Knowing what you know today about the enormous penetration of power supplies and their component parts at every level of electronic products and systems, would you believe that this huge industry rode in on the back of a single integrated circuit some 30 years ago? The year was 1976 when the then Silicon General company introduced the SG1524 regulating pulse-width modulator (PWM) integrated circuit. Invented by Bob Mammano, now a staff technologist with Texas Instruments, it was the first device to incorporate all of the circuitry needed to generate the adjustable frequency, pulse-width-modulated, 180-degree-out-of-phase control pulses that drive the power transistors of switching regulator power supplies.
Prior to that time, switching power supplies were a curiosity except for their use in military systems. But the advantages of switchers were beginning to get a foothold. In a well-known textbook published in 1977, Switching and Linear Power Supply, Power Converter Design, the late Abraham Pressman, a highly regarded designer and consultant on power supply design, wrote, “Switching regulators are in the process of revolutionizing the power supply industry because of their low internal losses, small size and weight, and costs competitive with conventional series-pass or linear power supplies.”
Switching power supplies languished in commercial applications up until around 1976 because of their then complexity and high costs compared with linear supplies. Even Mammano, who had experience in the design of military switchers, says, “Switching power supplies looked threatening. There was a lot of technology that we were not familiar with, so we wanted to make a control chip that would bring a lot of that complexity onto a single chip.”
Len Sherman, senior scientist at Maxim Integrated Circuits, recalls, “When it was a discrete circuit, very few people could do a switching power supply. When the 1524 came out, it wasn't easy to design a switcher, but a lot more people could attempt to do one.”
Looking back at the SG1524, it contained the basic functions for the control portion of a regulating power supply (Fig. 1). It had a 5-V regulator/reference, an error amplifier, an oscillator, a PWM comparator, a pulse-steering flip-flop, two uncommitted switching transistors, current limiting and shutdown circuitry. The uncommitted outputs were suitable for either single-ended or push-pull applications. It was rated for operation over the full military temperature range of -55°C to 125°C. Two other versions, the SG3524 and SG2524, had the same circuitry but operated over the 0°C to 70°C temperature range. All this circuitry was in a 16-pin dual in-line package (DIP). An interesting note is that a description of the SG1524 and its operation are the subject of the last three pages in Pressman's textbook.
Aside from being the first single-chip PWM IC introduced, what other significance can be attached to the device? According to Mammano, “Combining the analog functions, the references, error amplifier and op amps with digital circuits for doing PWM hadn't been done before on one chip. I always felt that was the most significant accomplishment.” And in today's technology, where mixed-mode (analog/digital) devices are commonplace, the SG1524 was one of the first mixed-mode ICs.
Other semiconductor companies were not standing still when the SG1524 made its debut. In short order, Motorola Semiconductor released its MC3420 based on much the same idea as the SG1524, and Texas Instruments came out with the TL494 (Fig. 2). The latter offered an improved oscillator design, which allowed easy synchronization of the control circuit to additional controllers or a system clock. In addition to these PWM ICs, manufacturers such as Signetics, with the NE5560, and Ferranti, with the ZN 1066, introduced their own versions on the market.
Once the door was opened to single-chip PWM controllers, new features and upgraded specifications flowed quickly as switching supplies made inroads on the power supply industry's mainstay linear regulators. For example, Silicon General brought out the SG1525, which increased the output drive capability from the original 100 mA to 200 mA, and provided a ±1% accuracy 5-V reference compared to the original's ±4%. Motorola beefed up its line with the MC3421, increasing the output drive capability to 250 mA compared to the original's 50 mA. And TI's TL497A was touted as being able to operate in three dc-dc voltage conversion modes — step up, step down and inverting — with up to 85% of the source power delivered to the load.
As PWM ICs added more functionality, a second class of power devices began to emerge in the form of support and supervisory chips. One of the most comprehensive was Silicon General's SG1543 output supervisory circuit, designed to replace several external circuits: overvoltage sensing with provisions to trigger an external crowbar shutdown, undervoltage sensing to either monitor the output or sample the input line voltage, and current sensing or current limiting.
A simpler IC, Motorola's MC3423 was essentially a crowbar-triggering device programmable through an external capacitor to provide delays that prevent the false triggering of an overvoltage protection circuit. Texas Instruments came out with the TL430 and TL431 adjustable shunt regulators designed to produce sharper turn-on characteristics than temperature-compensated zener diodes. These devices were analog ICs, and it should be noted that analog types were available years before the first PWM device arrived. Entire families of three-terminal series regulators, the µA7800 (for positive voltages) and the µA7900 (for negative voltages), were widely available from a number of semiconductor manufacturers.
Until the early 1980s, PWM ICs were based on voltage-mode control. At that time, researchers and designers began focusing on current-mode control because it offered certain advantages over voltage-mode control. In current mode, for example, the power supply looks like a current source to the input, so voltage changes at the input don't get through to the output. It also is easier to parallel current sources into an output capacitor than voltage sources. And since it can be implemented using cycle-by-cycle current limiting protection, current-mode control is more immune to current damage from short circuits at the load. (There was a crude form of cycle-by-cycle current limiting on the SG1524.) The Unitrode Corp., now part of Texas Instruments, introduced its 1842 and 1846 control ICs for single-ended and push-pull supplies using this principle.
