How We See Embedded Processing

Like many students, I worked my way through college in the foodservice industry. During my last few years as a student, I was a bartender at a popular upscale Italian restaurant in Columbus, OH. One of the most popular dishes on the menu was a Shrimp Pasta Fra Diavolo; diavolo is the Italian word for "devil" and this term is typically used to describe food with a spicy or peppery bite. The dish was pretty simple, consisting of a few large shrimp in a spicy tomato cream sauce over linguine, and it cost about $15.

Years later, as an engineer and a product-line owner, I was curious as to just how much of a profit the restaurant was making from this dish. I called an old friend, and he passed on the secret recipe to me (note: do not share any secret recipes with me). I then went on to netgrocer.com to price the ingredients, and here's what I found:

I couldn't help but notice that the cost of the ingredients is only about 44% of what the patrons of the restaurant gladly paid for the pasta. This is pretty high for pasta, but nonetheless, from a purely financial perspective, there is no way to justify going out to dinner—what a scam! So I ask this of you…why would you ever go out to eat when you could make the dish yourself for so much cheaper?

Of course this is flawed logic. There are a host of reasons that there is more value in the dining experience than can be captured with a simple calculation of ingredients. First, let's face it…most of us can't make this pasta taste as good as a professional chef can—and isn't it nice to have someone else cook for you? Second, can you think of a store that sells 2 teaspoons of olive oil for $0.18? Probably not…so you're going to have to buy a whole bottle and keep it on your shelf for future meals. Third, the restaurant does the dishes…I'll gladly pay to avoid the messy kitchen. Finally, there's something to be said for supply-chain management. I am pretty much restricted to the shrimp that I can get at my corner grocery store, while the restaurant can get higher-quality shrimp for less because of their high quantities.

This finally brings me to my point. Of course we all recognize that there are hidden costs of making pasta that a restaurant incurs and therefore, we are willing to pay significantly more for a plate of pasta than the sum of the ingredients. However, many engineers fail to apply that same logic to embedded systems.

Almost every day, I have a conversation that ends up with someone calculating the cost of an embedded board by adding up the bill of materials (BOM) and adding costs of hardware-development time. The calculation looks something like this:

Total Development Time = 6 months = $100,000

Total Materials = $50 BOM × 1000 units/year × 4 years = $200,000

Total = $300,000 = $75/board

Using this simple back-of-the-envelope calculation, you would not want to pay any more than $75 for an off-the-shelf board, since you could design your own for less. Unfortunately, this same logic often jumps up to bite newer companies who fail to account for the hidden costs of embedded-system development. The following are just a few of the additional factors that you will need to account for when making the calculation:

1. Inventory costs: Just like you need to buy the whole bottle of olive oil, you need to keep the components on the shelf in order to build up the board when it is needed. Whether you build yourself or use contract manufacturers, the cost is usually 20-30% of the BOM.
2. End of life (EOL) components: Many times, component vendors will EOL a part without a drop-in replacement. According to a VDC study, 27% of all embedded designs are started because of "mandatory changes due to EOL hardware and software." This unknown can drastically reduce the shelf life of a design.
3. Environmental regulations: The most recent example of this is the ROHS initiative. Most board-level products required a redesign to adhere to this environmental regulation, and with a new push toward green engineering, there are likely more such regulations to come.
4. Software costs: By choosing a fully custom design, software-development costs can increase by nearly 30%. That's right…software costs. This is primarily due to the development phase typically referred to as "board bring-up," where drivers, middleware, and OS development can take more time than the application development.
5. Opportunity cost: This is the hardest to quantify, but it could be the highest cost to your company. Essentially, the time that your engineers spend developing a custom solution could have been spent making your product better or pursuing other opportunities or making your product better.
6. Quality: PCB design is probably not your core business, and you probably don't look at the PCB layout as a key differentiator of your product. That said, wouldn't it be nice to let someone else do the hard stuff like manage signal integrity and consistent power rails so that you can focus on your UI or other high-impact features?

There are other factors too. My message to you is that you should take another look at your calculations before deciding to go ahead with a project. Maybe it's time to look at off-the-shelf hardware if your form factor can manage it. That way, you can make someone manage all of the hidden costs, and you and your company can focus all of your energy on making your product better.

Of course there are situations where custom designs make sense. Most of the time, these boil down to volume prices and/or form factor. I'm just saying that if you really think about it, there are probably a lot more designs where off the shelf hardware makes sense than you would initially think.

In other words, treat yourself! Let someone else buy the shrimp and do the dishes. That way, maybe you'll even have time to catch a movie too.

—P.J. Tanzillo