Print this Post

Hardware is HARD

Over the past month the biggest lesson I’ve learned is that hardware isn’t called that because it’s physical; it’s called that because it’s really difficult to do. In talking to people who have had successful and unsuccessful products, they are all consistently saying the same thing, and with all the stuff a company must endure to make a physical product, it’s amazing that people do it. Here’s a journey through the process of developing a consumer electronics product, complete with many of the things that will trip a person up and hidden costs:

First someone has an idea. They may spend a while building a prototype or sketches or something to convey the idea. Building a prototype can be expensive because everything is a one-off part, and the various methods for forming plastic are not cheap. Sometimes it’s better to use a softer material and a CNC machine to mill a demo part. Components cost 10x what they would if you were buying in bulk (1k or 5k) quantities. Even getting a circuit board can be a tedious process. Let’s delve into that a little more.

When designing a circuit board, there are a few software packages available. The famous professional one is Eagle, and you can get away with double-sided boards smaller than 160x100mm for free (as long as it’s not commercial). The price for professional work, though, is steep. A license can cost over $1k easily, or you can purchase a 1 year license for as low as $550. The most prominent free schematic software is KiCAD, which apparently has been getting better and better. Once you have your software, you design the schematic. This involves getting the parts you want from a library. You are almost guaranteed to have parts that aren’t in a library, so you have to learn how to make your own parts by reading the datasheets, which are sometimes hard to find. Or you might find the part, but when you order your component and place it on the board it doesn’t actually fit. That is fun, too. So you design your schematic, then switch to the layout and route all your components, trying to apply rules that aren’t really written anywhere, including things like minimum trace width, when to angle your traces, ground planes, trace length matching, and a whole host of other considerations that take a career to learn. Once you finish your design, you export to Gerber files, which you can then send to any factory, along with some specs like surface finish, solder mask color, board thickness, and other cryptic specifications. Hopefully the board you get back works.

So you’ve built a few prototypes and they work pretty well and you want to start mass production. Ha. Let’s start getting other ducks in a row first, because you have to do a lot of things in parallel. You have to worry about your brand name and product name. If you infringe on any trademarks, customs can stop your product at the border and you may not be able to sell in that country. Each country has different sets of trademarks, so some brand that works in one country might not work in another country. So you have to get that all sorted out and get trademarks figured out in the countries in which you want to sell.

If you want to retail, you’ll probably have to have an SKU, which is issued by a central authority and every SKU costs money, and every variation on your product needs a different SKU.

If you have USB, you’ll need a unique ID for your application. The USB regulating body charges a few thousand dollars to any new company that wants to sell USB devices, but once they have their own App ID, they can generate as many Product IDs as they want (up to 65000). There are a couple companies that have managed to sell individual Product IDs, but only because they grandfathered in before the USB board specifically prohibited this activity, so it’s a little sketchy. You also need to have a special license if you want to display the USB logo on your product. The same is true if you want to use the Bluetooth logo on your product.

At this point you can also start working on your FCC license. Pretty much any product with a circuit, and certainly any product with a microcontroller and crystal will have radio leakage and must be approved by the FCC. Best get started with the paperwork now, as it can take a while. Be aware that you will also have to worry about UL in the U.S. and CE in Europe.

Cute puppies will rescue you from the misery that is involved in this saga of product development.

Now you’re in Shenzhen and you’re starting to look at how to take your prototype to mass production. Unless you did it right in the beginning, you’ll probably need to work with someone to redesign all your plastic and metal parts for manufacturability, and this is another field that requires a career to master. For example, if you are doing injection molded plastic, there are books-worth of information on how to design correctly. Even thickness across the part, support structures, stress and cooling profiles, part lines, designing for assembly (there’s a whole book’s worth on making snaps or screw holes). The biggest thing is to design your parts so there are no slides, which increases the complexity and cost significantly, but is very difficult to design around. Once you have your design nailed down, having the tool milled will cost a few hundred to a few thousand dollars, so you want to get it right quickly. There are prototyping companies that can help you out with this, which is nice. We toured Star Prototype, and they have their stuff together to help with this part of the process. Once you have your tool, it’s yours, and you can take it to many other factories.

