Monolithic capacitors, paper capacitors, electrolytic capacitors, low-frequency ceramic capacitors (also known as ferroelectric capacitors), and polyester capacitors (generally with large capacitance and small size) are not suitable for high- and medium-frequency circuits due to their high dielectric loss, but can be used in low-frequency circuits and power supply filtering circuits. Mica capacitors, polystyrene capacitors, high-frequency ceramic capacitors, and air dielectric capacitors (generally with small capacitance and relatively large size) have low dielectric loss and are suitable for use in high-frequency and medium-frequency circuits.
I've always had a question: the inductive reactance of a capacitor is 1/jwC. A large capacitor has a large C, and at high frequencies, w is also large, so the impedance should be very small. Shouldn't it be more suitable for filtering high-frequency signals? However, the fact is that large capacitors filter out low-frequency signals.
The answer I found today is as follows: Generally, capacitors around 10pF are used to filter out high-frequency interference signals, and capacitors around 0.1uF are used to filter out low-frequency ripple interference, and can also play a role in voltage stabilization.
The specific capacitance value of the filter capacitor depends on the main operating frequency of your PCB and the harmonic frequencies that may affect the system. You can check the capacitor datasheets of relevant manufacturers or refer to the database software provided by the manufacturers to select according to your specific needs. As for the number, it's not fixed; it depends on your specific needs. Adding one or two more is fine. You can leave unused capacitors out for now and choose the capacitance value based on actual debugging results. If the main operating frequency of your PCB is relatively low, adding two capacitors is sufficient: one to filter ripple and one to filter high-frequency signals. If there is a large instantaneous current, it is recommended to add a larger tantalum capacitor.
In fact, filtering also includes two aspects, namely, large capacitance and small capacitance, which are decoupling and bypass, as some of you have mentioned. I won't go into the details of the principle, but for practical purposes, 0.1uF is generally sufficient for decoupling in digital circuits below 10MHz; for frequencies above 20MHz, 1 to 10uF is used to better remove high-frequency noise, roughly following the formula C=1/f. Bypassing is generally much smaller, usually 0.1 or 0.01uF depending on the resonant frequency.
When it comes to capacitors, the various names can be overwhelming: bypass capacitor, decoupling capacitor, filter capacitor, etc. Regardless of the name, their principle is the same: utilizing the characteristic of presenting low impedance to AC signals. This can be seen from the formula for the equivalent impedance of a capacitor.
Xcap = 1/2πfC. The higher the operating frequency, the larger the capacitance value, and the lower the capacitor's impedance. In a circuit, if the primary function of a capacitor is to provide a low-impedance path for AC signals, it is called a bypass capacitor; if its main purpose is to increase AC coupling between the power supply and ground, reducing the impact of AC signals on the power supply, it can be called a decoupling capacitor; if used in a filtering circuit, it can be called a filter capacitor; in addition, for DC voltage, capacitors can also act as energy storage devices, functioning like batteries through charging and discharging. In reality, the functions of capacitors are often multifaceted, and we don't need to spend too much time considering how to define them. In this article, we will uniformly refer to these capacitors used in high-speed PCB design as bypass capacitors.
The essence of a capacitor is to pass alternating current and block direct current. Theoretically, the larger the capacitor used for power supply filtering, the better.
However, due to lead wires and PCB layout, a capacitor is actually a parallel circuit of an inductor and a capacitor (and the capacitor's own resistance is sometimes also not negligible).
This introduces the concept of resonant frequency: ω = 1/(LC)¹/²
Below the resonant frequency, a capacitor is capacitive; above the resonant frequency, a capacitor is inductive.
Therefore, large capacitors generally filter low-frequency waves, while small capacitors filter high-frequency waves.
This also explains why STM-packaged capacitors with the same capacitance value have a higher filtering frequency than DIP-packaged capacitors.
As for the appropriate capacitor size, this is just a reference.
Capacitor resonant frequency
Capacitance value DIP (MHz) STM (MHz)
1.0μF 2.5 5
0.1μF 8 16
0.01μF 25 50
1000pF 80 160
100 pF 250 500
10 pF 800 1.6 (GHz)
However, this is just for reference; as the veteran engineer would say, it mainly relies on experience.
A more reliable approach is to connect two capacitors, one large and one small, in parallel, generally requiring a difference of more than two orders of magnitude, to obtain a wider filtering frequency band.
Generally speaking, large capacitors filter out low-frequency waves, while small capacitors filter out high-frequency waves. The capacitance value is inversely proportional to the square of the frequency you want to filter.
The specific capacitor selection can be done using the formula C = 4Pi * Pi / (R * f * f )
Selecting power supply filter capacitors is not difficult if you grasp the essence and methods.
1) Theoretically, the impedance of an ideal capacitor decreases with increasing frequency (1/jwc). However, due to the inductive effect of the capacitor leads, the capacitor should be considered as an LC series resonant circuit. The self-resonant frequency is the FSR parameter of the device. This means that when the frequency is greater than the FSR value, the capacitor becomes an inductor. If the capacitor is used for filtering to ground, the suppression of interference is greatly reduced when the frequency exceeds the FSR. Therefore, a smaller capacitor is needed in parallel with ground. Can you think about why?
The reason is that small capacitors have a large SFR (Self-Resonant Frequency) value, which provides a path to ground for high-frequency signals. Therefore, in power supply filtering circuits, we often understand it this way: large capacitors filter low frequencies, and small capacitors filter high frequencies. The fundamental reason is the different SFR (Self-Resonant Frequency) values. Of course, you can also think about why. If you think about it from this perspective, you can understand why the capacitor's ground pin should be as close to ground as possible in power supply filtering.
