How I made a cheap PCB
Even though I have a degree in electrical engineering, and even though I’m comfortable reviewing schematics, I haven’t done much circuit design in the decade that I’ve been out of school. The record was even worse when it came to printed circuit board design: I’d laid out only a single board, a single time, way back in 2003. I decided to change that.
Opportunity showed its charming face while I was working on a technical talk. I found myself in need of a high-bandwidth, high-gain, high-dynamic-range microammeter to do a specific type of analysis on certain devices. My usual go-to microammeter, a µCurrent Gold, was sufficient for preliminary work but fell short in the bandwidth and dynamic-range departments. Thus, I decided to design one myself.
The printed circuit board (PCB) design and manufacturing process turned out to be far easier and cheaper than I had feared. It took me about a day to learn how to use Eagle, which is the de facto PCB CAD program, another day to flesh out the circuit design and the board layout, and about four weeks of waiting for the boards to be made in China and shipped to me. The cost for printing 10 boards, including shipping? Just $14. Total cost for each board, including components, was about $6.
I half expected to get non-functional boards, or drill hits that were way off, or nothing at all, but instead, I got nice-looking boards with excellent registration and no electrical problems. For comparison, I could have had the same boards made at a plant about five miles from me, but the bare boards would have been at least $33 each, and the quality would have been no better.
The key was going through DirtyPCBs.com. They have some sort of deal with inexpensive Chinese board houses. As long as the board design is small (mine was 5cm x 5cm) and simple (2 layers), and you’re fine with receiving “about” 10 boards very slowly (1-8 weeks quoted; mine took 4 weeks), then I think they can’t be beat.
My layout job looks a bit amateur, and I made the anachronistic choice of going with several through-hole components instead of being 100% surface-mount, but the actual circuit works great. If my talk gets accepted, I’ll probably do another board spin to make it look nicer, but electrically nothing will need to change.
Cheap labor and heavily subsidized postage are an incredible combination.