2
\$\begingroup\$

Is there somewhere a description or flow-chart of the process of designing a circuit: Requirements analysis, schematic capture, simulation, PCB layout, etc. as well as various checks? Something especially for PCB design, and not so much about IC layout please.

\$\endgroup\$
12
\$\begingroup\$
  1. Agree on functional operation and performance specifications
  2. Start designing the hard bits first and use a simulator
  3. Make sure parts are available
  4. Go back to (1) if the task is impossible and redo the spec
  5. Start integrating the simulations to ensure different sub-systems of the circuit work together
  6. Go back to (1) if the task is impossible and redo the spec
  7. Start developing the schematic that is your definitive circuit
  8. Choose the rest of the parts and make sure the footprints you have are double-checked or have been successfully used before on previous PCBs
  9. Start designing the PCB and if can't be done in the space available go back to a relevant earlier stage such as (1) or (2)
  10. Complete layout and if you have the tools, and is deemed necessary simulate the effects that might make the circuit layout problematic such as when designing RF stuff.
  11. If PCB has to meet safety creepage and clearance requirements then get it double-checked
  12. Order PCB, order parts, get prototype built
  13. Test it to spec agreed in (1) - this could involve many many tests and there are probably reams of information that could be said here but space and time limits me!
  14. Go back to relevant earlier stage if specifications are not met.
  15. This is usually called Design verification
  16. Create/modify the production documents including a production test spec
  17. For first run of production, ensure Design is validated (extra special care to make sure results are exactly as verification results and if not go back to an earlier stage)

Hopefully, by the time you get to this point you'll have a pretty good design. Your PCB is fundamental to the schematic despite it never appearing on it (well maybe fixing holes etc). The PCB touches every single electronic component in your design - very important to do the job to what was agreed on day 1 (or amended via the process).

I'll also add that this is just off the top of my head and, there are plenty of variations on this such as when designing specifically for an end user - obtaining customer approval at several points along the way is a necessary part of the process and this should be agreed up-front. Also, some products require 3rd party independent testing to "prove" they comply with regulations such as EMC, safety etc..

Between #5 and #6 it is worth mentioning that bread-boarding certain parts of the design may be done where it is felt that the simulation may not give the full picture. This is definitely not uncommon and, designing a PCB just to rigorously test certain parts of a design, is a valid step along the way even though the bread-board/PCB may never get used on the "real" job.

Before #15 you should also start to consider (when production run quantities are not trivial) placing orders for production parts taking into account lead times. You should also be starting to think about any special test fixtures that need to be introduced and doing this in enough time so that the first proper production run is not delayed.

\$\endgroup\$
  • 1
    \$\begingroup\$ fda.gov/ucm/groups/fdagov-public/documents/image/ucm070635.gif. The FDA requires a "waterfall" design process, which nicely encapsulates your list. Of course "documenation" runs in parallel to the process \$\endgroup\$ – Scott Seidman Jan 15 '14 at 23:29
  • 1
    \$\begingroup\$ It's probably worth noting that it's very common for all sections where you mention use of a simulator to be replaced with "breadboard/have test-boards made". Depending on what you're doing, test hardware can often be easier to produce then a good simulation. \$\endgroup\$ – Connor Wolf Jan 16 '14 at 10:18
  • \$\begingroup\$ @ConnorWolf #13 does this but there are good reasons for breadboarding parts of designs at earlier stages especially if it's difficult to simulate. I'm doing a board right now that will never see the outside light of day but, for convenience, a lot of sub-circuits were grouped together just to see that they worked together. It's still a PCB but can be "called" a breadboard. \$\endgroup\$ – Andy aka Jan 16 '14 at 10:23
  • \$\begingroup\$ I've worked in a (somewhat) high-performance analog shop before, and at least the shop I worked at only used sims for logic verification. All the analog hardware was prototyped by just having small boards made. Parasitics and layout concerns were too major of a factor, and could not (and still cannot) be adequately simulated. Also, the hardware was for interfacing with transducers for which there were no models, and couldn't really be simulated anyways. \$\endgroup\$ – Connor Wolf Jan 16 '14 at 11:05
  • \$\begingroup\$ FWIW, I'd call test PCB's "breadboards", together with stripboard and a whole menagerie of similar one-shot construction methodologies (ex: wire wrap!). \$\endgroup\$ – Connor Wolf Jan 16 '14 at 11:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.