But the plastic parts aren’t it. You also may have a membrane switch, which also requires tooling costs. There’s the PCB (printed circuit board). There’s the components. There’s PCB assembly (putting the electronic components onto the board). There’s product assembly. There’s packaging. So let’s keep chasing the rabbit.

You’re reasonably satisfied with your plastic parts, and you have a good idea how the assembly is going to work. By the way, assembly is another area that requires books of knowledge, because every step adds to the cost, and there are ways to design parts to make them easier to assemble and reduce rework or defective parts (which either you or the factory will pay for, and which you have to make sure don’t sneak into the palettes of good products). So hopefully during the whole design phase you’ve considered all the steps for assembly and optimized your design accordingly to make it easy to assemble.

Now you want to work on the PCB. So you go to a variety of factories and explore and see what their quality control practices are and see a surprisingly low amount of automation and wonder how these people can breathe the acid fumes all day and realize that this is completely normal for a factory, or maybe even better than some. For larger factories they even build rigs with hundreds of little pins that are used to test every board, and people that examine under microscope random boards. You find a few that you like and start a bidding war between them, all the while hoping that they don’t bid themselves into a range in which they have to cut corners on your board, because you can pretty easily end up with a high failure rate if you don’t design your board to be easily manufactured (another book). You end up with a quote that’s 1/2 to 1/4 of what it would cost in the U.S. Cool. Except for that feeling in your gut that something could go wrong and you’d be screwed.

Next, there’s components. You have your bill of materials and go to the electronics market or alibaba.com to get quotes. All the booths look the same, and there is no way of telling if they are a direct supplier, or if you are dealing with a middleman or a second level middleman. So you go through the BOM and they can supply a surprisingly large percentage of your BOM, which is a little weird because you thought they were just an LED company. And the prices are all over the place. Sometimes it’s 1/10 of the price you saw on Mouser or Digikey, sometimes it’s 1/2. You ask what brand the part is and suddenly the price doubles. You see another booth where someone is pulling stickers off one reel and putting them onto other reels of cheaper knockoffs, trying to sell the cheaper ones as if they were the more expensive ones. You realize this is a jungle and without help you have no idea of the quality of the parts you are ordering or their provenance. And if you have critical components where a failure after a few hours will make your product inoperable, you are setting yourself up for huge risk.

Assembled components aren’t really any better. If you are building a low voltage consumer electronic device that doesn’t plug directly into a wall but uses a wall wart, then getting through UL is a lot easier because that wall wart should have gone through UL already. But this is China! So IF the wall wart has a UL print or sticker, and IF the license number on that UL listing is valid, and IF that listing actually refers to that particular supply and not another supply that factory has produced at some point in its past, and IF the supply actually has all the safety parts inside that it is supposed to, then you are probably ok. If any of those things are not true, then customs will stop your product at the border and you’ll be delayed or denied.

Cute puppy. He’s telling you not to get discouraged. Keep reading, though.

Now you have your plastic, your membrane switches, your components, your PCB, and you’re looking to assemble. And by the way, you don’t want to have all these parts done at a single factory. One reason is now a single company has all the details of your product, so a middleman handing a thumb drive off to a friend will make a competitor or knockoff expert’s job really easy. Another reason is that factories rarely have the capability to do all those parts, so they’ll tell you they do but they’ll really subcontract it out unknown to you, and you can bet they’ll subcontract to the cheapest possible bidder and charge you a markup for being a middleman. Anyway, you’ve got your parts and you’re ready to assemble them. You tour a bunch more factories to figure out what they are good at and what tools they have and what the expertise of their staff is. And just because a factory does something similar doesn’t mean they can do your thing well, and just because they have a specific tool doesn’t mean they have the staff to use it in the way you need. So you can plan on spending a few weeks on the factory floor working directly with the engineering staff and assembly crew to make sure that everyone knows their job and is assembling everything properly, and there is little to no waste, and the waste there is has no way of somehow making it into packaging as if it was a final product. Hopefully you’ve built test procedures and special rigs to test all the parts of your product to verify they work correctly, and hopefully you have a contract with an AQL (acceptable quality level) spec that details exactly what is required in order to pass the tests.

