This week in I2CDriver

The I2CDriver campaign ended last week, with nearly 400 backers helping to make it happen. That's really encouraging! Last week I sent the test stands over to the manufacturer (Boktech) in Shenzhen, and this morning I got the great news that everything is looking good. When things go wrong with the manufacturing and test, it's usually pretty awful. The distance and time zone differences make remote debugging quite painful. The Gameduino 1 and 2 had quite serious setbacks that nearly made them miss their delivery dates.

Here are the images - fresh from Boktech's bench yesterday.

The manufacturers first make the bare PCBs - whatever quantity you've ordered. I'm making 1000 of these boards. Typically boards are made in arrays, so they can be assembled in bulk. For example Gameduino 3 arrived in panels of five PCBs. The I2C modules here are being done in batches of two -- that's all up to Boktech; they decide on the optimal size. In addition, this design is really 10 smaller designs, that I have arranged in a 5x2 grid:


The modules each have four castellated holes, something I haven't tried before. It complicates the manufacturing process, and results can vary in quality. This is what castellated holes should look like - but the prototypes I ordered didn't look as good as this.
From the image above they're looking quite clean, again thanks to the engineers at Boktech.

After the boards return and all the parts arrive, the manufacturer could just manufacture the full run. They don't like to do this because of the risk of making a huge run of useless parts. So instead they assemble a single unit using the production PCB, paste stencil, and parts. This catches a bunch of possible problems:

• the design is incorrect (duh!)
• the tester doesn't work with genuinely blank parts
• any parts are fake or bad

So a lot of things can go wrong at this point! I was really worried about this particular one because the product is actually ten different modules -- there are a lot of things that can go wrong. Here's the sample that Boktech made yesterday:
And here is it sitting in the tester that I sent last week, with the engineer holding it down against the tester pogos.

As you can see, all ten indicators are green on the tester display - which means they all passed! This is is really good news, as any problems this close to Chinese New Year would mean that the project would slip into late February or March.

Thanks for reading. Next week I'll talk about some of the I2C firmware on the board.