Email to the optic@ list: http://wifisoft.org/pipermail/optic/2008-February/000075.html

I did however learned a lot this day (and the ones before) which I of
course like to share:

* RTFM, building electronics is a bit more difficult you could not skip
  a step and expecting it to work fine when the follow ups are ok.

* Test, test, test. Did i say test? Double,triple check your circuit and
  make sure to put visual and electronic tests into the building manual
  and run them. When designing something provide marked measure points
  which are easy accessible.

* Secure you electronics by using a voltage and current controlled power
  supply or else you will most likely produce the so called 'current
  aroma' when making a slight measure/connection error.

* Do not trust your soldering skill: when soldering a type
  component or board for the first/second or third time make sure it
  gets checked by someone else. Tracking down solder errors on a fully
  equipped board is pretty hard and time consuming.

* Do not use hardwire when connectors are provided: It might sound
  tempting to rush into it and just skip the connectors and hardwire it
  directly. Two reasons: It's difficult to debug one part of the board,
  it gets almost impossible to debug the whole board at ones. And second
  its a hell of a job to place connectors on the sometimes almost fully
  tinned holes.

* Mind the ground: It's tempting to use the negative and ground together
  while designing the board, but you could get into a lot of trouble. A
  simple example having power off a lamp/solder station did make my
  debug signal much more clear.

* Know your electronic components and combinations or more sure you know
  where to find references to it; It gets quite useful to recognize
  all components on the board, their identification, pin layout,
  placement and characteristics.

* Stick to the plan: It might sound tempting to make some inline
  hacks/improvements/changes during the build phase, but rather make a
  working version first and then start hacking.

* Document changes: Building might be fun, it gets even more fun with
  proper documentation. This include measurement results of your tests,
  any given problem and the solution and of course status. This will
  allow you to work faster and prevent errors on the long run. You will
  need to be systematic and spend at least 25% of your time on
  documenting. Without this your gain knowledge and newborn skills will
  never be exposed to the world.