SMT Power Inductors: Packing More Performance into Tiny Packages

Pity the poor inductor. Much like Cinderella, this neglected stepchild of the power supply is responsible for doing much of the heavy lifting, while its semiconductor siblings get all the attention. And it's getting worse. While silicon cheerfully rides the Moore curve, delivering quantum leaps in functionality in ever-smaller packages, magnetic elements must usually rely on less dramatic incremental improvements as they struggle to keep up with the increasing demands of both industrial and consumer applications.

This is no more apparent than with magnetic devices built for use with surface-mount technology (SMT) assembly systems where severe space, power density, and cost constraints are a fact of life. “Actually, surface-mounted ferrites are a bit of a misnomer,” says George Schaller, executive vice president of Fairfield, N.J.-based Ceramic Magnetics. “Since they are not an IC or monolithic device that has integral surface-mount pads,” he explains, “most inductors must be assembled on to a special carrier to make them compatible with surface-mount applications.”

However they're assembled, SMT-based inductors save space in assemblies and are easy to handle using automated equipment. This can help cut manufacturing cost and assembly time, but the very form factor and shape constraints that aid in the manufacturing process can make it difficult to meet some electrical design goals.

Fortunately, the past several years have seen many improvements in the materials, design, and manufacturing practices used in SMT inductors that have allowed power supply designers to keep pace with the demands placed on them by the rapidly evolving electronics industry.

Driving Factors

While several major factors drive the evolution of power inductors, perhaps the most important is size. It's important to remember that smaller is usually better when it comes to power supply design. Whether it's in consumer, telecom, or other OEM applications, power supplies must cram themselves into some pretty tight quarters.

Nowhere is the need for miniaturization more apparent than in the dc-dc converters used in cell phones, PDAs, and other handheld equipment — which must also shrink along with the ever-diminishing packages of the products they power. The cramped confines of a standard telecom rack, where the 0.5-in. to 0.75-in. spacing between cards allows an inductor or transformer a maximum of 6 mm to 10 mm of headroom, seem positively luxurious in comparison to a cell phone power supply. In these applications, designers are routinely demanding transformers and inductors as thin as 1 mm (Fig. 1) for their ultraminiature dc-dc converters.

Power density is another major issue in compact transformers — especially in telecom and datacom applications, where multi-gigabit transceivers, switch fabrics, and packet processing ICs conspire to push the power consumption of a blade toward 100W or more. At the low voltages that these ICs run (as low as 1.2V), the current demand can be huge, requiring a small supply to deliver 50A or more. This, in turn, requires the manufacturer to squeeze the most out of its tiny inductors by use of core geometries and winding schemes that allow for high flux densities that are uniformly distributed throughout their cross sections.

Compact size and high current density lead to another serious issue — heat. Even when properly designed, the core losses in a typical power transformer will turn several percent of the power they pass into abundant amounts of heat. Thanks to the wide operating temperature ranges of industrial equipment and the minimal cooling these same boxes usually provide, operating temperatures of 85°C to 100°C are common. In fact, at least one manufacturer (Micrometals, Anaheim, Calif.) has a regular demand for iron powder core materials rated for operation at up to 200°C.

Finally, SMT transformers must operate at considerably higher frequencies than ever before. David Kung, sr. applications engineer at BI Technologies (Fullerton, Calif.), says that today's power supplies often operate at frequencies measured in the megahertz. These higher frequencies are in good part used to accommodate the extremely rapid transients generated by the fast-switching logic found in most switching ICs, memories, and CPUs.

The lower inductive values found in these supplies (usually measured in nano-henrys vs. micro-henrys) permit faster response times, but of course, this comes at a price. The smaller number of turns used in these high-frequency inductors generate higher peak currents (70A to 80A in a 30A supply) that place heavy demands on core materials.

Mind the Gap

An air gap of one sort or another is often used to help the core support these high peak currents, without suffering from the sharp roll-off of inductive value due to core saturation. Forming these air gaps requires a high degree of precision to tradeoff current capacity and frequency response. This can be done several ways, depending on the materials being used and the level of precision required.

One of the most popular methods used in powdered iron cores is the “distributed air gap.” In this process, manufacturers produce the equivalent of an air gap by adding various materials (such as nylon) to the binder so that it allows the iron particles to be formed into a solid core, and separates the particles from each other. While this allows the initial gap in even very inexpensive magnetics to be precisely controlled, the value tends to change as the core ages — especially when operating at elevated temperatures.

When normal operating temperatures begin to approach 100°C, the binder can deteriorate over time (10,000 to 100,000 or more hours), allowing the particles to come into contact with each other. As this happens, eddy currents and other effects increase core losses, which produces more heat and further accelerates the binder's breakdown. Such “runaway” behavior may not be a factor in consumer gear, where product life is limited and temperatures are relatively benign. However, it becomes an issue in military, automotive, and industrial applications.

Some companies use new iron powders (such as Permalloy and Sendust) with nonorganic binder compounds that use materials like silicon or aluminum to produce a distributed air gap. These new materials can withstand operating temperatures of up to 200°C without degradation for hundreds of thousands of hours.

On the other hand, ferrite cores do not use a binder and are not susceptible to aging effects. They also enjoy several other advantages, including lower core losses at frequencies above 500 kHz, and possibly lower cost for some applications — at least compared to higher performance nonorganic powdered iron products. Unfortunately, distributed air gaps aren't an option if your application requires a ferrite core.

While there are several traditional methods to achieve a properly dimensioned mechanical air gap, the gapped toroid is a new concept growing in favor in small-medium (under 100W) supplies. Ceramic Magnetics Inc. (CMI) was an early developer of gapped toroid technology, and has an extensive line of products which have two micro-gaps spaced 180° apart. The total gap size can be precisely adjusted to meet the A_L (inductance factor) required by the designer.

