What capacitor and inductor are the best for a switching power supply? - Capacitor -ESR of the Output Capacitor Exerts a Significant Impact on Output Fluctuation at Decreased Load

－ In addition to the output ripples, are there any other items that require particular attention for the output capacitor?

The output capacitor has a function, in addition to the output ripple, of maintaining regulation with respect to the fluctuations of the output load current. As you may know, when the CPU transitions from a sleep state to the run state, a large load current flows suddenly, and this condition causes the phenomenon of an instantaneous drop of output voltage.

－ Yes, I understand that the load transient response characteristic is one of the important characteristics of a power supply unit.

Changes in output with respect to load fluctuations occur on a sudden increase in load, as in the example just mentioned, and conversely on a sudden drop in load. In the example of the CPU, the condition arises when the CPU shifts from the run state to a sleep state. In this case, the output voltage rises instantaneously. What I’m going to discuss is the case where “although the level of output voltage drop during a sudden increase in load is within the margin of tolerance, the increase in output voltage on a sudden load decrease is unexpectedly large.”

I’d like to discuss this topic in reference to waveforms and state diagrams. The top waveform diagram shows the output voltage (red) for a synchronous step-down converter, the inductor current (blue), and load current (pink).

First of all, the load current begins to decline a little before the vertical broken line ① (blue), ultimately reaching zero. Imagine a situation where the load current, 3A in magnitude, virtually ceases to flow due to the CPU, for example, or something else shutting down.

Regarding the next item, the inductor current, because the load current begins to diminish during the high-side switching off (low-side on) period immediately before the vertical broken line ① (blue), the off period seems to have extended to some extent. However, the next cycle (high-side on, low-side off) unwittingly begins at the time of the vertical ① (blue) broken line. Due to this fact, the inductor current increases even though the load current is extraneous. After that, at the time of the vertical ② (green) broken line, the high-side switch turns off. In this condition, the off-state continues, and the inductor current continues to decline to near zero. Bear in mind that up to the mid-point, the condition continues in which the inductor current is larger than the load current.

And in view of the change in the load and inductor (switching) currents, let us look the output voltage. When the load current begins to fall, the output does not immediately decline to an extent that is necessary. Even though we would like to keep the high-side switch off, due to controls exerted on the power supply IC, the high-side switch unwittingly turns on, with the result that the output voltage rises suddenly (during the period marked by ① blue and ② green vertical broken lines). At this time, although the on-time becomes short to some extent, because the load current continues to diminish, at this time the power supply ends up passing a great deal of current to the output capacitor.

After that, the high-side switch turns off and the inductor current declines. However, given the fact that the inductor current is greater than the load current, the difference flows into the output capacitor, causing the output voltage to continue to rise. The waveform chart at the bottom shows the capacitor current.

The output voltage starts a declining pattern beginning with around the red vertical broken line ③. The reason is that the difference between the inductor and load currents, that is, the capacitor current, begins to decline as time passes. In the lower waveform chart, compare the difference between the inductor and load currents and the change in capacitor current. The reversal of the load current waveform represents the capacitor current waveform. It is apparent that at the intersection with the inductor current, the capacitor current is zero, and in the subsequent reversed segment it is negative, after which it returns to zero.

The change in output voltage is V_c＋V_esr, and in either case the capacitor current is involved. In particular, because the term V_esr arises in ESR × capacitor current, it is inevitable that an increase in ESR produces a large output change.

－ You have not mentioned ESL. Does it mean that it’s not relevant?

In the case of the conditions governing this example, I believe that there is no particular need to consider ESL. However, in situations where the decrease in load current is even more abrupt, the impact of ESL will present itself.

－ In this example, you used a functional polymer-type output capacitor. What differences do various types of capacitors make, especially when an MLCC is used?

Here we have data indicating the results of using various output capacitors. We have conducted experiments using three representative functional polymer types, and altogether 16 types of difference capacitances and sizes, including multi-layer ceramic (MLCC). The top waveform diagram indicates significant output voltage changes as a result of a sudden decrease in load. The waveform diagram below it provides an enlarged view of the change, which varies significantly depending upon the type and capacitance of the capacitor involved.

Be that as it may, because frankly the changes are not easy to see, we created a mapping view.

As noted above, because in this example the principal driving forces for output voltage change are V_c and V_esr, irrespective of the type of capacitor involved, the capacitor with a large capacitance, that is, consequently a small ESR, is the key to reducing the output voltage fluctuations when the load declines suddenly in this manner.

－ What advantages does the MLCC provide?

As in the case of ripples, because the multi-layer ceramic capacitor (MLCC) has low ESR and ESL values, from the standpoint of these parasitic components, for a given performance level an MLCC with a smaller capacitance than the functional polymer type can be adequate, and obviously it offers advantages in terms of size reduction. For a given job, the MLCC with a two-third capacitance of the functional polymer type will be acceptable.

－ Well said. Are there any other points we should be aware of?

In the example we just discussed, the prime factor responsible for a significant rise in voltage when the load drops suddenly lies in the control of the power supply IC employed, that is, the response characteristics that are exhibited in the event of a sudden drop in load. Conversely, the issue described here may be reduced to an acceptable level by using a type of power supply IC that can quickly respond to a decrease in load.

Another point I should mention is that if similar problems occur, they can be addressed by using an output capacitor with a small ESR. Additionally, including the case of ripples, although output capacitors with small parasitic components, such as ESR and ESL, offer advantages, in some power ICs a small ESR in the output capacitor can cause problems. This issue deserves a careful study.