Why does a 104 capacitor need to be placed on the chip power supply pin

    First of all, let‘s take a look at capacitors. The function of capacitors is simply to store charges. We all know that capacitor filtering should be added to the power supply, and a 0.1uF capacitor decoupling should be placed on the power pin of each chip. Why is the capacitor next to the power pin of the chip 0.1uF or 0.01uF? Is there any particular requirement. To understand this, one needs to understand the actual characteristics of capacitors. The ideal capacitor is just a memory of charge, that is, C. However, the actually manufactured capacitor is not so simple. When analyzing Power integrity, our commonly used capacitor model is shown in the following figure.

    In the figure, ESR is the series equivalent resistance of the capacitor, ESL is the series equivalent inductance of the capacitor, and C is the true ideal capacitor. ESR and ESL are determined by the manufacturing process and materials of capacitors and cannot be eliminated. What impact do these two things have on the circuit. ESR affects the ripple of the power supply, while ESL affects the filtering frequency characteristics of the capacitor.

    We know that the capacitance impedance Zc of a capacitor is 1/ ω C. Inductive reactance of inductance Zl= ω L（ ω= 2 π f), the actual complex impedance of the capacitor is

    Z=ESR+j ω L-1/j ω C=ESR+j2 π f L-1/j2 π f C. It can be seen that when the frequency is very low, the capacitance plays a role, while when the frequency is high to a certain level, the role of the inductance cannot be ignored. At higher frequencies, the inductance plays a dominant role. The capacitor loses its filtering function. So remember, capacitors are not simply capacitors at high frequencies. The filtering curve of the actual capacitance is shown in the following figure.

    As shown in the above figure, the best filtering effect we want is at the bottom of the "valley", which is the sharp point where the curve is concave. At this point, the filtering effect is good. When the logic gates inside our chip IC are executed within the range of 10-50Mhz, the interference generated inside the chip is also within the range of 10-50Mhz (such as a 51 microcontroller). Please carefully look at the curve in the above figure, The valley of 0.1uF capacitors (there are two types, one is plug-in and the other is patch) falls precisely within this range, so it can filter out interference in this frequency band. However, when the frequency is very high (50-100Mhz), that‘s not the case. At this point, the filtering effect of 0.1uF capacitors is not as good as 0.01uF, and so on. No matter how high the frequency is, the magnitude of the selected filtering capacitor needs to be reduced, as shown below

    Tantalum capacitors or aluminum electrolysis above DC-100K 10uF

    100K-10M 100nF (0.1uF) Ceramic capacitor

    10M-100M 10nF (0.01uF) Ceramic capacitor

    >100M 1nF (0.001uF) Ceramic capacitor

    So, in the future, don‘t see capacitors that put 0.1uF in everything. In some high-speed systems, these 0.1uF capacitors simply won‘t work.

    Finally, it should be noted that when laying out a PCB, 104 should be close to the chip and the power and ground circuits should be shortest, otherwise it will not have a bypass effect. The reference is as follows