Have you had any experience with a permanent (scald) imprint of charging a buck regulator for a full power test, followed by a permanent inductive terminal temperature test? Excessive core loss and AC winding loss are The culprit. There is generally no problem with the 100-kHz switching frequency because the core loss is about 5% to 10% of the total inductor loss. Therefore, the corresponding temperature rise is the problem. In general, when choosing an inductor, simply calculate the maximum load current and establish the inductance by allowing 20% ripple current. Due to the insignificant core losses, a temperature rise similar to that shown in the product manual appears. However, as the switching frequency rises above 500 kHz, core losses and winding AC losses can greatly reduce the allowable DC current in the inductor. Using a 20% ripple current to calculate the inductance results in the same flux increase in core material, regardless of frequency. The general form of the core loss equation is:

Pcore = K × F1.3. Therefore, if the frequency (F) rises from 100 kHz to 500 kHz, the core loss will be eight times the original. Figure 1 shows this rise and also describes the allowable copper loss as the core loss increases. At 100 KHz, most of the losses are present in the copper wire while it is possible to utilize full DC rated current. At higher frequencies, the core loss becomes larger. Since the total allowable loss is determined by the sum of the core loss and the copper loss, the copper loss must be reduced as the core loss increases. This condition continues until the losses are equal. Best of all, losses remain stable at high frequencies and allow for maximum output current from the magnetic structure.

Above 0.5 MHz, the core loss greatly reduces the effective conduction loss. Figure 1 and Figure 2 are based on fixed core volume and winding area, only the number of turns is variable. Figure 2 shows the core loss inductance and allowable dc current shown in Figure 1. At frequencies below 1.3 MHz, the inductance is inversely proportional to the switching frequency. The inductance reaches a minimum around 1.3 MHz. Above this frequency, the inductor must be raised to limit the core flux, thus controlling the core loss to 50% of the total loss. The inductor's current rating is also calculated at the same time. Low frequency, the core loss is not large, the rated current from the winding power loss decision. In the following equation, the number of turns is proportional to the reciprocal of the square root of the frequency, so doubling the frequency (reducing the inductance by half) yields 0.707 turns. L = μ × A × N2 / lm This situation affects the winding resistance in two ways. The number of turns is reduced by 30% while the available area per turn is increased by 41%. Since the winding resistance is related to turns / turn area, the resistance decreases linearly with frequency, for example, the resistance drops by a factor of 2 in this example. At higher frequencies, core loss begins to limit the copper losses until they reach the same point. At this point, the inductance is increased to reduce the flux by adding more turns and increasing the winding resistance. In this way, the inductor current rating is reduced. Therefore, the best frequency is obtained from the viewpoint of inductance size.

In summary, the idea of increasing the switching frequency to reduce the core size is correct, but only at the point where the core loss and the AC winding loss are equal to the copper loss. After this point, the core size actually increases. In addition, designers should note that there are many high switching frequency products available today, some of the corresponding application manual does not clearly indicate that there are some potential problems of excessive core loss.