Tweet

Posted on 01 March 2019

Powering the Modern Electronic Revolution

Free Bodo's Power Magazines!

 

  

 

Digital control has the ability to raise power supply performance to the next level

If a single event were to define the start of the modern electronic revolution, it would have to be the 1958 invention of the Integrated Circuit (IC). The independent work of two inventors, Robert Noyce, co-founder of Fairchild Semiconductor and Jack Kilby of Texas Instruments, resulted in each being awarded succeeding patents.

By Steve Mappus, Systems Engineer, Fairchild Semiconductor

 

The initial roadblocks likely revolved around concerns regarding feasibility, manufacturing, test and which applications might most benefit from this technology. However, it wasn’t until 1961 when Fairchild Semiconductor introduced the first commercially available IC when a more interesting problem was noticed. While the IC had initiated the trend toward circuit miniaturization, other electronic systems, namely linear power sources, remained large and bulky by comparison. In many ways the invention of the IC, the start of the modern electronic revolution, drew attention to the fact that power technology was lagging. If commercial power sources were to keep pace with IC development, then drastic size reduction would be necessary. Ironically, it would be the invention of the IC itself that would, in time, enable the technology to overcome this obstacle.

Robert Noyce, Ph.D, 12-12-27 – 06-03-90

Linear power supplies operate from a rectified, low frequency (50 Hz - 60 Hz) off-line transformer and require a large amount of bulk capacitance proportional to the power demand of the load. The bridge rectifier and series pass element make them inherently inefficient, especially when low output voltages are required. Even prior to 1961, it was well understood that switching the energy storage elements at a higher frequency would be necessary to reduce power supply size. The problem was that the transistors required to switch off-line power at higher frequencies were not readily available. Driven by the smaller size and higher efficiency requirements of the aerospace industry, much of the power research done in the early 1960s was concentrated in the area of low voltage DC to DC conversion, which resulted in the development of the buck, boost and buck-boost, switch-mode topologies. The introduction of high voltage, bipolar power transistors in 1967 then made it possible to apply similar switching techniques, previously used in DC to DC conversion, to off-line power conversion.

Throughout the early 1970s some companies began offering commercially available switch-mode power supplies targeted at the military and industrial market segments. The half-bridge converter topology was the preferred choice because the voltage rating on the two primary transistors is limited to the peak input voltage. The DC output voltage was regulated using the pulse width modulation (PWM) control technique switching at 10 KHz - 20 KHz, offering major improvements in terms of size, weight and efficiency compared to linear power supplies.

PWM control is considered the essence of switch-mode power supply design and as IC development continued to progress through the early 1970s, many of the required analog and digital circuit blocks were beginning to appear in IC form. The idea of combining the individual circuit functions into a single IC eventually led to the industry’s first integrated PWM controller, the SG1524, designed at Silicon General in 1975. The invention of the SG1524 achieved much more than merely combining existing circuit blocks into a single chip. The functional requirements of the analog circuits demanded the IC be developed using a bipolar process but digital ICs were not typically produced this way. By monolithically integrating analog and digital circuits, the SG1524 challenged the notion that ICs had to be developed using either an analog or a digital process. The SG1524 also included the ability to protect the power stage via an internal current limiting function. The desire to achieve even higher levels of protection, monitoring, control and drive integration is still ongoing as more advanced power ICs continue to flood the industry.

Even though the SG1524 was able to switch as high as 300 KHz, bipolar power transistors are hindered by slower rise and fall times. Therefore, power supplies, for the most part, remained stalled around 25 KHz until the first available discrete power MOSFETs appeared in 1975. Compared to bipolar transistors, similarly rated MOSFET devices can switch at higher operating frequencies.

Techniques used to manufacture early MOSFETs were similar to planar IC development processes until Siliconix released the first trench power MOSFETs in 1994. Using a refined open cell trench process, Fairchild Semiconductor introduced their PowerTrenchTM family of MOSFETs in 1998, which resulted in even higher cell densities while maintaining low gate charge. As the number of competitors providing power MOSFETs grew, the process of how they were manufactured became as proprietary as the devices themselves and quickly became the second component necessary for achieving higher performance, smaller size power systems. As MOSFET characteristics rapidly improved, power supply designers often reaped significant efficiency benefits by sometimes doing little more than swap earlier generation devices with newer ones. Achieving similar improvements today place more emphasis on the skills of the designer and less dependency upon device improvements. Nonetheless, the topic of gate drive and MOSFET selection is still considered a critical step toward achieving the highest possible converter efficiency.

From the mid 1970s, much of the advancement experienced in the field of power electronics was driven by the demanding needs of emerging growth technologies. Electronic industries such as aerospace, industrial, telecom, medical, automotive, consumer and wireless each had unique requirements that continue to challenge semiconductor companies and power designers alike. It is impossible to mention all the significant power contributions that have resulted from such an exhaustive collaboration but in 1981 IBM released the Model 5150, which was adopted as the first personal computer to have the PC name associated with it. Today’s modern computing industry consists of PCs (more commonly known as desktops), workstations, notebooks, mainframes and servers. Computing would turn out to be a “sleeping giant” to the power supply industry, pushing the limits of how power is processed, delivered, managed, cooled and packaged.

