Updated: Jul 20
I suppose it’s actually building good hardware that’s hard, and there are a lot of reasons for it. Here are five key steps that I’ve used to make life easier:
Write a Good Spec
This could easily be the topic of another post, as writing a spec which is a useful and living reference vs. burdensome paperwork is not that straightforward.
GD&T is Important
Yes, those tedious gd&t symbols are necessary for two reasons. One, they are a great way to know if your engineers know what they’re doing. But more importantly, they define the uncertainty in your designs and define the cost of manufacturing. And, until you have defined your tolerances and measured them, you’ll never know if your equipment is failing because you designed it badly or because someone used a blunt cutter.
Don’t Buy Junk Parts
On the topic of blunt cutters… don’t buy junk. Building and testing a prototype takes months- don’t jeopardize it by sending custom parts out to the lowest bidder. At the beginning, buy from someone you trust, who will deliver quality and ideally tell you how to make your parts better. Cut costs later once you know it works.
Have Professionals Build your Prototypes
Who doesn’t like working in the shop, doing some wrenching, and soldering on the baby you designed. Maybe this is ok on a first rough, middle of the weekend build, but anything you want to test or build more than one of, you should hire a professional for. A true skilled technician will do it better, faster, and won't make excuses for your lousy unassembled design.
Test as Soon as You Can All of the above culminate in this point. Once you’ve built it, run it. There’s nothing like test data to really know if it works, how it works, what affects it, and when it will break. Instrument your machine and environment around it- I would suggest adding a logbook so you know when people mess with the equipment, and then collect as much data as you can. You’ll be guaranteed to learn a lot.
Peter Kostka holds a M.Sc. in Nuclear Engineering and Engineering Physics. He has 10 years of experience leading hardware product development teams and using and testing simulation tools to optimize hardware design in the aerospace and automotive industries. Peter noticed that companies find it difficult to give meaning to data. That is why he created Machinery Analytics, which is a software that uses machine learning. He is now helping innovators increase product quality and get to market faster.