I skipped the negotiation part. You’ve found a couple factories that you’re interested in. You don’t want the top tier factories because you’re too small-fry for them and if you get any attention at all you have no leverage anyway. You don’t want the bottom tier because you will get screwed. You want a factory that needs your business and wants a famous product, but already has a few wins and can demonstrate that it does good work. Then you start negotiating; they’ll nickel and dime you, and slip in strange fees and engineering costs, and if you’re new they’ll demand payment up front, and try to take any leverage away from you. You will have minimum quality levels, and hopefully a friend who also manufactures in that factory who can act as leverage, and hopefully you’ll have developed a relationship with the factory boss who understands your product and is investing in its success with you.

Now you have a factory producing units (assuming all the parts were ordered and successfully delivered to the factory and are up to spec). It has taken a few weeks to iron out the kinks, but there’s a palette sitting on the loading dock. Hopefully you’ve figured out what happens next. The product will be shipped by boat probably (which takes a couple months), or if you are in a hurry you can ship by air. It will arrive at customs, and they will check for FCC approval (you got that, right? wait, what?), UL or CE approval (you got that, too, right?), verify that it isn’t obviously violating any trademarks, if it’s for kids will verify that it meets those standards, too (you have that as well I assume), and then send it on its way. Well, it will of course look for any tariffs it can levy on you.

With these bombshells, you’re possibly in trouble. Fortunately you could smuggle a few units in as prototypes and send them to the FCC and UL for testing, but since you’re supposed to send the final product to them, you have to wait until you are most vulnerable to do it; when you already have a bunch of units waiting to come over and any changes will be at great cost. The FCC will probably not pass your device. Everything leaks, and will be probably out of compliance somehow. You either need to fix it or negotiate something with the FCC (not a bribe). UL and CE may care about you device, too, and if your product is for kids, or Customs decides it may be for kids, then you have a whole bunch of other regulations to worry about.

You’ve navigated that minefield, and now you’re ready to get it to retailers. Actually, you get it to distributors, who get paid a lot of money just to make a phone call to the retailers and get a place on their shelves for your product. And you’d better get it to the distributor on time, because if your shipment is late and you miss the date, then you don’t end up on shelves. Except that you have to pay for those empty shelves. So you can’t sell your product, and you’re paying for the empty space on the shelves.

But assume that somehow your product made it to shelves and you are selling units. You have to deal with returns and customer service, because it’s quite possible a fraction of your units are coming out of the factory or making the journey to the shelves (you had a packaging engineer work on making sure your product could survive the extreme conditions of the cargo ship, and transportation by truck, right?) defective. This cuts into your margins, but hopefully not much.

Until something big is wrong with your product. If it’s for kids then maybe the kids licked it and got lead poisoning, or if you dropped it a special way a small part breaks off that can be eaten by infants. Or maybe a cheap knockoff capacitor is causing fires in some units. Now you have a PR nightmare, a customer service nightmare, and you’re getting sued, which may put you out of business pretty quickly.

But if you can avoid that, and your product is wildly successful, then you’ll have a small margin of time before the knockoffs and similar products start cropping up. And any company with a patent similar to what your product does will come knocking on your door looking for a handout or to shut you down. Your only protection is to have filed for a patent, no matter how flimsy, just so you can claim to have a patent pending on your product, but it won’t stop the knockoffs in countries that don’t care about that.

If you’ve made it this far, somehow you’ve navigated a million things getting in your way, but you’re not rich. The standard formula for retail price is 4x the cost of goods sold. But the retailers and distributors want at least 1/2 of that. The cost of goods is 1/4. That means you get 1/4 to split between all the employees and all the administrative costs and all the engineering and tooling costs. If you’re lucky, you can get away with higher margins by selling online, but those higher margins are usually at the cost of significantly lower volume.

So at the end of the day, a small hardware startup is tasked with a nearly impossible feat. If you don’t know the right people, the right regulations, the right organizations, and the right engineers and designers, you have a slim chance of a successful consumer electronic device. You don’t have the knowledge or the leverage to keep from getting screwed by any of the million little regulations or shrewd people.

And that’s just the overview. Getting into the nitty gritty is even scarier.

Don’t worry, a cute puppy is here to save you from your misery and make everything ok.

Permanent link to this article: http://engineerinshenzhen.com/hardware-is-hard/

1 comment

  1. Rodrigo Cardoso

    Great insight. Thank you

Comments have been disabled.