Designing an efficient power inductor for mobile-computing applications
A new approach to power-inductor design enables VRMs (voltage regulator modules) to handle more current, operate more efficiently, and give higher performance, all while reducing costs.
By Gil L. Opina, Jr, Pulse Electronics -- EDN, August 14, 2007
Constrained by battery capacity, notebook manufacturers seek more efficient as well as lower-priced components for their designs. Active and passive devices used on VRMs (voltage regulator modules) that power microprocessors have been evolving to increase overall efficiency while lowering the cost of building the part. For power inductors, however, where cost is always proportional to the material used and the complexity of the design, it has been difficult to both increase overall efficiency and to reduce cost. This article explores a new way to design power inductors that enable VRMs to handle more current, operate more efficiently, and give higher performance, at the same time reducing costs.
Increased efficiency in the power-management circuit of notebook VRMs increases battery operating time. Efficiency is the ratio between the output power and the input power:
For a given output power (Pout), the efficiency (ξ) can be increased by minimizing the power (Pin-Pout) drawn by the system. This can be achieved by minimizing the loss on the active and passive devices used. For inductors, the total loss is given by:
Copper loss (Pcopper) is a function of the loss due to the direct current resistance (DCR) and alternating current resistance (ACR) of the conductor.
Core loss (Pcore) is a function of the magnetic flux density (ΔB) created in the inductor by the ripple current and the frequency of operation. The amount of core loss as the result of a particular application is mainly affected by the core configuration and size as well as the core material's hysteresis and eddy current loss properties. This can be mathematically expressed based on the Steinmetz equation:
Where: K, m and n are core material constants
ΔBpk is the peak ac flux density in Gauss
f is the switching frequency in Hz
Core material selection
There are a number of soft ferrite materials in the market right now. Each material offers a variety of advantages and disadvantages to the inductor's design. Currently, two materials are widely used for power inductor designs that are used in notebook VRMs. One is manganese zinc (MnZn) and the other is iron powder.
Iron powder can be further classified into two material grades. The first grade is the high temperature grade which can handle temperatures up to about 200°C without exhibiting thermal aging. They are made of iron alloy which allows operation at elevated temperatures. The second grade is made of normal grade iron powder (an example is Micrometal's -52, -26, -18, etc.) material which, depending on the core structure, flux density, and switching frequency, can only withstand temperatures below 125°C.
Thermal aging is a condition where normal iron powder material exhibits an irreversible increase in core loss primarily due to exposure to an elevated temperature (usually >125°C) for a prolonged period of time. This increase in core loss subsequently increases the inductor temperature, which tends to accelerate the aging process over time. Eventually, it comes to a point where the inductor temperature is so high that it will permanently damage the inductor as well as the system.
Each grade of iron powder material has its own advantages and disadvantages when compared to MnZn. For the same condition (same core size, turns, flux density, and frequency), the MnZn material has a better core loss performance, especially at higher frequencies, compared to iron powder materials as seen in Figure 1. Normal VRMs operate at 300 kHz and around 400-500 Gauss. At this condition, the iron powder core loss density is around 10 times more than that of the MnZn material. Depending on the core size and structure, this can translate to a couple of Watts of loss and eventually reduce overall system efficiency as much as 5%-10%.
Since the core loss of iron powder materials is extremely high compared to MnZn, and the objective is to increase efficiency, it is a logical conclusion to use the MnZn core material for notebook VRMs, which will result in a 5%-10% improvement in the efficiency of the VRM. If a 40W VRM is used, about 2 to 4 Watts of power are preserved by using the MnZn.
On the other hand, iron powder cores, due to an inherent distributed gap within the core structure, have a better bias performance compared to MnZn. Typically, iron powder cores have a saturation flux density (Bsat) of greater than 10 kilo-Gauss (kG), while a MnZn power material typically ranges from 4.8 kG to 5.1 kG at room temperature. Another advantage of the distributed gap within the iron powder material is a soft inductance roll-off, especially nearing the knee area (see Figure 2). MnZn material exhibits a drastic drop near the knee area and if a sudden increase in load occurs, the core may immediately go to saturation causing the output current to spike. This makes it impossible to regulate the output of the VRM, perhaps causing permanent damage to the VRM or processor.
