My Caster Wheel Design
In the last blog post I went through my thoughts on what type of stabilisation attachment I wanted to use for the Tumbller. I ended up choosing to design a caster wheel for a number of reasons you can find in that post. There were however still a lot of details I needed to figure out before drawing up the CAD file of my design.
I had no idea until recently but there are actually quite a number of different designs for caster wheels. They aren’t too dissimilar to each other, but enough to make a difference in mobility and load bearing. As far as I could tell the three most common designs are the stem and socket caster, traditional (kingpin) swivel caster and kingpinless swivel caster. I’m not going to go through these in too much detail because the choice in caster is most important in applications where there is a high load. This is not the case for our small robot and so there is very little chance of the caster failing in that way. The three designs are show in the images below.
As you can see, all three look very similar and in reality they are. They all generally work on the same basic principal, of two ‘plates’ making up two halves of a raceway in which the ball bearing can roll around. This allows the the caster to swivel. The main difference between the Kingpin caster and the Kingpinless caster, you might have guessed is the kingpin. This is the a central pin that holds everything together, seen in the section view. The kingpin is often considered a weak point within caster wheels with a lot of cases of it failing under heavy loads. This is why the kingpinless casters were developed, which have began to replace a lot of traditional/kingpin casters.
Again this is not particularly important for our application the Tumbller but I have to base my design off something and the kingpinless swivel caster seems to be the better choice. It is also less complex than the traditional caster which is always nice when drawing something up on CAD. I also believe the kingpinless design is better suited to the Tumbller than the stem and socket caster. Although the stem and socket is mainly used in smaller applications such as tables and chairs which don’t generally have huge loads, they are fixed in place through a single point, the stem. This is usually a screw, as seen in the image, which is fitted into the leg of furniture or any other application. The Tumbller tray has quite a wide surface area (with respect to scale), unlike for example the leg of a table. This is something I can use in my design to increase the spread of the reaction (normal) forces on the Tumbller rather than have them concentrated in a single point. For these reasons I decided to go with the kingpinless design, which you can see below.
I also mentioned in the last blog that I might consider putting in two caster wheels rather than one. After thinking about this I couldn’t really see a huge benefit of it. Another wheel would of course increase stability to some extent but it does also have some disadvantages. For example if the Tumbller was to overcome some sort of irregularity, a three wheel design would allow the robot to always have three points of contact with the ground, maintaining static stability. This would not be the case with four wheels, where the robot would tip to get three points of contact. In some cases this could cause the robot to get knocked off course or fall. Another disadvantage would be that the two caster wheels would have to be made significantly smaller in order for them not to interfere with each other as there is only a certain amount of space on the Tumbller tray. This would mean less contact area with the ground.
Another area I considered when designing my caster was the shape and tread of the wheel. This can have a very significant effect on the performance of a wheel even in small, lightweight applications. The tread of a wheel is the area on the circumference that makes contact with the ground. This can change a lot depending on the shape of the wheel.
The wheel shape can vary from flat tread to round tread, shown in the image above, or anywhere in between. Having a flat shape gives a larger surface contact area with the ground leading to better traction and stability. It does however make turning a lot more difficult. For a caster wheel to turn it needs to be able to rotate about itself or skid. This is much more easily done with a round tread. There is an clear solution to this, or rather a compromise, where we design something in between. This is generally called a crowned tread and what I decided to use in my design. This can be seen more clearly in the image below.
There were a whole load of other factors that I could have also considered such as caster offset or increasing traction through making the center of gravity as close to the drive wheels as possible. These would most likely have required me to change the design of the Tumbller as well which is not something we would be able to do if the robot does come. As mentioned earlier a lot of the smaller adjustments made to caster wheels for the most part only come into effect in a significant way in large load bearing applications. I may make slight design changes in the future to my caster but I am quite happy with it for the moment.
Check out a short video of the full Tumbller Assembly I made using motion study on solidworks.
Thanks for reading through my post. If you enjoyed it you might like keeping up to date with more regular updates of everything going on here. You can do that my following my Instagram page (link in my bio).
Thanks again, more soon.