2) In practical design, we often have questions: How do I know the SFR (Surface Rate of Return) of a capacitor? Even if I know the SFR value, how do I choose capacitors with different SFR values? Should I choose one capacitor or two?
The SFR (Signal Frame Rate) of a capacitor is related to its capacitance value and the lead inductance. Therefore, 0402 and 0603 capacitors with the same capacitance value, or through-hole capacitors, will not have the same SFR value. There are two ways to obtain the SFR value:
1) Component datasheet, such as the SFR value of a 22pF0402 capacitor, which is around 2G.
2) Measure its self-resonant frequency directly using a network analyzer. Consider how to measure S21.
Once you know the capacitor's SFR value, use software simulation, such as RFsim99. Choose one or two circuits depending on whether your power supply circuit's operating frequency band has a sufficient noise suppression ratio. After simulation, the next step is actual circuit testing. For example, when adjusting a mobile phone's receiver sensitivity, the LNA's power supply filtering is crucial; good power supply filtering can often improve the sensitivity by several dB.
To put it simply, think of a capacitor as a leaking tank, and the arrival of AC peaks as adding water to the tank. If the amount of water leaking is the same, then if the frequency of adding water is high, you need to use a smaller tank to ensure a high water level. Conversely, if the frequency of adding water is low, and the tank is too small, the water level will drop significantly before the second water arrives. Therefore, you need to use a larger tank to mitigate the drop in water level caused by leakage.
Article 2: Why is a small capacitor connected in parallel with a large capacitor?
Because large capacitors have a large capacitance, they are generally also large in size, and they are usually made using a multi-layer winding method (those who have disassembled aluminum electrolytic capacitors should have a good understanding of this, and those who haven't can try disassembling several different capacitors). This results in a relatively large distributed inductance (also called equivalent series inductance, abbreviated as ESL) in large capacitors.
As we all know, inductors have a high impedance to high-frequency signals, so large capacitors have poor high-frequency performance. Conversely, some small-capacity capacitors, due to their small capacitance, can be made very small (shortening the leads reduces ESL, since a section of wire can be considered an inductor). They often use the parallel-plate capacitor structure, resulting in very low ESL and excellent high-frequency performance. However, due to their small capacitance, they have a high impedance to low-frequency signals.
Therefore, to allow both low-frequency and high-frequency signals to pass through effectively, we use a large capacitor connected in parallel with a small capacitor. A commonly used small capacitor is a 0.1uF ceramic capacitor. For higher frequencies, even smaller capacitors, such as a few pF or hundreds of pF, can be connected in parallel. In digital circuits, a 0.1uF capacitor is typically connected in parallel to ground on the power supply pin of each chip (this capacitor is called a decoupling capacitor, or power supply filter capacitor; the closer it is to the chip, the better), because the signals in these locations are mainly high-frequency signals, and a smaller capacitor is sufficient for filtering.
Capacitor capacitance formula for series and parallel connections - Voltage divider formula for series and parallel connections of capacitors
1. Series connection formula: C = C1*C2/(C1 C2) 2. Parallel connection formula: C = C1 C2 C3
Supplementary section:
Series voltage division ratio V1 = C2/(C1 + C2)*V ........The larger the capacitor, the smaller the voltage divided; this applies to both AC and DC conditions. Parallel current division ratio I1 = C1/(C1 + C2)*I ........The larger the capacitor, the larger the current passing through; however, this is under AC conditions.
A small capacitor is connected in parallel with a large capacitor.
Because of their large capacitance, large capacitors are generally also large in size, and they are usually made using a multi-layer winding method. This results in a large distributed inductance (also called equivalent series inductance, abbreviated as ESL) in large capacitors.
Inductors have a high impedance to high-frequency signals, so large capacitors have poor high-frequency performance. Conversely, some small-capacity capacitors, due to their small capacitance, can be made very small (shortening the leads reduces ESL, since a section of wire can be considered an inductor). They often use a parallel-plate capacitor structure, resulting in very low ESL and excellent high-frequency performance. However, due to their small capacitance, they have a high impedance to low-frequency signals.
Therefore, if we want both low-frequency and high-frequency signals to pass through well, we use a large capacitor and then connect a small capacitor in parallel.
For commonly used small capacitors, 0.1uF CBB capacitors are preferred (ceramic capacitors are also acceptable). For higher frequencies, even smaller capacitors, such as a few pF or hundreds of pF, can be connected in parallel. In digital circuits, a 0.1uF capacitor is typically connected in parallel to ground on the power supply pin of each chip (this capacitor is called a decoupling capacitor, or a power supply filter capacitor; the closer to the chip, the better). This is because the signals in these locations are primarily high-frequency signals, and a smaller capacitor is sufficient for filtering.
An ideal capacitor has impedance that decreases with increasing frequency (R=1/jwc). However, an ideal capacitor does not exist. Due to the distributed inductance effect of the capacitor leads, at high frequencies, a capacitor is no longer a simple capacitor but should be considered as a series equivalent circuit of a capacitor and an inductor. When the frequency is higher than its resonant frequency, the impedance exhibits the characteristic of increasing with increasing frequency, which is the characteristic of inductance. At this time, the capacitor is like an inductor. Conversely, an inductor has the same characteristic.
The use of large capacitors in parallel with small capacitors is very common in power supply filtering, primarily due to the self-resonant characteristics of capacitors. This combination of large and small capacitors effectively suppresses power supply interference signals from low to high frequencies. The small capacitor filters high frequencies (high self-resonant frequency), while the large capacitor filters low frequencies (low self-resonant frequency), complementing each other.
