Power Integrity Analysis and Management for Integrated Circuits: Power, Delivering Power, and Power Integrity
1.3 Power Delivery
Electrical power is delivered in two distinct ways, through conductive coupling and through electromagnetic coupling. In conductive coupling, power is transferred through electric conduction or physical transfer of charges from one body to another through a conductive medium. In the electromagnetic coupling method, power is transferred through electromagnetic induction or communication, where charges are induced to move in the presence of electromagnetic energy.
This distinction is useful in differentiating DC power from AC power; all direct current power is delivered through conductive coupling, while AC power may be delivered through either method. Examples of AC power delivered electromagnetically include Nikola Tesla’s 1893 demonstration of wireless energy transfer used to illuminate vacuum bulbs, and William C. Brown’s demonstrations of power transmission using microwaves from 1961 to 1964. Ordinary transformers are everyday examples of AC power delivery through electromagnetic coupling. Electrochemical batteries and wires connecting to a circuit are demonstrations of conductively coupled DC power.
1.3.1 Central DC Power Delivery Module
Figure 1-1 shows a power supply module integrated in a personal computer system. Most such power supply modules in stand-alone electronic systems convert available AC power into a number of distinct DC voltages that serve other modules in the system. In the illustrated module, for example, DC voltages of +12 V, –12 V, and 5 V are generated from AC power entering through the chassis-mounted socket seen above the switch. These voltages power subsystems such as a hard disk drive or a microprocessor motherboard. Such a power delivery system is central, developing necessary DC voltages at a dedicated location and distributing them within the system through wires seen leaving the central module.
Figure 1-1 Example of a power supply module in a personal computer conversion.
Source: Author mboverload, Wikimedia Commons. 
Figure 1-2 shows a motherboard seating a microprocessor, memory, and peripheral cards in a typical personal computer system that integrates the power supply module seen in Figure 1-1. An output connector from the power supply plugs into a connector seen near the middle of the left side of the board. Just below this connector, the board seats a microprocessor on an approximately square socket. To the right of this socket, there are electronic components—transistors, magnetic-core inductors, and electrolytic capacitors—that perform a further conversion of the DC power supply into a voltage that the microprocessor can use. Such power conversion at or very near the receiving component is called point-of-load (POL) power conversion.
Figure 1-2 Personal computer motherboard housing a microprocessor and including adjacent DC-to-DC conversion.
Source: Gary Houston, Wikimedia Commons. 
This concept originates in the days of the “War of Currents” between Tesla and Edison, when Tesla showed that transporting remotely generated AC power at high voltages, converted to lower voltages where necessary, rendered the distribution of power over large geographic areas very feasible. Tesla solved the problem of energy loss due to high currents in the transmission pathways; a similar problem is solved in electronic systems by the use of POL power conversion. Microprocessors, operating at voltages approaching 1 volt, consume large amounts of current, of the order of 100 amperes or more. If transmitted through the wires of the power supply module, this would lead to unacceptably large voltage and energy loss among other difficulties. In the subsystem of Figure 1-2, power conversion takes place adjacent to the microprocessor, ensuring that high currents only flow across a very short distance of board interconnect.
Microprocessor POL power delivery is commonly performed through voltage regulator modules (VRMs ) that communicate with voltage-level control logic within the chip. A VRM is essentially a voltage down-converter, also called a buck regulator, that obtains a DC voltage of 12 or 5 volts and converts it to the voltage required by the microprocessor. This conversion is accomplished through high-efficiency switched voltage regulation, as discussed in Chapter 3, Section 3.1.2. VRMs may be replaceable or may be soldered to the motherboard, and are often optimized to work with a specific microprocessor.
1.3.2 Integrated Power Delivery
Integrated power delivery is a method whereby power conversion and delivery are integrated with load circuits. Methods of integration vary; examples include in-package voltage regulation,3 monolithic integrated power conversion,4 and three-dimensional integration of power delivery circuits with load integrated circuits.
