Jernberg pi
The Case of the Missing PDN Owner
With many disciplines contributing, who will manage the process?

As technology trends toward smaller, faster, cheaper, the challenges around good PDN design get more difficult. With multiple requirements needed from many disciplines, the PDN’s demands will only increase and become harder to maintain.

Over the past few months, we have discussed elements essential to power delivery and PDN requirements. Now that we have a better understanding of this, it’s time to explore what is needed to create the ideal PDN product, and who is best equipped to bring together all the elements of the PDN.

What is a good PDN design, and how do you achieve it? Power-related design objectives tend to be similar in nature for all PCBs: to provide sufficient current at a stable voltage to each device. What does vary widely is complexity, however. Said objectives can range from simple single-supply, powering a solid power plane, to a multi-source, hot-swappable, self-monitoring, thermally sensitive, complex design that accounts for most components and a large amount of copper on the PCB. Simply put, good PDN design delivers power adequately and reliably.

Achieving this is not as simple as it sounds, however. Multiple disciplines and requirements play a role in the success of the PDN and, ultimately, product functionality. These requirements influence other design decisions and, if not considered at the right time, can be almost impossible to address as the process moves forward. The PDN, both parts and etch, typically are the most difficult to change.

PDN parts, particularly magnetics, filters and regulator-type devices often come with placement guidelines from different disciplines that can be conflicting and may ultimately need mediation. While never an advocate of the cookie-cutter approach, my advice is wherever possible to “route power first.” Circulate it, review it, involve the other disciplines, and seek agreement. Use it for the canvas as you floor plan. Power problems are echoed on every signal on the rail and are a nightmare to fix.

In addition to the power delivery role, the PDN also typically carries the return current associated with the digital signals on the board. As a result, the copper shapes of the power nets are often influenced by adjacent signal routing, being reshaped to provide an uninterrupted return plane above and/or beneath routed trace. It is important to “size” the copper (i.e., define shapes and vias) based on simulated current density and voltage drop. This “best practice” can be achieved through visual inspection or with simulation tools.

Experienced design teams recognize solving power-related problems always involves “multi-physics” trade-offs, such as placing components far enough apart to permit heat to dissipate, but close enough to meet EMI requirements. This is in addition to any contributing factors that provide additional limits: for example, placing devices as close as possible for optimal signal integrity but not too close to prohibit rework.

Therefore, the PDN needs to be addressed as a system. Each individual power net, even if routed with exceptional precision, can be compromised if demands aren’t properly considered. Additionally, each net in the power system must be capable of delivering adequate DC current on its own, while limiting its losses to an acceptable level (voltage drop). Individually, the nets must also be able to deliver that current not just in adequate quantity, but with the necessary responsiveness such that the voltage is stable, even to high-speed transients.

Like a plumbing system, we must ensure each pipe is sized to handle the flow. Not only does this pipe need to handle its own requirements, but it must also account for the requirements of the additional sources connected to it. To achieve our DC goal, we need to approach the PDN similarly. Therefore, before making important layout decisions, we must consider the requirements of adjacent disciplines such as thermal, EMI, mechanical constraints, safety, etc., for the best possible chance at first-pass success (as all affect PCB layout).

Figure 1. All design requirements converge at the layout.

Figure 1. All design requirements converge at the layout.

Opportunity knocks. Everyone has a hand in the PDN, but no one owns it, unless you have an in-house power integrity engineer (FIGURE 1). Everyone cares about their own design requirements, but no one owns how they all come together. This lack of ownership is the root of many PDN issues. Each of the disciplines involved in product development has requirements that must be met and are often determined far too late in the design cycle. For better or worse, all requirements will collide at layout.

Figure 2. When electrical information is displayed on a board, the choices are data-driven.

Figure 2. When electrical information is displayed on a board, the choices are data-driven.

Figure 3. Example of power integrity used directly within the CAD tool.

Figure 3. Example of power integrity used directly within the CAD tool.

The PCB design engineer must either meet or manage every requirement for all disciplines, making them best positioned to manage the PDN (FIGURE 2). The impact of the PDN on the traditional PCB design process can enhance our ability to make informed decisions. For example, consider the task of splitting a plane to contain two distinct voltages. Now reconsider this knowing one of those nets carries nearly the full current of the power supply and the other carries but a trickle. With this knowledge, plane cuts would be different, affecting overall board performance.

Within the design process, every action has a multi-effect that needs to be balanced, and as time goes on its importance will increase. Someone needs to take on the role of managing these actions, and the PCB design engineer is the best person to do so. PCB design engineers understand how to balance design requirements with manufacturability and have the most influence over the success of the PDN. The multitude of tools available directly within the PCB canvas enables PCB design engineers to seamlessly incorporate both signal and power integrity within their design process, while minimizing time concerns.

The PDN is evolving. We must evolve with it. We’ve always had power, so what’s different now? The PDN is evolving, and margins are shrinking. Operating voltages dropping from 1.8 to 1.5 and then 1.2 as the input current doubled for memory is a hallmark example of this evolution. As complexity increases, not having an owner increases the potential for failure.

The PCB design engineer has the most influence when it comes to ensuring good power delivery, as they control every element needed to build the PDN. The PDN is essentially a physical problem, and while it can be planned early in the design process, full definition and realization doesn’t occur until PCB layout.

Most companies don’t have the luxury of a designated power integrity engineer. As the needs of the PDN evolve, we must evolve with them. With the role of the electrical engineer morphing with the PCB designer, now is the ideal time to take this opportunity to incorporate power integrity with minimal disruption to the existing design process.

Terry Jernberg Headshot
Terry Jernberg
is an applications engineer with EMA Design Automation (ema-eda.com), with a focus on PCB design and simulation. He spent his early career on signal integrity simulation for the defense industry and was fundamental in the adoption of these tools at EMC and Bose. A vocal advocate for simulation, his enthusiasm for physical modeling has expanded to include power and thermal capabilities.