Welcome to Power Tips! With the emphasis today on the need for more efficient and cost-effective power solutions, we created this column to provide helpful tips on a variety of power management topics. This monthly column is geared towards design engineers at all levels. Whether you've been in the power business a long time or just coming on the power scene, you'll find some nuggets of information that just might help you with your next design challenge. For more information about our plans for this column, click on my video here. Look for this column the first week of each month in Power Management DesignLine. Your comments are always welcome—RK
Selecting the optimum operating frequency for your power supply is a complex tradeoff involving size, efficiency, and cost. In general, low-frequency designs tend to be the most efficient, but are the largest and most costly. Moving higher in frequency improves size and cost at the expense of circuit losses. In the following paragraphs, we use a simple buck power supply to illustrate these tradeoffs.
Let's start with the filter components. They dominate a substantial portion of the power supply's volume, and a filter's size is inversely related to the operating frequency. On the other hand, each switching transition has an accompanying energy loss; the higher the operating frequency, the higher the switching losses, and the lower the efficiency. Then again, higher frequency operation usually means that component values can be smaller. Hence, operating at a higher frequency can lead to significant cost savings.
Figure 1 shows the frequency-volume relationship for the buck supply. At a frequency of 100 kHz, the inductor dominates the power supply's volume (dark blue area). If we assume that inductor volume is related to its energy, then the volume shrinks in direct proportion to frequency. This assumption may be a little optimistic because at some frequency, the core loss in the inductor becomes significant and limits further size reduction. If the design uses ceramic capacitors, output capacitor volume (brown area) decreases with frequency—required capacitance decreases. On the other hand, input capacitors generally are selected for their ripple current rating. The rating does not change appreciably with frequency, so their volume (yellow area) tends to remain constant. In addition, the semiconductor content of the power supply is constant over frequency. So with low-frequency switching, the volume of the power supply is dominated by the passive components. As we go to higher operating frequencies, the percentage space taken up by semiconductors (i.e., semiconductor volume, light blue area) starts to dominate.
Click to enlarge
Figure 1: Power supply's component volume will be semiconductor dominated
While the curves show that semiconductor volume is essentially constant with frequency, that relationship may be an oversimplification. There are two types of losses associated with semiconductors: conduction and switching. Conduction losses in a synchronous buck converter are inversely related to the MOSFETs' die area. The larger the MOSFET area, the less its resistance and conduction losses will be.
Switching-related losses have to do with how fast the MOSFET switches, and how much input and output capacitance the MOSFET has. These are both related to how large the device is. A larger device will have slower switching times and more capacitance. Figure 2 shows the relationships for two different operating frequencies (F). The conduction losses (Pcon) are independent of operating frequency, while switching losses (Psw F1 and Psw F2) are directly related. So a higher operating frequency (Psw F2) produces higher switching losses. The minimum total loss at each operating frequency occurs when switching and conduction losses are equal. And total losses will be greater as the operating frequency increases.
At higher operating frequencies, however, the optimum die area is less, which can lead to cost savings. In practice at the low frequencies, optimizing losses by adjusting die area produces an unaffordable design. However, as we push to higher operating frequencies, we can start to optimize die area for loss and, hence, reduce the volume of the semiconductors within a power supply. The downside is that if we don't have semiconductor technology improvements, power efficiencies will fall.
Click to enlarge
Figure 2: Increasing the operating frequency results in higher losses overall
As mentioned previously, higher operating frequency reduces the inductor's volume; the required core material is less. Higher frequencies also reduce the requirement on the output capacitor. With ceramics, it allows lower capacitance or fewer capacitors to be used. That tends to reduce the semiconductor die area, which also works to cut costs.
Next month, we'll talk about how to tame a noisy power supply.
 Robert Kollman is a Senior Applications Manager and Distinguished Member of Technical Staff at Texas Instruments. He has more than 30 years of experience in the power electronics business and has designed magnetics for power electronics ranging from sub-watt to sub-megawatt with operating frequencies into the megahertz range. Robert earned a BSEE from Texas A&M University, and a MSEE from Southern Methodist University.
|