In recent years, current mode has become more prevalent because of the high current loads that have to be switched in systems that contain many low-voltage, high-current microprocessors, DSPs and other digital circuits. Don Ashley, strategic marketing manager at National Semiconductor, explains, “Current mode is a bit of a complex control system, but it has the most robust transient response. It will quickly correct a big load change on the output. With new processors that switch high currents in microseconds, voltage mode can't cut it.”
Although much of the credit for the widespread adoption of switching regulator power supplies accrues to PWM ICs, other components of power supplies play key roles. As Mike Briere, executive vice president of research and development at International Rectifier, points out, “It's not just the advent of the PWM IC but the coming of the power MOSFET that enabled the simpler drives and the faster switching that could take advantage of the PWM concept. None of the power supply components stands alone; magnetics, ICs and switches all play an important part.” Before power MOSFETs came along in the mid-1970s, bipolar transistors performed the function of power switches. But since MOSFETs can switch faster, they enabled the switching frequencies of power supplies to move from the 25-kHz to 50-kHz range to hundreds of kilohertz and even megahertz, thereby reducing component size and leading to smaller, faster, more efficient supplies.
Power ICs Grow Up
Today's power control IC industry has migrated from that first PWM IC in 1976 to what Venture Development Corp. called in a 2003 report a “Global Market for Power Supply and Power Management ICs.” The report estimated that the size of the market was more than $5 billion and, growing at an annual rate of 8.8%, was slated to reach close to $7 billion by 2006.
With that growth has come a huge number of specialized device types to support the computer, communications, automotive, consumer appliance and industrial applications that barely existed 30 years ago. There are buck regulators, synchronous buck regulators, charge pumps, MOSFET drivers (both high side and low side), voltage- and current-mode controllers, low dropout regulators (LDOs), hot-swap and soft-switching controllers to name a few. Portable products such as cell phones, cameras, PDAs and other battery-powered gear have created an entire array of power-management devices that did not exist 30 years ago just to serve those applications. Included here are charging circuits, protection ICs, battery-management chips and gas gages (to determine the amount of power remaining in a battery).
Some of the PWM ICs available now look remarkably similar to the first devices from the mid-1970s. For example, Fairchild Semiconductor offers the KA7500C switching-mode power supply (SMPS) controller IC, a device in a 16-pin DIP whose description and features closely resemble the SG3524 and other early PWM ICs. Of course, modern semiconductor processing results in one major difference. The KA7500C sells for 37 cents in 1000-piece quantities, whereas an SG3424 went for around $13 in 100-piece quantities (and this price is not adjusted for inflation).
Another device that has elements of the early PWM chips is the NCP5425 dual synchronous buck controller from ON Semiconductor (Fig. 3). This device is much more integrated because it packs two independent buck regulators that can be used for single two-phase or dual single-phase outputs. In a dual output mode, the second output tracks voltage transients from the first. It provides four gate-drive outputs, two each from its dual controllers. Like earlier PWMs, it can perform cycle-by-cycle current limiting, can be programmed for operating frequencies from 150 kHz to 750 kHz and has a programmable soft-start feature.
With its LM5041 cascaded PWM con-troller, National Semi-conductor has defined what it claims are the elements of a modern PWM power control IC (Fig. 4). Throughout the years, power distribution architectures have under-gone significant changes to deal with processors, DSPs, ASICs and other digital ICs that operate at low voltages — under 2 V — while drawing high currents — up to hundreds of amps. One of these distribution systems, the intermediate bus architecture (IBA) uses point-of load (POL) converters, usually in the form of buck converters that deliver the large number of different voltages and currents required by the system components.
The LM5041 is equipped with many of the features necessary to operate in such an architecture. Cascaded topologies are suitable for multiple-output and high-power applications, and with multiple controllers in a single package, it saves space on system boards. It has four internal gate drivers, two of which are intended for driving a push-pull load and well-known features such as undervoltage lockout (UVLO), thermal shutdown and user-programmable dead time. The device also has a high-voltage startup regulator and the ability to sense fast transients with leading-edge blanking.
A major trend in today's power IC development is to create so-called “smart power” devices. This means integrating intelligence, analog and power on the same power control IC. The idea is to combine semiconductor technologies to perform analog and digital functions on the same chip. In today's technology, the usual division of processes and functions is to use BiCMOS for oscillators, drivers, amplifiers and voltage references, DMOS for the power switches and CMOS for the control logic.
One of the more common uses of this technology is to incorporate a PWM IC's power switches on the device rather than having on-chip gate drivers that control external power MOSFETs.
One device that uses this technology is the voltage-mode MAX8566 from Maxim Integrated Products. It is a 10-A PWM stepdown regulator with internal switches. The device can run in PWM mode at frequencies from 250 kHz to 2.4 MHz, and being at the higher end of the industry's present frequency range, the size of external components is significantly reduced. The on-chip switches are n-channel power MOSFETs having very low RDS(ON) values — 8 mΩ — to reduce switching losses. Having the switches integrated with the controller saves board space, an important consideration in today's high-density POL applications since a system may require a large number of such power supplies to support a wide range of voltages operating at heavy current loads.
The power IC industry has come a long way from the days of the first PWM ICs, but it will always be in transition because it is driven by advances in digital IC technology. As gate widths drift ever lower, more complex devices will be available, operating at voltage levels below 1 V with currents running into the hundreds of amperes. Then the technology must once again rise to the challenge of creating new devices and architectures for powering the systems of the future.