CAD modeling, virtual prototyping, and 3-d printing all have their place; but at the end of the day, You Still Need to Build Real Hardware. In this post, I’ll share a few examples of my own hard won knowledge that no book or classroom lecture ever covered. The real lessons of the day came from some time in the shop with experienced and very talented machinists.
I was doing work designing a socket drive to connect a steel shafted motor to an aluminum component. The socket and motor were both modeled in CAD, tolerance stackups were taken into account, and we had something like a .0025″ clearance between the walls of the socket and the drive shaft.
The parts came from the machine shop, and like a little kid on Christmas, I tore open the packaging and started putting parts together. We put the mechanism (and associated motor) through it’s paces and took the pieces apart to inspect for wear.
Much to my surprise (and dismay), a hard wear line appeared on the inside of the socket. After some head scratching, careful measurement, and further inspection, we found that the socket had a shallow taper to it. The end mill used to cut the socket into the aluminum had deflected inwards. The walls it cut were not straight, so the socket pinched the ends of the motor shaft. This was lesson #1.
For our redesign, we decided to use a harder wearing steel insert and a wire EDM process which would cut straight walls and press fit that into the aluminum. Now, I had run all my equations on the interference fit and the friction forces holding the thing in should have been enormous. Much to my surprise, we went through testing and the steel insert was rotating in it’s pocket.
I talked it over with the machinists and the more senior engineers. It seemed silly in retrospect, but the steel insert, instead of stretching the aluminum hole as it was inserted, merely gouged it’s way through. I had nowhere close to the interference fit I thought I was achieving. This was lesson #2.
We solved the problem by pinning the insert, but that’s another story.
What I want to point out is that CAD modeling, tolerance analysis, and all these other tools are just that – they’re tools. For example, modifying the CAD model to add the steel insert was much faster than starting a new paper drawing from scratch. That saves us time and money as engineers. But at the end of the day, we’re paid to make things that work, not things that theoretically work, and there’s no proof better than a working prototype.
Do you have any examples of a real world model showing you the real mistakes in your design? Let us know in the comments!