Advantages of integrated power delivery include extremely short lengths of interconnect between power conversion and load components, resulting in reduced energy losses, as well as the potential for faster, symbiotic functionality of the power conversion system and the load device. In recent years, integrated power conversion and delivery has enabled substantial energy savings for integrated circuits through techniques such as dynamic voltage scaling and adaptive voltage scaling. Integrated power delivery assists in improving power integrity.
1.3.3 Power Distribution Networks
A power distribution network is formed by the interconnection of electrical devices that transfer power from a source to a load. Its principal function is to effectively transfer electrical power from the source to the load, doing so with minimal energy loss and minimal degradation of the power delivered.
As discussed in Section 1.1.1, energy loss in its simplest form is encountered in overcoming opposition to the flow of charge in conducting materials. For a simple direct current source and load, the power distribution network is designed to minimize resistance in the connecting electrical pathways. Depending on the nature of the source and load, more complex power distribution networks are often required. For example, the DC energy source may be derived from an AC input, through rectifiers that make current flow unidirectional. Voltage varies substantially in such a DC energy source, due to the sinusoidal variation of AC voltage waveforms from 0 to peak amplitudes. Such variation may well be unacceptable to the load, and the rectifier’s DC output may require further conditioning. This conditioning is performed by filters and supply decoupling devices. These devices block or bypass variations in energy transmission and smooth out power supplied to the load. Filters and decoupling devices are comprised of capacitors and inductors; capacitors, most commonly employed as supply filtering or decoupling devices, are discussed in later chapters.
1.3.4 Power Delivery Regulation
A key aspect of power delivery is providing electrical power in a controlled or regulated manner. Electrical circuits perform predictably when the supplied voltage and current are regulated according to their needs. Voltage regulation is most common in DC electrical systems, permitting load devices to extract as much current and power as changing conditions may require. Some applications, such as the charging of electrical batteries, require regulated current flow.
A voltage regulator is in many ways similar to a voltage source, such as a voltaic pile, that provides a fixed potential difference. The regulation function keeps the output voltage constant while load currents change, whereas the output of a typical voltage source drops by the product of the drawn current and a finite impedance intrinsic to the source. As drawn load currents increase, a regulated voltage also drops in value but to a much smaller extent. This characteristic is represented in the definition of load regulation, which is the ratio of the variation in output voltage, from minimum to maximum drawn load current, to the nominal voltage. Load regulation is an indication of the load-bearing quality of the voltage regulator. It is improved through feedback of the load voltage.
Voltage regulation, particularly in DC systems, is accomplished either by switched conversion or linear regulation. In a previous section discussing central and POL voltage regulation, we noted that power conversion near the load minimizes the distance over which high currents flow, thereby minimizing energy loss. This is an important aspect of voltage regulation as accomplished through switched DC-DC power converters. Load requirements of high currents at low-supply voltages are translated into much lower currents drawn from high-voltage supplies. This permits the connection of varied load devices to any given DC voltage source, as long as the power output capacity of the DC source is not exceeded. Switched DC-DC converters are also capable of “boosting” output voltage, providing high voltages at low currents, while drawing power from low voltages at high currents. Additionally, since switched converters employ active devices as switches, energy losses are typically very small as compared with linear regulators, where energy loss is directly related to the product of load current and the difference between input (higher) and output voltages. Hence, switched power converters and voltage regulators are very efficient, approaching 95% efficiency in commercial designs. Due to these advantages, switched power regulation is the most common method employed today, while linear regulation powers systems that demand high bandwidth and an absence of the output ripple inherent in switched converters.
Voltage regulation is the most common example of regulated power delivery, and it assists in maintaining the required voltage across a load. Practical loads will, however, impose significant challenges to any regulated power supply system. A resistive load applied to a DC-regulated power source, in the instant of application, will demand a near-instantaneous rise in current supplied by the power source. This “transient” or fleeting operational requirement may well exceed the capabilities of the power source, whose output voltage may not remain steady during such an event. Such considerations of the reliability and robustness of supplied power are generally categorized under power integrity.