Printed Circuit Boards Part 1: Printed Circuit Board Basics
Printed Circuit Board Design
Suppose you decide that you want to design a product and need to make a custom printed circuit board (PCB) which needs to fit in a certain location within your product. If you are an experienced electrical engineer, then you may already know where to start. Sometimes the considerations are also thermal, which would require a mechanical engineer. However, if you are trained in another engineering discipline, or a non-technical sort, or just need some general knowledge, then the following blog may be useful to you.
Electronics Thermal Management
The need for a PCB assumes that you already have a circuit designed. One can use engineering experience, coupled with a SPICE model in software to understand how the circuit theoretically works. The actual mathematics and physics are very complicated, but have been reduced to modules that can be plugged into the software for analysis. Most components have SPICE model. IC thermal management is very important for reliability of components, but when they are integrated on a PCB, the thermal analysis needs extension and is usually known as “electronics thermal management.” Thermal analysis requires other software to understand the generation and dissipation of the heat generated by components. Heatsinks, cooling fans, and other methods are used to keep PCB and the components within their operating temperature range. A mechanical Engineer takes into consideration strain relief, shock mounting, vibration effects, and related issues. PCBs can be tested thermally, and on shake tables. This is sometimes called “shake and bake.”
The most basic aspect of PCB design involves making the devices fit on the board, and then connecting all of the relevant pins. Each electronic component manufacturer, be it a microcontroller or a programmable logic device, would already have a reference design which is always a good place to start. Once you settled on a circuit and component list, the next step is the design and routing of the board. Each component would have a certain footprint, such as DIP, SOIC, BGA, etc…, with a various number of pins. Your CAD system is more than likely to have the part in its library. If the part is new, one can usually make a new package footprint within the software being used. Choosing the number of layers of board depends on many factors, not the least of which is the number of outputs on the components with largest number of pin-outs, such as a central processing unit or a digital signal processor.
Interconnecting traces that are routed through the PCB requires both good engineering practice, and specific knowledge of the devices, frequencies, and sensitivies to interference, cross talk, and noise. Sensitive routes in some applications may require shielding which may also increase the number of layers needed. Power components would also require extra attention, if integrated within the same board along with low-power electronics. The most common configuration is the two layer board, which has a core FR4 (fire-retardant) layer and two metal layers on each side. This is often coated with the dielectric layer (solder mask) once routes are fabricated in the metal layers. At this stage, once the shape of the board, number of layers and component footprints are locked-in, one has to place the components and draw up the wires or routes. The unconnected routes would crisscross throughout the board forming a “rat’s nest”. Design packages make auto-placement and auto-routing much easier than in the past, but some clean-up and manual routing for some nets is still often required. Extra metal in boards may be required to dissipate heat and control warping of the boards, and the design engineer must be aware of the requirements.