The Design Process: How to Get from Nothing to Something - Building Open Source Hardware - page 66

record what you built, what you changed, and why you changed something between
design iterations, it will be much easier to step back and revisit old designs. Formally, this
information is collected in a design history file . You can keep track of this information on
paper or in project management software, or you can even keep your design files and
notes in a version control repository. How you manage documentation isn't as important
as the fact that you do documentation. The lessons you learn during the testing phase
might be fantastic inputs to your product's user manual . You might discover a new
design requirement from the user beta testing. Either way, having that information con-
veniently captured in a design history file will give you an edge on both the competition
and the inescapable forces of human error.
Every Master Plan Has an Exit Strategy
Normally, some external force (like a deadline) will compel you to wrap up your project
before it's perfect. Even without a deadline, constraints always crop up. It's impractical to
run a project under the assumption that all features, all requirements, and all details are cre-
ated equal. If you are forced to wrap up with details left unaddressed, then make an errata
sheet, clean up your manufacturing package as best as you can, and ship it.
Focus on Critical Features First
If at any stage of your process, you can't build a prototype circuit that satisfies your bare
minimum requirements for functionality, go back to the drawing board and question your
assumptions. Rope in other engineers. Be inventive. But if you revisit your scope and you
still can't come up with a design that works, and you're frustrated and taking stabs in the
dark, consider shelving the project. Sometimes as engineers, we have flawed premises, or
we need more domain knowledge, or the integrated circuit we need doesn't exist yet. Either
way, it's vastly important to recognize when you're spinning your wheels and call it quits.
Hail Mary Plays Only Work in American Football
If you are forced to wrap up a design that works functionally but has unaddressed man-
ufacturability or cost problems, it can be worth your time to redouble your efforts, solve
the issues, and get that project out to production. However, if you've gotten to the final
last-minute deadline on a project, you have a design and fabrication files for a board
that's untested, and you're about to place a $30,000 bet on getting something that works
first-time-right— don't. Your best bet (if you absolutely have to ship something) is to step
back and ship the last version that worked. Sending untested hardware to production is the
stuff that nightmares are made of.