Encoder Front Page
SRS Home | Front Page | Monthly Issue | Index
Search WWW Search seattlerobotics.org

'Wheels the hard way...'

Kenneth Maxon

Lots of discussion (and many Encoder articles) focus on wheels for mobile robotics. The premise, usually, is how to do it faster, better & cheaper. Here I present another view. How about a full-on custom design that takes painfully long to fabricate and should not be mass-produced by anyone? Think I'm joking? Read on...

This informal article focuses on the creation of wheels for the author's small robot project. Along the way, I'd like to quickly touch on the design criteria that guided me along the way to the solution I finally implemented.

The need for these wheels was driven by the same rules governing all amateur robotics ventures. I need it, I can't afford to pay somebody else to do it, and I need it to work with the rest of the mechanical components in my design. The figure below is presented to have a visual image to keep in mind while considering the design criteria presented below. Think about how each of these criteria might apply to the specifics of this robot! The wheels discussed in this article are clearly visible in the center of this CAD representation.

Side View:

Front View:

One of the more important driving aspects of this design came in determining the correct size of the wheels. This simple parameter is one of the most often (in my opinion) over looked design criteria in designing mobile platforms. Following, I present a free form roughly associated list of design criteria. If you take a minute to read through them carefully, hopefully, you will be able to draw the causal links between the associated members and see the importance that wheel size really carries.

Wheel Design Criteria:

Coeficent of friction. Wheel to floor / wheel to mounting.
Torque required to spin.
Gear train to apply torque during standard operation or dynamic loading.
Wheel mounting. Shaft in hole in plastic or bearings? Thrust Bearings? How much torque will be lost due to the gear train. 1/2? More?
Wheel splaying? Inward tracks true when traveling straight, Outwards turns better. Dynamic with spring suspension? Transmission of power to and from splayed wheels?
Encoder mounting? To motor (small gear ratio in drive train) To wheel(large gear reduction, lots of slop)? Tach?
Drive train: Belts, chains, Gears, Direct drive? Top speed available after gear train? Tooth pitch if toothed belts or chain used. In larger robots, tooth loading for gears and chain sprockets. Torque/power transmission ratings of chain, belts, gears etc.
Mounting of drive train and motors? Screws? Hot glue? Nylon Straps? Rubber Bands? Can the mounting system be taken apart and reassembled with ease?
Torque requirements of motor? At speed? do they provide any torque at the speed I want to operate with? I.E will they be able to climb a ramp at speed? Acceleration speed. 0.8 G? Continuous Duty? Reversible? DC? Brush wear?
Weight of motor / gear train to provide such torque?
Current drive for motor? Stall Current? Running Current?
Motor technology? 2 pole? 5-pole? Brushes? Operating characteristic at extremely slow speed?
Amplifier to switch high current. Can my system switch 24-V @ 6Amps per motor? Over current protection?
Battery life when running 2, 4 .. X motors at X amps. Yeah, it works great, but it only runs for five minutes before I've got to recharge the batteries. Battery Size? Battery Weight? (can the small motors I've chosen move these huge batteries?)
Fabrication technology? Will I need a milling machine, lathe, indexer or other specialty tools?
Construction time frame? 1 day, week, month, year? Can I actually finish the work myself or will I need to outsource some of the fabrication for this design?
Cost & availability?

Do you address each of these criteria when designing or is it more of the, 'round wheels roll' idea? Food for thought...

The only premise I'd like to really put forward with this article is to point of that amongst the free association above, cost and availability come last. When designing a robot for personal interest, why not design exactly what you want, and worry about where the money comes from later? Worst case you may find yourself having to fabricate a special screw or retaining ring, because the exact one you want/need does not exist. When it's your personal project such eccentricities are acceptable.

In my personal design regime I always build a complete CAD model first to check fit and function. Later the CAD modeled parts are used as the basis for the generation of CNC code used in fabrication. Quite often when working with assemblies of five or more pieces this invaluable step identifies fit / clearance problems and saves time and money. Much of this work can be done on engineering paper as well, but becomes more complex when the parts have interactions / interfaces in three dimensions.

In the design presented below, you may note, that aesthetics should be worked somewhere into the list above. Buried in the support column are four bearings to support both radial and axial forces. There are two needle thrust bearings and two radial bearings as this design uses an offset single sided support system for the main axle. The main section of the wheel (rims) are made in two halves that bolt together with #6-32's to locate the tire. While manufacturing the toothed belt pulley I cut a six-inch log and then 'lopped' off the sections you see here. Note the correct angle has been cut onto both the leading and trailing edge of each tooth.

With the above said, I present a few photo's of the wheel's I've built for my personal robot. These wheels are NOT cheap to make, fast to make, nor are they mass producible! But they are what I wanted, and pretty cool!

I went ahead and cut back on the size of these pictures in order to make them a little faster to download. Unfortunately this looses a bunch of the detail present in the original. Sorry...

The pair of these wheels took about six weeks to design. Much of this takes into account the massive number of iterations put into going through the design criteria listed in the free form association above. The actual time spent on the computer in CAD software boiled down to about three days as aesthetically, I went through several revisions. From the time I hit the machine shop until I had a pair of finished parts took another full three weeks. Unfortunately, I've got a 60+ hour a week (sometimes much more) day job that really slows this kind of work down.

Below is a disassembled view....

Finally, the picture below shows the connection of the toothed drive belt and drive pinion:

Is it right? Is it horribly wrong? Should I have bought a set at the hobby shop? I'm not attempting to answer that one just presenting my solution to a design problem, as it's vastly different from the solutions provided by others up to this point.

I wrote this article nearly two years ago and must apologize, as I've never taken the time to finish it. As such this has been presented in it's informal, unfinished state, but I hope the article has inspired you to take a moment or two to stop and think about designing something closer to what you want for your design at the expense of getting it done in a hurry.

fs Drinking Diet Coke - N - Gettin' it done. - Kenneth

Come visit my web site to check out more on this and other projects:

Max's Little Robot Shop