Think about people you graduated high school or college with. I bet they went off and became doctors, lawyers, marketers, salespeople and a range of other things, right? Some even be entrepreneurs, starting something brand new. Now answer this question:
How many of them started factories?
I am willing to bet more of them started food trucks.
There’s a lot of focus on hardware these days. Yes, it’s true that it’s easier than ever to get hardware designed and manufactured. No, it’s not true that it’s “easy” now.
I want to put a line in the sand and state something unequivocally: manufacturing is neither easy nor glamorous. No amount of flowery blog posts or meet-ups with high profile speakers will change that. And I love both. The truth is that it’s a grind getting your project out the door.
Line in the sand notwithstanding, the people doing manufacturing make it all worth it. Why? They’re superheroes! If you meet a successful engineer who regularly deals with manufacturing, it means you know someone who:
- Gets things done under pressure
- Has attention to detail
- Is dedicated to their work
So go ahead and hire a person who has successfully manufactured products…but don’t put the institution on a pedestal.
Why is manufacturing still hard?
Hardware manufacturing is difficult is because of the interdependence of so many pieces.
Imagine a product that has a mechanical enclosure and one PCB mounted inside it (my go-to example). If one component is not available or working properly, it puts the entire device at risk. Not only that, you don’t control the entire process. Especially with electronics, you can’t walk down to the machine shop and re-do a part to fix the tolerance. No, you’re buying all your parts from external sources, each of them with their own internal design constraints and tolerances.
In the case where you are working on a mechanical piece and think you could just go and modify the part or create a new one, well…not quite. You need to have the material on hand to allow for that sort of thing. Even blocks of aluminum take time to order. Margins on machining aren’t particularly high either, so there’s little incentive to keep large amounts of stock on hand. And finally, assuming you have the means to create a modified part that instead goes into your assembly, you still need to do that over and over. What’s more, you need to track which units have the modified process as part of their assembly and track it for customer support and any issues that come up.
Though I’ve proposed an idea like Continuous Integration for manufacturing in the past, the tracking of those integrations can be almost as cumbersome as the change itself.
The pretense of glamour
There are no particular articles extolling the glamour of manufacturing that I’d like to refute. It just seems to be a trend that I see. As more and more is written about manufacturing (especially about the importance of it to the US economy), people I talk to seem to idolize the process.
Over time, I find myself looking back on my own time with rosy colored glasses. The benefits of being close to a manufacturing floor cannot be understated in terms of process and design innovation, but that is the subject of another article. For now, I want to catalog my past observations as a reference for myself and others.
Factories are dirty
Making things creates waste and dust and scrap. It’s impossible to have a 100% efficient use of materials. As such, there are waste products that ends up cluttering up the manufacturing floor or creating by products like dust and grime. There are extreme examples, of course. Looking at pictures of manufacturing in rural parts of Asia evokes feelings of concern for the workers and the quality of the products. But even in clean 1st world facilities, it’s not quite the cleanliness level as working in an office. It’s hard to describe, but even when I was working in a class 10,000 clean room, something feels different about it. So while “dirty” might not describe something that by definition has no more than 10,000 particles per unit area, it is a long way from the bean bags and massage chairs of the high end software development environment.
It’s still very dependent upon people
People are great, right? They catch mistakes and have an eye on the process so that when something is awry, they note it and give the engineers feedback. Talking to assembly workers used to be a great way to understand how to optimize a process for more efficient assembly in the future. Talking to a calibration technician helped to understand which steps failed and took the longest.
The other thing about people is, they’re not so great. They are superstitious (“when I plug it in like this, it works…but not when I do this other thing”). They are stubborn (“I think we did it better before”). In short, they are human. Dealing with their every day problems is a reality and there is much an HR aspect to working with them as there is the practical “how to build this thing and get it out the door” aspect. Lack of repeatability in the process (likely with any humans “in the loop”) also can lead to lower quality products or higher customer returns due to mistakes.
When your factory floor 1st shift reaches capacity, you have two choices: stop making product or add more shifts. Most companies decide upon the latter, assuming they can recoup their costs from adding more people at different parts of the day/night. After all, if they are making money by running the first shift, it follows that they can continue doing that at night, possibly even with higher output due to fewer interruptions.
The downside is that people working with manufacturing have to flow with the lack of boundaries this schedule carries. Emergency on the factory floor at 2 am? You’re going to get called. Used to coming in at a leisurely 9:30 am start time? Not when you have a shift change meeting to attend at 7 am. The hours are long and if you’re on salary, you’re viewed as someone who is “free” (they don’t pay anything more for you to be there at the 7 am and the 7 pm meeting, for better or worse).
There is no red carpet at the end
Sure, some products have launch parties. Especially large consumer products with a bright future. But you don’t ever see manufacturing people on the carpet, do you? It’s because they’re still in the factory, cranking out product.
In manufacturing, there is no end. There is the next product that needs to start as soon as the last one is done.
What about the robots?
When I started talking about dirty factory floors, filled with manual labor, you were preparing your argument about how robots are the future, right? I’d agree with you if you weren’t 30+ years behind. Industrial robots have been here since the 70s. You can go to a scrapyard and pick up an aged industrial robot or assembly line.
The trend of increasing robots on the factory floor will continue, but won’t change the glamour of the profession.
First off, a lot of the line will not be capable of replacing with a robot, at least not in the way that people envision it. Robots require a highly optimized process, which also means high volume to justify the costs. While there are lots of downsides to humans, their adaptability is another of their positive traits. Yes, you can get a Baxter robot, which is meant to do learn the “everything else” on a manufacturing floor. But those have many years until adoption.
The pick and place machine is another relevant robot that has been on the floor for many years. This machine has allowed countless electronics assemblies to be manufactured at high speeds. However, ask an operator of a pick and place machine to see just how “automated” it is. Once you have set up the machine, it can (in theory) run unfettered. But the reality is that it’s another piece of automation that generally speeds up the process but is far from automated. More on this below.
While a “lights out” factory is the holy grail of factory automation, those lights are off for a reason: because making of things still isn’t glamorous. It’s just something that has to get done.
The material aspects of manufacturing cannot be ignored. The most automated factory in the world will fall to its knees without a well-developed supply chain. Making hardware requires material input, be it raw materials (chemicals, metal stock) or sub assemblies and components (resistors, power modules, bluetooth modules).
As stated at the beginning, each additional component of an assembly represents another potential failure point. Risk is everywhere.
While the cost of failure is high, there are often thin margins in manufacturing, representing a need to seek cost reductions in any way possible. Each change from the originally specified part might be justified in terms of reducing cost, but that introduces another possible point of error. If a purchasing agent finds a good deal on a part and subsequently substitutes a 5% resistor when a 1% was needed, the product implications could be dire. The need for communication is crucial between all levels of an organization, from manufacturing floor, through purchasing, to the original design engineer.
You know what’s not a glamorous job? Approving 37 requests for swapping out resistors via email on a Monday morning.
It’s a grind, but that’s why it’s important
The negativity of this post is not to say that we shouldn’t strive for a pristine, automated and efficient factory floor. That really is the end goal for every factory. This post is meant to act as a foil against articles, here and otherwise, that talk about the grand future of manufacturing. It’s a difficult job and that’s why you don’t hear about your college friends racing to start another factory. That’s why it’s important to continue improving processes. There is opportunity and monetary gains for those who figure out how to make the entire process better.