduction hole drilling.

Wow! I could do a 30lb'er that would basically be a brick! And still have a 1 3/4 pounds for extras like a wedge.

It was then that the design came together. I'd build the brick, and then build a set of add-on weapons. The brick would be the drive, power and control module, and the weapons would attach via a set of hardpoints. I'd do simple ones at first, compete as a 30lb'er, then based on the experience gained, build additional weapons, and perhaps upgrade the motors, and compete in the lightweight class as well. Or even build a duplicate bot and compete as a multibot. And all that battery juice could even be used to power some active weapon modules…


As the various parts I needed trickled in, I began construction. Step one was to make some 1/2” ID spacers to hold the wheels on in the right place. Lowes had just what I needed, but I beveled one end on my Sherline lathe so that it matched up with the inner race of the bearing in the gearbox. The first image shows a cute trick, using a metal flat to get the spacer perfectly aligned in the chuck; you just get it snug, then turn on the lathe and press the flat against the piece, and the vibrations walk it into perfect alignment.

Next, I did a lot of drilling and tapping to create the various frame pieces. There are 12 bits of channel in Anorexia; the front/rear and left right edges, and 2 small bits that partially surround the each of gearboxes. These small support elements are included for several reasons. First, even though the gearboxes have only top and bottom mounting holes, they provide positioning and support for them. Second, for extra gearbox support, as well as frame rigidity, on the top corners of the robot, just over the gearboxes, the aluminum armor is replaced by a steel plate. This provides a much more rigid mount for the motor, as well as stiffens the whole frame.

The plan is to get the whole thing bolted together and tweaked out just perfectly, and then weld the steel.

The bottom plate was precut by my friendly local metal supplier. Turned out I gave them slightly wrong dimensions for the notches needed for the wheels. I had originally planned to CNC drill the top and bottom plates, but my source for free CNC work was stacked, so I decided to hand-drill the bottom plate. Made a few minor mistakes, but a bit of dremelling fixed that. I was originally planning to use flathead cap screws, but the precision required, not to mention the huge amount of countersinking I'd have to do, made me move to buttonheads (and accept some minor loss of ground clearance), at least for the time being. If I make new armor plates, and can CNC them, I'll try countersinking.

Preliminary assembly went well. Note the different lengths of the rear gearbox supports; I forgot to cut the right number of these when I was at a friend's shop using his chop saw, so I made them using some scraps, and will cut them down when I go to weld. You can also see one of the motor mount plates. They bolt to the front and side frames, rest partially on the gearbox supports (the tapped holes in the supports are for bolting on the armor), and will be welded over most of their perimeter eventually.

I also built some aluminum stiffening members (upside-down T stock). These serve two purposes; first, they fit very snugly in the robot, and transmit stress from a direct impact in the center of one of the edges to the other side of the robot, reducing the chance of buckling. Second, they keep the various bits of the robot from sloshing around, and all the weight reduction holes drilled into them are handy for threading wires.

I'm also planning on using all the victors I have left over from the first robot, one per motor. Running such a small motor at only 14.4V, they shouldn't even get warm, but if they do, if I slightly trim down the fan mounting posts and use half-height fans, I can get them to fit in the 1.5“ space – barely. I'd have to drill some ventilation holes in the armor.

Some last details and I'll be ready for my first test assembly. First, I drilled out all the motor mount plates using the first one as a template. Lesson: you can never have too many sets of step blocks!

Then I drilled and tapped all the various weapons mount hardpoints, as well as two small holes in each side for threading electrical connections. I used the cute “shift the base while the drill is in a previously drilled hole” trick to precisely space the holes even though the Sherline mill isn't quite wide enough to handle the whole piece. I also made myself some matched spacer blocks that came in very handy.

The wheel holes needed to be 7/8” in diameter, and there was no way I was going to be able to get a 7/8“ drill bit with a 3/8” shank, the largest the Sherline will take. I was resigned to having to schlep to the “real” machine shop (and hope their non-CNC mill motor, which had conveniently blown the week before, was fixed), when I noticed this nifty step-drill at Lowes for 30 bucks. It worked like a charm, and cut through the steel like butter; I just had to be careful that the tip didn't hit the mill! As you can see, the motor shafts (and my spacers) fit the holes just right.

