Comments on: Programming the CupCake Motherboard and Extruder

By: Keith Neufeld

Keith Neufeld — Thu, 03 Sep 2009 14:25:21 +0000

Here’s how we avoid this at work.

The first person to develop a new procedure takes notes onto the internal wiki as they’re researching and exploring. When they’ve refined the process, they edit the notes into a set of instructions.

The second time the procedure it done, it’s strongly encouraged to have a different person do it. Regardless of who does it, they follow the wiki instructions strictly, to identify gaps, errors, and other shortcomings. Because it’s a wiki, they correct the instructions on the spot, as they go through.

Every subsequent time the procedure is performed, it’s done by following the instructions, even if it’s routine and commonplace. It’s too easy to get distracted, called away, or whatever, and skip a step; then you deliver something to the customer (in my case, thousands of people who use the university network; in MakerBot’s case, people who build the CupCake) that doesn’t work. For procedures that are totally routine, we may make an abbreviated checklist at the top to streamline things, but we still use at least the checklist.

By: Kevin Reid

Kevin Reid — Thu, 03 Sep 2009 13:02:52 +0000

It’s easy to develop habitual procedures and not think at all about that they’re not documented for others.

By: Keith Neufeld

Keith Neufeld — Thu, 03 Sep 2009 10:52:33 +0000

Steve, I take your point; but it’s dulled by the facts that MakerBot seems to be building a CupCake for their own use from almost every batch and if this is a general problem they must have run across it themselves, and that it keeps coming up in the forums but no one has done anything about it. That’s why I think I must be missing something and it’s not really as simple as outdated/missing instructions.

By: Steve

Steve — Thu, 03 Sep 2009 03:40:54 +0000

This sort of thing happens a lot. You design something, document the steps to build it, then add a feature after the fact. Problem is, the new feature changes the build sequence in unexpected ways. It’s likely that the MakerBot folks never went through the assembly instructions with the final version of the board and just assumed that it would still work.

There was a similar issue with the RepRap Darwin instructions where one of the test steps would fail intermittently because of a floating input on the MCU. At the time the instructions were written, that pin wasn’t an input. No one ever went back to verify that the test still worked with the latest firmware.