Hello again, since starting work on my Elegoo Tumbller Robot last week (which you can read about here) I’ve made good progress on the SolidWorks model. After writing the first post, I continued to draw up the initial designs for the rest of the individual components. As expected, some parts ended up being way out of proportion and I also had to make some quick fixes to my list of dimensions along the way. Thankfully, the overall shape of the robot is looking a lot better compared to earlier in the week. I’ve also made a start on refining the aesthetics of the more tricky components, such as the motors.
This is probably a good time to also outline my approach to this aspect of the project. Given the somewhat unique nature of designing a CAD model for an artifact that you don’t have the actual measurements for, I think there is good reason to shift from a systematic design philosophy to a more agile, iterative one. The benefit of this, is the inefficiency of updating part dimensions over time ends up being less time consuming than the requirement of drafting initial working drawings (with inaccurate dimensions) for each component and the assembly, and then constructing the model on SolidWorks as well. The beauty of SolidWorks, is that the model can be iterated upon much more quickly than initial sketches & drawings.
True to the iterative approach, my first model was less than impressive. It’s a little to short, the wheels are completely disproportionate and the motors are too bulky. However, I was able to take solace in the fact that the base plates and sensors were indeed accurate. I then went about adjusting the dimensions of the more problematic components with a more systematic approach. For example, the motor measurements were adjusted using existing motor designs, and I could verify it’s compatibility with the some calculations and drawings before making it in SolidWorks. The rest of the components you can see where also re-designed using the same approach and as you can see, the current model now looks far better than the initial assembly.
All was not smooth sailing on SolidWorks to reach this point of course. In particular, I felt that there were a couple of specific issues I ran across this week that are worth highlighting:
- The wheel design: As mentioned last week, I thought that the only way I could achieve the curved spoke design was to use the Surfaces option in SolidWorks, which I haven’t really used since secondary schools. It turns out there’s a much easier way. With the help of YouTube I figured out that by just using the Revolved Boss/Base feature on a suitable sketch, I could get the desired shape far more easily and quickly. On top of this, I made the classic mistake of using the length of the wheel’s radius instead of it’s diameter for one of the sketch parameters. Hence, the wheels in the first draft are half the size they should have been.
- The 4 long structural rods weren’t long enough: It took me a while to figure out why the initial design was looking so disproportionate in terms of it’s length-to-width ratio. The problem was that the 4 main rod structures actually each consist of two rods (5cm and 2cm) attached together, resulting in a longer rod of roughly 7cm. This was hard to glean from the images on the Amazon website as the joint between the two rods is hidden from view. It was an easy fix anyway, and the robot immediately looked a lot better and had a closer resemblance to the pictures.
- Redesign of the motors: For the initial render I just used a motor design I found on Grabcad.com as a placeholder. The design was far too big for purpose and not an accurate depiction of the motors used in the actual robot. However, it served as a really good guide when I was designing a more suitable motor for the current render. In fact, the current design is really just a scaled down model of the Grabcad model, but with a few other tweaks to make sure it would be compatible with the wheels and the brackets keeping them in place.
Finally, the other big change I made since the initial draft is colours! I only did this today to spruce up the overall aesthetics but it’s clearly very effective and it looks even better in PhotoView mode — but I’m saving that image until the model’s completely done. The use of textures, which you can find online, is also useful for touching up the design of the ultrasonic sensor. Note to self: the top plate of the robot shown should actually be transparent, not blue, whoops.
Looking forward, there’s still plenty to do. In particular, my focus for the coming week is to accurately design the motherboard which is very bare at the moment. This is probably going to be the most tedious part of the design, there seems to be a lot of pins and tiny electronic components to incorporate in the model and there’s not a lot of details or dimensions to go off from the website’s images. Currently, I think the best way forward is to continue with my iterative approach to the project. The first draft probably won’t look great but I can then tweak and improve the design over time.
That’s it for now, thanks for reading! Hopefully next week I’ll be a lot closer to the finished product and by the week after, I reckon I’ll be very close to finished if all goes well.