[sanddrag]

A lot of it comes from knowing what works, either from your team in the past or even better, other teams in the past. I like to go with what I think will work, and a little less if I'm feeling brave and a little more if I'm feeling worried. You can build something very reliable with very few calculations of any sort involved. You just have to be able to envision what kind of forces will be applied and where, and what the possible failure points will be. Intuition is what I use most. If you don't have a lot, calculations are good I guess. I get a lot of my knowledge of how to build things and of what works well by walking around the pits and literally sticking my head into other team's robots. I don't usually ask too many questions, but I observe taking painting every detail into a mental image. If you look at any mechanism for a minute or less a walk away satisfied that you know enough about it, you are doing something wrong. Spend a lot of time observing what other teams do successfully, and eventually you'll pick up on it.

Can you point out a specific problem you had and maybe give us a picture of the mechanism?

EDIT: Also, when you have a problem, even more important than getting it fixed is figuring out why the part/assembly failed. What went wrong? If you take a good hard look into why things fail, you'll know what you need to do to prevent it from happening next time. And after you can do this, then you need to develop the "power" to prevent failures from happening the first time.