Fascinating. I used to do decades ago with my ears what their devices do now: Listen to machinery while walking around the plant I worked in to see if I could pick up on a trouble spot, because I learned over time that mechanical equipment that was heading for failure sounded a bit different then the equipment that was still running smoothly. Belts, bearings, rollers and plates would wear and I would know to pay more attention to them more often, rather than just react when something happened. People thought I was nuts when I told them I could hear that something didn't sound right through all the noise, which was considerable (you had to wear hearing protection all day).
Some of the more fancy machines these days have embedded microphones to tell if a bearing is going to fail soon. They make a distinctive noise based on shaft rpm and bearing diameter when they start to fail, so you know which one to order before you spend a day tearing the machine appart to get at the faulty bearing.
Bearings also tend to start running quite hot when they are at the end of their lives. Bearings on trains often have smoke bombs on them that go off at a specific temperature.
As a hardware startup founder, I would like to note that the lean model only really works for hardware when you not creating something complex (ex lots of mechanical parts or lots of integrated electronic systems). In cases like that, the lean method completely falls apart because of the cost, and it becomes really important to make sure your calculations and simulations are correct.
Agreed. There are some important other steps as well. It's a bad idea to sell unreliable junk. So having a sense of what kind of reliability testing you need is key. Reliability testing should start as soon as possible. This testing often can identify significant problems in the design or assembly that are difficult to model. Things like temperature cycling, damp heat, and high temperature storage at a minimum. Also a big piece of HW design is designing for the whole distribution of parts you might get not just the parts you have now. Will your design fail if you get parts that are at the limit of the data sheet specs? This is where the modeling @FreedomToCreate is talking about is key. Lastly, making sure you really understand the concept of yielded cost. If you get your yielded costs wrong you will be wrapping dollar bills around every product you ship. If your gross margin is below ~35% on a new design you should be worried.
I'm sure it's true and I don't mean to diminish your comment but.. Isn't that kind of true for anything? Start small and build up.
You're not going to be a contender of, say, Apple in a year or even 10, but there are several companies in the pipeline already at more advanced stages
In some complex and mature markets if you start too small you will never be able to compete with existent players. "Starting up in a garage" works mostly on new and undeveloped markets.
Depends on what you can leverage. You can start a car comapny by modifying existing cars, go up market, build a business and use the same suppliers as the other car company's. Give it 40 years and you really can start a major car company from scratch. Granted this generally takes long enough that most people fail or get bought out.
As an extreme example, before starting Tesla, Eberhard took an existing electric car [A], which was itself built on an existing kit car [B], and had a different battery technology installed [C]. And then used it as a proof of concept to start Tesla.