By varying the thickness of an isolation compound that separates both toroid halves, a gap between .001 in. and .0024 in. can be produced (Fig. 2) without changing the core's manufacturing dimensions (ID, OD, and gap size). Since varying the gap requires no production setup, customers can order custom-value cores at no extra charge.

“But sometimes high-frequency designs require more than just a clever core,” says David Smolik, BI's director of new product development. He explained that at higher operating frequencies, ac resistance factors also becomes an issue, with skin effects lowering the effective conductance of a wire. Using smaller wire reduces the skin effect, but also its current capacity. Smolik says that if needed, BI can wind its inductors with rectangular wire that contains 27% more copper for an equivalent ac resistance as compared to round wire.

While more costly and difficult to work with, square wire can be the answer to some sticky problems; however, you should also consider using a helical winding instead. A helical coil reduces the proximity factor, something which also contributes to unwanted ac resistance. While this requires specialized winding machinery that only some manufacturers can provide, it can save money if specialized (i.e. square) wire is not required.

New Frontiers

For the most part, inductor technology is a mature technology, where innovation ambles along at a leisurely pace, with few “quantum leap” breakthroughs. However, a few developments promise to help power supply designers continue to push size, power, and efficiency barriers.

One of these new frontiers is planar coils, which use photolithographic manufacturing processes developed for making printed circuit boards (PCBs) to form inductors on single- and double-sided and multilayer materials. Depending on the application, PCB-based planar inductors use magnetic cores mounted on the top and bottom of the board, and often run through it, as shown in Fig. 3.

Paul Leibman, vice president of marketing at Coilcraft (Cary, Ill.), explained that planar inductors might have several advantages over the traditional ones. They have immense power density because the printed assemblies avoid the space consuming core-and-bobbin assemblies. However, manufacturers must strive to make planar inductors cost-competitive with the mature wound-coil products. This is especially true with prototypes, where it takes two to three weeks and up to $3000 to produce the photomasks and PCBs required to form the inductor. Leibman says they're working on a prototyping system to deliver low-cost, quick turn samples of custom planar products.

Perhaps the most intriguing technology on the market today is the “multiple-core” approach, which was developed by Flat Transformer Technology, Anaheim, Calif. Flat transformers distribute the work being performed across several smaller cores, each using a single turn on primary and secondary (Fig. 4). By running the secondary coil across multiple cores, you can build an effective turns ratio of your choosing.

James Lau, vice president of marketing at Flat Transformer explains that their multiple core architecture minimizes leakage inductance because it is directly related to the number of turns on the primary side. Low hysteresis and eddy current losses claimed for this approach make it an excellent candidate for low output voltage converters and applications where high efficiency is essential.

Lau also says that the transformer's design ensures that its output current is evenly distributed across the transformer's multiple taps — something that might allow a designer to save component costs by using several less expensive low-power semiconductors to do rectification instead of a single high-current device.

The theoretical cost for manufacturing is also much lower, because the flat-sided transformer elements can be mass-produced and stacked together for a given application (Fig. 5). As an example, a fully wound transformer element with a 10A rating costs around $3.50 each in single-unit quantities. It would take four elements to make a 40A, 4:-to- step-down transformer, for a total cost of around $14.

Think Ahead

In the end, whether you use conventional or exotic magnetic technologies, a good design requires careful attention to your requirements and to the properties of the components at hand. “And,” says Leibman, “making sure you don't put off the magnetics design until the end of the project.” A 20-plus year veteran of the magnetics industry, he has seen countless designs compromised because of last-minute requirements for expensive or nonexistent parts that could have easily been avoided. Thus, to optimize the solution from the start, the designer must work with the magnetics manufacturer early in the design cycle with a clean sheet of paper.

After serving nearly 20 years as an electrical engineer, Lee Goldberg has been writing for the electronics industry for nearly a decade. In addition to occasional freelance articles, Lee is co-editor of analogZONE.com, where he covers networking, I/O, and green engineering. He also publishes a monthly column on green engineering in EE Times.

References

Center for Power Electronics Systems, at Virginia Tech., www.cpes.vt.edu/cpespubs/volumes/volumes.html
“The Effect of Direct Current on the Inductance of a Ferrite Core,” Fair-Rite Corp., http://www.fair-rite.com/Directcurrent.pdf
Brian D. Wiese and George E. Schaller, “The Micro-Gapped Toroid — A New Magnetic Component,” Ceramic Magnetics Inc., http://www.cmi-ferrite.com/Papers/microgap.pdf

For more information on this article, contact the author at [email protected] or CIRCLE 331 on Reader Service Card

Getting the Lead Out

While it does not represent a breakthrough in cost or performance, no-lead solder technology is also an important development for the SMT inductor industry. Designers must also pay attention to environmental regulations that are now coming into effect in many of its target markets. Of most immediate concern are the serious restrictions on lead content in most electronic products that will soon be in effect in Europe, Japan, and other Asian nations. Within a year or two, these countries will require computers, appliances, consumer electronics, and even medical equipment to eliminate lead from their solder and other assemblies before they can be sold.

Many magnetics vendors already offer components that are pre-tinned with lead-free soldering materials. These are often Sn/Cu or Sn/Cu/Ag alloys designed to closely mimic the electrical, mechanical, and working characteristics of traditional Sn/Pb solders. The only real significant difference is their higher working temperatures — about 25°C to 50°C above regular solders. This can cause thermal sensitivity issues with some components, yet not typically with the high-temperature materials used in most inductors. Should you need it, vendors such as Cookson Electronics Assembly Materials, Jersey City, N.J., can provide a full range of lead-free soldering supplies.