Back to Basics: A Guide to Different Power Architectures

March 2, 2015

architecture imageThere are a several different approaches to developing power architectures, each with its own benefits and disadvantages. It’s even possible to use the same component in different architectures. Choosing the right approach is critical, as the architecture selected frequently plays a major role in determining the performance and cost of the system. This is particularly true of modern systems, where the trend is towards lower voltages, higher currents, higher speeds and more on-board voltages.


Centralized Power

The centralized power architecture (CPA) is one of the oldest, and simplest ways to develop a power system. All voltages are generated at a central location and distributed to loads via buses. Providing that currents are low, voltages high and the distance between the power supply and the load is small, this can be an effective approach.

Figure 1 - Centralized Power Architecture (CPA)

Figure 1 – Centralized Power Architecture (CPA)

In modern systems, however, voltages are frequently low, currents high and power must be distributed widely. The I2R power loss in the busses becomes unmanageable, requiring impractically large bus cross-sections. With all aspects of the power conversion located at some distance from the load, the ability of CPA systems to respond to transient power demands is seriously limited.

Systems built using a CPA are also highly inflexible. As one power supply is used to generate all the voltage rails, it can be costly and time consuming to redesign it to meet a change in requirements: for example a particular voltage rail requiring additional current.


Distributed Power

In the early 1980s distributed power architecture (DPA) evolved to take advantage of the availability of new power devices, bricks. By breaking the DC-DC conversion into building blocks, isolation, regulation and voltage conversion could be performed near the point of load, allowing power to be distributed in the system at a higher voltage.  This dramatically reduces the loss in the busses as the current is reduced.

Figure 2 - Distributed Power Architecture (DPA)

Figure 2 – Distributed Power Architecture (DPA)

As systems began to require an increasing number of voltages on-board, engineers were forced to use more and more bricks, which meant a significant increase in the real estate required and the system cost. Furthermore, even though the power conversion is closer to the load than in CPA, DPA systems using bricks are often unable to deliver the speed of response required for modern loads.


Intermediate Bus

Some of the drawbacks of DPA were addressed by intermediate bus architecture (IBA).

This approach is a development of DPA, where a semi=regulated bus – typically at 48 V – is used to distribute power in the system. Intermediate bus converters (IBCs) are used to generate a voltage to supply the board and small point-of-load converters (PoLs) are used to generate the different voltages required.

Figure 3 - Intermediate Bus Architecture (IBA)

Figure 3 – Intermediate Bus Architecture (IBA)

In IBA, the IBCs provide isolation as well as regulation and conversion, allowing lower-cost non-isolated PoLs (niPoLs) to be used to optimize the use of board space and system cost.


Factorized Power

Factorized power architecture (FPA) was developed by Vicor, and is the newest of the power architectures. In this approach, power is still distributed using a DC bus, but the regulation function is separated from isolation and voltage transformation. The main power distribution bus is regulated by a pre-regulator module (PRM), and then the voltage is isolated and stepped down at the point of load by a voltage transformation module (VTM).

Figure 4 - Factorized Power Architecture (FPA)

Figure 4 – Factorized Power Architecture (FPA)

The FPA approach offers a number of solutions to challenges faced by power system designers. Firstly the VTM is highly efficient, even when operating with a large step-down ratio. This allows the low voltages required by many modern semiconductors to be generated with minimal energy loss. The regulation stage (PRM) can be located remotely, saving board space, and the outstanding transient response of the VTM allows bulk capacitance to be placed before this power component: the reduction in required capacitance can save space, increase reliability and reduce cost.

Factorized power systems can also use a combination of a bus converter module (BCM) and standard niPOLs. Here, the BCM performs isolation and transformation functions, while the niPOLs regulate and transform the voltage. As with a PRM/VTM combination, regulation and isolation are separated.

Power system architecture have evolved over time to meet the ever-increasing demands of the loads that must be powered. The introduction of power components has enabled approaches such as FPA, and made developing complex power systems much quicker and easier. Today online power system design tools allow the creation of complex power chains with a click of a mouse, eliminating much of the hard work of architecture selection and power chain design.


Further Information

Tags: , , , ,

Comments are closed.

Find out more about our Cool-Power Buck Regulators DCMs - A better Brick subscribe to vicor newsletter Contact Us

Get Connected