One example worth mentioning is the distributed power architecture (DPA) commonly used today. The realization of the DPA arrived through a combination of requirements for higher power density, more flexibility, lower EMI levels, higher reliability and a higher level of system power performance. As shown in Figure 1, a DPA first converts the AC line voltage to a distributed DC voltage bus such as 48 V, although for high power, server applications this voltage can be as high as 350 V.

Distributed Power Architecture example

The integration of a PFC converter (AC to 400 V) and DC to DC converter (400 V to distributed DC) stage became commonly referred to as a “rectifier”, within the context of a DPA. The rectifier output is then used as the input to an intermediate bus converter (IBC) which then steps the voltage down to a quasi-regulated 12 V or lower. The 12V IBC voltage gets converted to the usable point of load (POL) DC voltages which usually range between less than 1 V to 5 V.

This advanced level of power processing, driven mostly by the microprocessor requirements, continues to spur the design of many novel PWM controllers and drivers, discrete and passive devices and technology patents. And, just as the previous 20 years, power supply design continued its progression within the shadows of IC (microprocessor) development.

Throughout the mid 1980s the need to squeeze more power out of smaller sizes continued to press on and in 1984 Vicor introduced the first “brick” module, a 4.6”x2.4”x0.5” DC to DC converter that achieved a power density of about 25 W/in3. This small converter size was made possible by switching at extremely high frequency, made possible through zero-current switching techniques. Using families of high density bricks, Vicor had proposed a configurable, building block approach to power supply design. As power designers gradually began to accept the brick concept, the converter module industry quickly became crowded with numerous competitors each proposing their unique spin.

The demand for smaller packaging extended beyond the design of brick modules to also include discrete power devices. In 1985 the surface mount DPAK was introduced which started the campaign toward power, surface mount technology. Power packages such as the DPAK and its successor, the D2PAK offered additional thermal benefits over more commonly used through-hole packages. Modern surface mount power packages such as the quad flat no-lead (QFN) shown in Figure 2, can include multiple power devices co-packed into a single module.

QFN packages for multi-die modules and high power applications

Fairchild Semiconductor is one of the few suppliers of power management ICs able to co-package discrete devices with gate drive, protection and control IC functions. The success of multi-die modules are just one example of how the power industry is embracing technologies that optimize system power through higher levels of component integration.

The 1990s ushered in on the explosive growth of the internet. By virtue of the sheer amount of hardware infrastructure required to support the World Wide Web (WWW), power electronics is to the internet what gasoline is to automobiles. There are currently more than one billion people worldwide who regularly use the internet. Every access point requires a PC (or wireless handheld device) to surf through the power hungry network of modems, routers and servers that make up millions of host computer sites all over the world. Furthermore, with global power consumption growing at a yearly rate of nearly 2%, the issue of adequately maintaining aging power-grids has become a huge concern. Initially, reducing demand at the load makes more economic sense. However, the problem would still need to be addressed from the power grid. With government legislation and various special interest groups stepping in, power efficiency is now in the global spotlight. The One Watt Initiative is one example of a proposal by President George W Bush in 2001 which aimed to reduce standby power losses to below 1 W.

Historically, designers have been most concerned with power supply efficiency during a systems full load or active state where operating temperatures tend to be highest. The One Watt Initiative and other similar energy saving programs have challenged designers to rethink the efficiency issue. The result is a new generation of power management ICs, such as the Fairchild Power Switch (FPSTM) FSQ-series, highlighted in Figure 3. The FSQ-series FPS contains an integrated PWM controller and current sensing MOSFET which consumes less than 0.2 W when operating in standby mode. Power controllers that enable power supplies to achieve higher efficiency during standby mode are commonly referred to as “green mode” due to the environmental benefits they offer.

FSQ-series green mode FPS

Even as more “green mode” features are expected to enter the industry, power systems have mostly remained analog in the sense that they are optimized using fixed passive components chosen to meet a specific set of known operating specifications. As load requirements become increasingly complex, the limitations of a pure analog approach become apparent. Digital control has the ability to raise power supply performance to the next level in many respects. Imagine being able to manage multiple power rails, change regulation set points based upon load dynamics, achieve high efficiency across a wider load range, dynamically adjust the way faults are handled and achieve real time control loop optimization under all conditions. Digital control can yield better overall efficiency results because the system power is managed more efficiently according to the demands of the load.

As the power supply industry is sitting on the cusp of what appears to be the next defining milestone, will digital power be able to meet the demands of the modern technology revolution? As with any new technology, optimism is often met with skepticism but with several proposed solutions already revealed and more on the way, the early adopters will be the first to answer.

 

 

VN:F [1.9.17_1161]
Rating: 0.0/6 (0 votes cast)

This post was written by:

- who has written 791 posts on PowerGuru - Power Electronics Information Portal.


Contact the author

Leave a Response

You must be logged in to post a comment.