Gary Griffin, President of Fast Interconnect
After having the privilege of managing PCB designers for over 30 years, I
have determined that common sense is not usually their strongest point.
I don't say this to insult anyone. It is merely an observation. Even the
most seasoned designer needs help now and then, if they are only willing
to ask for it.
So, with that said, I'd like to offer a few common-sense tips for PCB
designers. First, I would have to say that the most common mistake
made by designers is that they end up designing themselves into a
corner, much like a novice house painter. Then, it is usually suggested
that the designers delete most of their work and approach the design
from a different angle.
The best person to provide such advice is usually a non-designer who is
familiar with the project, because he can usually see the bigger picture
better than a designer who has been staring at the computer screen all
day long.
Problems like this often arise when designers attack the easiest routes
first. In fact, the most critical trace routes should be considered and
routed first. We all know that PCB layers get expensive on production
runs and the customer wants to save layers to cut costs, which end up
increasing the design complexity and density. Thus, the most critical
routes should be done first, and everything else is secondary.
Today's PCB designer has access to some very good design tools,
unlike back in the old days when we were forced to use architectural
design packages to lay out boards. Current tools allow the user to shut
off the rats (the little lines that look like string art and show what is
connected to what) that you don't want to focus on. This allows the
designer to focus on critical routes without a lot of unnecessary
clutter.
For complex designs, it is also a good idea to take the time up front to
select rat net groups and color them. This allows the user to easily view
groups of nets, thus helping the designer focus and increase efficiency.
For designs that require controlled impedance, always contact your PCB
fabrication house prior to routing the first trace. Ask the engineering
group or manager for the information necessary for setting up trace width
and spacing as well as the layer stack. I have seen projects fail because
this simple action was not taken, and plenty of time has been wasted
going back to adjust trace widths and spacing.
And it is best for the PCB designer to speak directly with the customer.
This way, the designer gets the info he needs quickly, plus it decreases
the chance that data and answers to specific questions will be
reinterpreted by a third party. Communication is paramount to a
successful, on-time design, and allowing the designer to speak directly
with the customer only makes common sense.
Some projects are over-engineered during the initial design cycle. It is
the responsibility of the customer and the design manager to recognize
this and come to an agreement before the endless cycle of design
enhancements take place, often do nothing to make the product work
better.
As the design nears completion, it's important that the designer, the
project manager and the customer all agree on the end results. I was
once told by a designer with many years’ experience that designs are
never completed, just abandoned. This means that, given the chance,
the end-customer (often an engineer) will see areas that can be tweaked,
re-done, removed or replaced. This over-engineering cycle must be
stopped at some point so the project can be completed.
The common-sense rule applies here: Good enough is indeed good
enough. A first-time project will require some tweaks or corrections
before it goes to production. However, if the prototype is never
completed no one will know what these tweaks should be. Engineering
reviews cannot possibly catch everything that may go wrong in a final
product. Actual testing must be conducted in order to determine the most
effective changes. More often than not, these changes end up being very
simple and very quick.
And lastly, do not fall for the old "The prettier it is, the better it will work"
scenario. This is never the case. The customer deserves a good-looking
product and is paying for that; however, aesthetics and functionality do
not necessarily go hand-in-hand.
An efficient design is a pretty design. A design that works is beautiful by
any standard.