The core function of bypass capacitors is to filter out high-frequency noise. In power networks, the switching action or signal transitions of integrated circuits generate high-frequency noise (such as ringing and glitches). Bypass capacitors provide a low-impedance path (ideally impedance Z = 1/(2πfC)) to directly conduct noise energy to the ground plane, preventing interference with other circuit modules. For example, a 0.1μF capacitor is typically connected in parallel to the power supply pins of digital ICs, which can effectively filter out noise below 100MHz (refer to the Murata technical datasheet).
To stabilize local voltage, when the load current changes abruptly (such as a sudden high CPU load), the power supply line will experience a voltage drop due to parasitic inductance (typically 1-10 nH/cm). Bypass capacitors, acting as "miniature energy storage units," can release charge within μs to compensate for the voltage drop. For example, a 10μF + 0.1μF combined capacitor (Xilinx UG-583) is recommended for each power supply pin in FPGA power supply.
Turning off the PMOS transistor does not generate impulse noise because the PMOS transistor was previously on and no current was flowing through it. Simultaneously turning on the NMOS transistor creates a loop consisting of the transmission line, ground plane, package inductor Lg, and the NMOS transistor. A momentary current flows through switch B, causing a disturbance at the ground junction inside the chip, raising the reference level. This disturbance is known as ground bounce noise in power systems.
In a real power supply system, all interconnects, including chip pins, PCB traces, power layers, and bottom layers, possess a certain inductance. Therefore, the SSN and ground bounce noise analyzed at the IC level above exist in the same way when performing board-level analysis, and are not limited to the chip's internal components. From the perspective of the entire power distribution system, this is what is known as power supply voltage sag noise. Because the switching operations of the chip output and the operations within the chip require instantaneous extraction of large currents from the power supply, and the power supply characteristics cannot respond quickly to these current changes—even high-speed switching power supplies have switching frequencies only on the order of MHz—to ensure that the voltage on the power supply line near the chip does not drop beyond the tolerance specified in the device datasheet due to SSN and ground bounce noise, an energy storage capacitor needs to be provided near the chip to meet the high-speed current demand. This is what we call a decoupling capacitor.
Therefore, there are three important distributed parameters for a capacitor: equivalent series resistance (ESR), equivalent series inductance (ESL), and equivalent parallel resistance (EPR/Rp). Among these, ESR and ESL are the most important. In practice, when analyzing capacitor models, only a simplified RLC model is typically used, i.e., analyzing the capacitor's capacitance (C), ESR, and ESL. Due to the influence of parasitic parameters, especially ESL, the frequency response of a real capacitor exhibits a "V"-shaped curve between impedance and frequency. At low frequencies, the capacitor impedance decreases as the frequency increases; at the lowest point, the capacitor impedance equals ESR; thereafter, as the frequency increases, the impedance increases, exhibiting inductive characteristics (thanks to ESL). Therefore, selecting a capacitor requires considering not only the capacitance value but also other factors.
All considerations are aimed at reducing the inductive reactance between the power supply and ground (while meeting the maximum capacitive reactance of the power supply) so that when a large instantaneous current flows through the power supply system, it will not generate large noise interference to the power supply and ground pins of the chip.
Frequency characteristics of capacitors
At high frequencies, capacitance is no longer considered a lumped parameter, and the influence of parasitic parameters becomes significant. Parasitic parameters include Rs, the equivalent series resistance (ESR), and Ls, the equivalent series inductance (ESL). The actual equivalent circuit of a capacitor shows C as the electrostatic capacitance, 1Rp as the leakage resistance (also called insulation resistance), with a larger value (typically above the gigawatt-hour level) indicating lower leakage current and more reliable performance. Because Pp is usually very large (above the gigawatt-hour level), it can be ignored in practical applications. Cda and Rda represent the dielectric absorption capacitance and dielectric absorption resistance, respectively. Dielectric absorption is an internal charge distribution with hysteresis that allows a capacitor in an open-circuit state after rapid discharge to recover some charge.
Working principle and frequency characteristics
1. Relationship between capacitive reactance and frequency
The impedance of a bypass capacitor decreases as the frequency increases, but it becomes inductive after the self-resonant frequency (SRF) due to the parasitic inductance (ESL). For example:
- 0.1μF MLCC capacitor in 0805 package, SRF approximately 20MHz (ESL approximately 1nH)
- The SRF of the same packaged 1μF capacitor drops to 5MHz (TDK parameter manual).
2. Multi-capacitor parallel connection strategy
To cover a wide range of noise frequencies, a combination of "large capacity + small capacity" is often used:
- 10μF aluminum electrolytic capacitor: handles low-frequency ripple below 1kHz
- 0.1μF ceramic capacitor: suppresses noise from 10-100MHz
- 1nF high-frequency capacitor: for GHz-level interference (such as RF circuits).
III. Key Points for Selection and Layout
1. Selection of key parameters
- Capacitance: 0.01-0.1μF is commonly used in digital circuits, while pF level is required for RF circuits.
- Voltage withstand: at least 1.5 times the power supply voltage (e.g., select 10V specification for a 5V system).
- Material: NP0/C0G ceramic is preferred for high-frequency applications (temperature stability ±30ppm/℃).
2. PCB Layout Specifications
- Keep it as close as possible to the IC power pin (distance <3mm)
- Prioritize the use of short, wide traces to reduce ESL.
- Avoid vias interrupting the return path (refer to Intel PCB Design Guidelines).
Energy Storage Principle: Electrolytic capacitors generally possess energy storage capabilities. For capacitors with specific energy storage needs, their energy storage mechanisms are mainly based on double-layer capacitance and Faraday capacitance, i.e., supercapacitor energy storage. As a capacitor utilizing the double-layer principle, a supercapacitor stores positive and negative charges respectively on its positive and negative plates when an external voltage is applied, forming an electric field. Under the influence of this electric field, opposite charges are generated at the interface between the electrolyte and the electrodes to maintain the internal electric field balance of the electrolyte. These positive and negative charges are arranged in opposite positions with extremely short gaps on the contact surface of the two plates, forming a double layer, thereby significantly increasing the capacitance.
The basic principle of capacitor bypass is that it's a circuit design technique that connects a capacitor in series next to a circuit element, allowing current to flow more smoothly through the circuit. When high-frequency noise occurs in the circuit, capacitor bypass also helps reduce electromagnetic interference.
The basic principle of capacitor bypass is to use the equivalent circuit model of a capacitor to eliminate high-frequency signals and improve the operating stability of the circuit. