From Napkin to Final Product, One PCB Design’s Journey (Part 2)

The only thing that I am qualified to have an expert opinion about is the final product: the outcome of all their effort. For this reason, I really don’t feel the need to visit facilities, shake hands, and struggle to remember names. I am not against doing it for the sake of learning, but to make it part of the standard operating procedure is really an unnecessary use of time.

By the way, I expect that a lot of people (especially members of the old guard) will disagree with this opinion. They would probably say that building a relationship with these prospective vendors is of paramount importance. Me? I am looking to get a job done. I really don’t want to come off as jaded or cynical here, but experience has taught me that relationships like these develop quite nicely once money starts changing hands.

The situation can be boiled down to three questions for any vendor. The three questions are my method for sourcing almost any custom parts—not just printed circuit boards. Getting those questions answered quickly and making decisions based on that information has been my strategy for making rapid progress over the years.

The three questions:

1. Can the vendor do what you need?
2. Can the vendor get it to you when you need it?
3. How much will it cost?

Let's address the specifics of each of those questions:

Can the vendor do what you need? This is a simple question than can connote a whole lot of excruciating detail. Your project specifications will drive your board design, and your board design will shape this question.

SPEC-->PCB Design-->The Question

Each board house is going to have design preferences, so depending on your project specs, you may have the freedom to adjust the design so that they can do exactly what you need.

How much will it cost? Price between manufacturers can vary a lot. Your job is to try to get a minimum acceptable level of quality for the lowest price. There is a compulsion common to engineers, especially the newer ones, to design perfect products made of the best materials. Unchecked, this compulsion would result in titanium-tantalum alloy, gyroscopically leveled coat hangers with SIL 3 failure ratings.

You will have to fight that urge. Maybe you don’t need tolerances as tight as you think. Maybe your board doesn’t need to be that small. After all, smaller boards usually require smaller tools to cut, and smaller tools must be run slower. That takes more time, and manufacturing time costs. Figuring out where you can compromise is how you drive costs down.

Another issue to consider:

The location of the board house. For instance, if the vendor is in China, it is going to be difficult to ship product back and forth quickly. Consider this scenario: Your boards are made in the United States and the organization that assembles and tests your boards is in China. Rather than shipping all of that product back and forth across the globe, it may be more cost-effective to find a single source for all your needs…or at least try to locate the two services on the same continent.

When embarking on a project, you are going to need to manage risk. If a manufacturer’s response to a specific requirement is wishy-washy, or if they can’t commit to the specs that you can’t budge on, move on. This isn’t bad manners, it’s business. You have to ask yourself this question: Can you and/or your organization afford to lose time while a manufacturer “tries” to meet your expectations? The only time “trying” is acceptable is when you are developing something new that hasn’t been done before. Or maybe if the manufacturer is offering a deep discount in exchange for very long lead-times, and you can afford to wait.

There aren't a lot of Yelp reviews for PCB manufacturers ( I checked). So unless you are personally touring the manufacturer’s facilities and have the bona-fides to judge what you see, you will have to rely on word of mouth and forum posts, and maybe the testimonies the board houses themselves post on their websites.

I am leaving out names here, but recently I went with one board house over another because of the images of their work I found on the net. The pictures I found showed that the house I went with produced much cleaner work.

Ordering the parts. I set out to get all the support components I needed: plenty of 0.1-in. spaced pin headers, through-hole diodes, USB sockets, and controller boards.

Pick any electronic parts distributor; they will have what you need (Element14/Premier Farnell, Digikey, Mouser, etc.). For the prototype, I used Amazon. With a smattering of the big parts companies and individuals selling components, I had no issue finding what I needed.

Ah…Amazon Prime two-day shipping, where would I be without you? Most of what I needed could get here in two days. But, one part, a Teensy 2.0 controller board, was a bit harder to find. eBay had a few, some odd sites here and there. In the end, I picked the cheapest one. Above are the diodes I picked, through-hole for ease of assembly. The header pins are from Amazon, too. However, they claimed the position of the plastic break-away pin guides were adjustable; they were not. Either that, or you had to be a monster to move the divider. I left them long.

(Note: I couldn’t find headers I liked that were cheap. I’ll keep searching, for sure.)

I realized that a final product should not use a teensy (or any) dev board. The dev board costs $15.  The final version must not hinge completely on finding some discontinued development board. Requiring the user to have to also buy another board is unacceptable. I could mount all the components for the AVR microcontroller onto my board directly, saving about half the price in parts. Even this is too much.

In our next installment, we'll move on to the build itself.