Power supply performance design

Every engineer wants to design the best performance for the system, power supply or product they are working on. But what exactly is performance design? For power solutions this may refer to the typical trade-off between size, weight, and power (SWaP), and even the cost factor (SWAP-C).

Design for energy performance (usually measured by efficiency metrics) focuses on optimizing the energy performance of the power supply, i.e. emphasizing operating expenditure (OPEX), which is the basic energy cost.

If the power solution is optimized for its form factor performance, this may reduce the maximum conversion efficiency, meaning that the optimization of the capital expenditure (CAPEX) design focuses on how to save upfront costs rather than reducing amortized costs through lower OPEX. This distinction is critical in applications where power OPEX dominates total cost of ownership (TCO), such as large data centers.

For unrestricted applications, power OPEX can be determined based on fuel, range, or battery life. Often, these limited energy sources will act as control factors to maximize system performance.

Therefore, engineers must first understand the relationship between power supply, load, and the operating environment before they can begin to understand which performance factors are the focus of optimization. In the case of power solutions, most design parameters ultimately focus on the design of thermal performance, such as keeping critical components below critical temperature thresholds under demanding operating conditions of maximum input voltage, full load and high ambient temperature, such as semiconductor junction, package surface temperature, printed circuit board or PCB temperature.

Handle uptime performance

If output voltage adjustment rate and accuracy are Paramount, then optimizing power control loop performance (feedback loop stability and load transient response) is a priority to ensure that power delivery is not unstable or unpredictable after sudden load changes or during supply voltage dips and surges.

When operational reliability is a top priority, the mission-critical performance indicator is the uptime of the application or system itself. In this case, the requirements of the system may even require sacrificing power and other equipment to maintain operation as much as possible, even if the operating conditions are out of specification. This is quite different from designing a power supply with built-in shut-off protection due to short-term overload, overcurrent, or overheating.

Although power and thermal metrics are not generally considered to be major bottlenecks in application performance, they become limiting factors in basic performance due to their basic physical properties. The reason behind this may be the maximum junction temperature of a power semiconductor component, the maximum current of a power cord, or a power inductor, in any case performance is ultimately limited by power or heat dissipation limitations.

Sometimes a derating of system performance is required to maintain overall insulation or thermal zones. For example, the processor is capable of handling additional million instructions per second (MIPS) or the radio has additional space to further amplify the RF signal, but the system lacks sufficient thermal management technology to allow for the increase in power consumption.

Emphasizing energy performance

Power is often not given much attention, which underestimates not only the complexity and special needs of power solutions, but also in terms of availability. As mentioned earlier with the thermal bottleneck, it is often seen that there is a significant gap between the peak demand for system load and what the power supply can deliver in order to save costs or compress the power supply into a smaller space. Ignoring loop control and transient design challenges (based on the scope discussed in this article), power gaps can occur when the safety margin of the power subsystem design analysis is too low, while not fully considering all loads provided by the same common rail, or even incorporating the upstream power rail into the larger power solution.

Most power subsystems involve multiple levels of voltage conversion, from off-line (ac) to intermediate bus voltage (usually 48/24/12 Vdc) to low voltage (usually ≤5 Vdc) of ASics and other logic circuits. In general, since the load current tends to increase as the bus voltage decreases, more attention is paid to the efficiency of power conversion solutions for low-voltage rails; In this case, the dissipation loss becomes more important and critical to the thermal performance of the overall system.

Even with the higher focus on load terminals, it's easy to overlook the impact of upstream power conversion solutions. It is therefore critical to develop an interactive model of the system's power budget that considers the load and efficiency curves of all power supplies, as well as the overall impact of transient performance from the terminal load all the way up to the upstream offline power supply.