[Electronic Hardware and Software Design Solutions]

The Three Ps for PCB Design Success

Randy Nuss
Copyright 1999 Idea Consulting

Over the course of 15 years and many board designs, I have learned a few lessons, most of them the hard way. There is nothing more disappointing than getting a call from manufacturing 4 weeks after signing off on a new design and discovering that one or more components "don't fit" on the prototype boards. Or worse yet, to find a power plane short in an otherwise perfect design.

Lets face it; everyone makes mistakes. Despite the best of intentions, all who participate in the design process will continue to make mistakes. Given that mistakes will be made in the design of an electronic assembly, we must attempt to create a design process which finds and corrects these errors before they find their way into the manufacturing process.

The engineer must take responsibility for success or failure of the entire design process. Problems with an assembly are never the fault of the PCB designer. Why? The engineer is responsible for checking the work of the PCB designer. Only through careful checking at each stage of the process can the possibility of producing first time success be achieved. CAD based design-rule checking does find and correct many errors, but there is no substitute for careful checking by the design engineer. Checking isn't glamorous and usually not that much fun, but it will save a lot of hair-pulling later.

Electronic design textbooks describe the means for selecting component values based upon a particular circuit topology, but they rarely, if ever, describe the process for ensuring an error free PCB design. To find and eliminate errors introduced during the design process, it is useful to practice the three Ps of electronic design. The three Ps stand for PACKAGE, PINOUT and POWER.

Package

The first P is for Packaging. Electronic components are generally available in several package types. The most common differentiator is whether a part is through-hole or surface-mount. Many surface mount memory devices for example, are available in different widths. It is vital that the design engineer selects and documents exactly which package will be used in a particular design. It is critical that the engineer also verify that the PCB designer has built the CAD part outline correctly.

Relying on the PCB designer to build the CAD part outline correctly is a recipe for failure. As stated before, PCB designers (being only human) make mistakes. There is no DRC program that will verify that a part outline matches the specification provided by the manufacturer. The engineer must check every outline particularly new part types..

Scrutinize new component types. In general, components which have been used on previous assemblies will be error-free, so it is important to spend the most time checking the new components. Have a current Bill of Material for the assembly which includes the exact manufacturers part number. Use that number to locate the physical drawing on the manufacturers spec sheet.

Most PCB design programs have a way to verify dimensions, and its simply a matter of sitting down with the PCB designer and the spec sheet and asking him or her to measure a few dimensions. The whole process goes relatively quickly. Generally, there are several component types which share the same footprint such as 0805 resistors and capacitors. These commonly used components will almost always be correct since their use is quite common. Spend more time on connectors and other parts whose package dimensions vary from vendor to vendor.

Remember, the part that purchasing buys must match the part you intended which must match the part whose outline is on the PC board. Its easy to get it wrong so CHECK THE PACKAGE!

Pinout

OK, lets assume the packaging is correct. What about the pinout? How do you know the CAD designer who created the schematic body didn't make a mistake when assigning the pin numbers to the schematic component body. This can be particularly troublesome with large pincount devices such as processors and special purpose gate arrays. Pincounts on many devices are in the hundreds with thousand pin devices on the horizon.

Debugging problems with component pinouts can be extremely frustrating and time consuming. It only takes one pinout problem to mess up an otherwise working circuit. The circuit may fail in some unusual way which makes it hard to trace. Or worse yet, the defect may remain latent, waiting for a particular state of the hardware to exhibit a failure.

In any event, pinout problems are easily prevented. When submitting a device to CAD for inclusion into the CAD library, insist that you (the engineer) be required to check it before allowing its use. In this way, the engineer can prevent disaster by finding and correcting pinout problems before the components make it into the library. Never assume the CAD designer built the part properly. Check it yourself.

Many companies have an electronic design process which requires sign-off by engineering for any new part to be added to the library; all should. Trust nobody, CHECK THE PINOUT.

Power

What happens if some of the pins on a large IC are properly connected to their power supply, but others are accidentally left unconnected? It might work. Often, power supply connections are neglected, and errors are not discovered until the debug process. Correcting power supply problems through rework can be uglier than correcting simple signal flow problems and generally result in a lower quality power connection. The sad truth is that errors in the power distribution to components on a PCB assembly are amazingly common. Why should this be?

First, power connections are customarily not drawn on schematics. This is generally done to prevent clutter. In most CAD systems, power connections are pre-programmed into the component as properties and are difficult to alter. This can lead engineers into a false sense of security; out of sight means out of mind.

Secondly, engineers consider the power supply connections as mundane compared to the "glamour" of high speed state machine design for example. There is a tendency to not want to bother with verifying the power supply connections.Everyone generally concentrates on signal flow. It is assumed that the power connections are correct.

Checking for power supply problems occurs at three distinct points in the design process. First, verify that the power pins will be correctly connected to the appropriate supply when checking the CAD schematic body pinout. The power supply pins will generally not be shown graphically, so check the device configuration file. Verify each pin's pin number and power supply connection against the data sheet.

Carefully check your schematic for any power pins which must be manually connected. Op-Amps, comparators and many analog components use graphical power pins as their power supply connections may use a variety of voltages. Make sure that bypass capacitors are present on the schematic.

Finally, check the PCB artwork for proper power supply connection. Ensure that conductors are wide enough to prevent excessive temperature rise or voltage drop. Long, narrow power supply traces can lead to ground bounce and a variety of other, hard to diagnose problems. Ensure that adequate capacitive bypassing is present and properly placed as well.

Make sure that there are no power plane shorts, by overlaying the power plane film and drillmaster. Check that for each hole, there is at most 1 power plane connection. This vital check prevented disaster several years ago when a PCB designer created a component but forgot to put clearances on the power planes. If this board had been fabricated, a direct short from +5V to ground would have rendered the boards as scrap.

The bad news is that plane shorts "designed in" to a board are almost impossible to rework without a lot of drilling, and even then the boards will never be reliable. The good news is that plane shorts are easily found at the film stage. Its easy to do, CHECK THE POWER.

Finally

So there they are, the three Ps. If you use the power of the three Ps, I guarantee you will have much better success with PCB designs. It really comes down to attention to detail. No college class ever taught me that engineers had to check so many things. Other engineers taught me, mistakes seen and mistakes made taught me. Expect to find errors and you shall most certainly not be disappointed. Happy hunting!

HOME

Copyright 1999 Idea Consulting