Choosing ferrite material due to its lower core losses raises concern on the hard saturation of the inductor, but there are ways of increasing the peak current that an inductor can handle. This can be done by shifting the knee area of the saturation curve further right, so as not to put the inductor at risk of early saturation.
The ferrite ERI10×10×4-mm used in populating the inductance curve in Figure 2 was redesigned, resulting in a better saturation as shown in Figure 3.
The knee area of the inductance curve is improved by 8.5A or by around 30%. The new saturation of the improved ferrite core will now become 33.5A from the previous 25A. This will give more headroom especially during system full load and transient conditions. This will also pacify concern, especially from notebook designers, regarding the early hard saturation of ferrite inductors currently available in the market.
Coil design
Currently, magnetic wires have shapes that are either round, square, or rectangular. Round wires are the most commonly used in the industry but square, and particularly rectangular wires, are becoming more popular, especially for high-density, low-profile applications as in notebooks. Copper loss for an inductor is given by the equation:
And Rdc of any conductive material is given by:
Where: ρ is the resistivity of the conductor, for copper
it is 6.878×10-7 Ω-inch or 1.68×10-5 Ω-mm at 20°C
L is the length of the conductor
A is the cross sectional area of the conductor normal to the flow of current
|
For a given winding space and number of turns, flat wire will have a better copper fill factor. Thus it enables designers to put a larger copper area inside the core window than round wire. From equations 11 and 12 it can be seen that an increase in the copper cross sectional area used reduces the Rdc and eventually the copper loss. Typically, flat wire will have around 22% less Rdc compared to round wire.
Experimental results
The Dell XPS M1210 notebook was subjected to a notebook battery life benchmark using BAPCO's MobileMark 2005 software. This software initiates three test environments that include the DVD playback, Office Productivity 2002SE, and Reader 2002SE. The purpose of this test was to compare the battery life of the Dell notebook when the three existing microprocessor voltage regulator inductors are replaced with higher efficiency inductors. The original inductor uses an iron powder core material while the replacement inductor uses a low core loss ferrite material.
During the duration of the BAPCO MobileMark 2005 testing, one of the inductors used on the voltage regulator had its temperature monitored. This was done by placing a thermocouple on top of the inductor, capturing the readings by a data logger, and saving the results to a separate desktop computer. The Dell notebook was subjected to the three tests, first with the original inductors (see Figure 4A) and then with the power ferrite core inductors (see Figure 4B).
After each test environment was done, the BAPCO software generated a report. Table 1 summarizes the test results.
The temperatures gathered for each test and for each inductor used are presented in Figure 5, Figure 6, and Figure 7.
From these experimental tests, it appears that using the inductor made from the ferrite material will result in a lower temperature as well as longer battery operating time (up to 6% improvement) compared with existing inductor solutions utilizing iron powder core materials.
Conclusion
Notebook computers are now being required to incorporate a more efficient power management circuit in their designs, mainly because of the growing concern for increased energy consumption. The drive for increased energy consumption is not only being implemented by large scale energy users in industry, but by consumer electronics as well. Notebook designers, who are used to incorporating iron powder inductors in their VRD/VRM applications because ferrite materials have drastic saturation, must reconsider their use of these materials because of the high core loss properties of iron powder inductors. The reason that ferrite has not been used in such applications is primarily due to its hard saturation properties.
This article has proposed that with a carefully designed core structure, a ferrite material can handle the high saturation requirements of notebook computers. There are inductors on the market today that can bridge the low core loss properties of a ferrite core with the high saturation current capabilities. This has been accomplished by carefully designing the core structure to enhance its saturation current. A further advantage is that these inductor designs incorporate a rectangular wire winding process which decreases the direct current resistance (DCR) by almost 30% compared to its iron powder inductor counterpart which uses helical coil winding technology. With a lower DCR, and a ferrite material, these inductors increase overall system efficiency and show remarkable results during the BAPCO MobileMark 2005 battery operating time tests. The inductors increase notebook battery operating time by an average of 4% and lower the inductor temperature by at least 3°C compared to the iron powder inductors.
Author information
Gil Opina is a senior design engineer with Pulse Electronics (Singapore), where his responsibilities include designing custom and standard power magnetics. He is currently pursuing a master's degree in Power Engineering at Nanyang Technological University in Singapore. His interests include playing basketball, tennis, golf, and billiards, and learning about advances in space exploration.
