Sam Jacoby

The journey continues. This is part of my work as a research assistant in the High-Low Tech group at the MIT Media Lab.

This is the design that I drew the board up from. Looks pretty good, right? Well, I was happy: lot’s of complicated little parts, colors—enough for me to feel like I was doing my job.

I sent off the gerber files, and had another board made by the good men and women of AP Circuits. They did bang-up job, naturally. I was walking around the lab with them, and Mark said, “That from AP Circuits?” You can spot one of these gems from across the room—it’s in the fancy-pants solder mask, apparently. Whatever the reason, you can spot one of these gems from across the room: circuit bling. I happily toasted it up.

The finished board, bristling with muxes and PGAs and all the rest. Looks pretty good.

Now, some of the savvier among you may note that there are substantial errors in this circuit. To you, I tip my hat. I was oblivious.

At the moment, our hero is plugging the board into his computer, ready to bask in the radiance of its green, power LED.

Nothing. Poke around a bit. Nothing. Pull out the multimeter. Could there be a bridge in thin traces of the mini-USB connectior? No. 5V streaming in. That looks good. But, hmm, streaming out? Ah…well, not so much. And then I see it. None of the grounds are connected. Only one half of my circuits are routed.

Here’s the story. It’s not so grand. At some point in the development process, I did something bad. To simplify my design, I used a ground plane—a single sheet of copper that the ground of all of the traces plugged into. After “pouring” the ground (kind’ve like the fill tool in paint, or something), the screen was looking a little crowded. I turned off its display. Great. In so doing, though, I also turned off the airwires¹ that it came with. Naturally, I then promptly forgot I did any such thing and proceeded on my merry way. Getting the board made and all the rest.

It turns out, though, the job wasn’t really done. The pouring algorithm had done its job, sure—but it had left a number of sections for me, it’s human-companion to resolve. The groundplane needed a number of additional airwires that I never added back in. My circuits, then, were a series of orphans, plugged into nothing but themselves! I needed 17 additional wires. I was crushed.

At any rate, lesson learned. Don’t switch off views and think you’ve solved any problems. You haven’t.