The step drill also came in handy when drilling out some unneeded metal in the gearbox supports, as well as enlarging some of the lightening holes in the aluminum braces.

Valentine's Day Assembly

With all the bits mostly built, I had to put Anorexia together to see if it would work, and get a better idea of what's left to do. And surprise, everything went together like a charm! I was shocked… shocked! The bolts in the frame are just placeholders for the hardpoints, by the way. And those little white booties around the motors are strips of tyvek from an old floppy-disc sleeve, the purpose being to make sure the motor wires don't touch the frame – this would be “a bad thing”.

I was even more shocked when I lugged it down to the garage and tested it. The darn thing worked! Very controllable, very powerful, and very fast!

But this was as nothing to my shock when I weighed it. Even with those gearbox supports that I have to cut down (and haven't drilled out yet!), it comes in at a skinny 26.5 pounds, almost 2 pounds under my estimate.

2 weeks until Battle On The Beach, and all I have to do is weld it up, attach the top armor, build a wedge attachment (and anything else I can think of…), and paint it. I might just make it!

Battle On The Bugs

I had a great time at Battle On The Beach. Red Menace lost again, but Anorexia had better luck; I won a fight, lost another in a 23-22 split decision, engaged in Bunny Soccer (our side were all running the T-Box), and had a great grudge match with Eat Hitch and Die.

Robot debugging consumed a lot of my time. The first problem was that the insides of the bot got so hot that the endbells of the motors started melting (the tyvek booties just vaporized!). After my first fight, I'd used up all my spare motors; thank goodness for overnight delivery from Tower Hobbies. This issue was solved by drilling a lot of vent holes in the top and bottom plates. Not horribly pretty but it worked. I also reconfigured the transmitter so that most of the time, the robot ran on 7.2v instead of 14.4; it appears that the hobby motors really can't be overvolted for long periods of time.

My second battle was with an intermittent short in the PWM cables, one of which had melted due to the heat in a non-obvious way, smoking my mixer (but not, thank goodness, my radio!). Lucky for me Dan Danknick had a trunkfull of spares.

And then I had to deal with radio range problems, which I eventually fixed by mounting a Dean's antenna on the side of the robot.

Apart from these issues, the robot performed very well. It was fast, responsive, and very sturdy. The only damage sustained was a slightly bent axle, and that could have been fixed onsite if need be. Fortunately, Terry of Team Whyachi had an extra.

Am currently preparing for the April Robot Club & Grille event, but it's mostly some very minor tweaks. In particular, I'm building an anti-spinner attachment.

I am also going to test the small johnson motor to see if it works well with the T-Box. If it does, great, because they look to be sturdier, more powerful, and cost only $3.50. Such a deal.

Robot Club & Grille April Tournament

Anorexia did very well, taking second place! In the process, a couple of minor design flaws were exposed; first, it needed stops on the wedge to prevent getting chocked up on it. And second, I need to drill some access holes so that motors can be swapped without removing the bottom plate.

Once back in Wilmington, I redesigned the wedge attachment to make it more sturdy, and to have an integral stop (the big bolts) that prevents the chocking when right-side-up. I need to test it some more when inverted.

Drilling the motor access holes was also quickly accomplished. And as a side-effect of the new wedge attachment mount, I added more mounting holes on the other 3 sides of the frame.

In addition, I am adding cooling fans. The obvious mounting place was on top of the Victor speed controllers, but the standard fans are much too bulky. However, I was fortunate enough to find 40x40x6mm 12V fans at Mouser Electronics that were just the ticket. When mounted on a standard Victor, the complete assembly is exactly 1.5“ high, but I milled 0.100” off the Victor support posts to allow for some wiggle room. (Actually, I milled too much off, thus the washers in the final image). I also milled off the PWM cable housing so I could insert standard hooded PWM cables.

Lots of pictures from the event here.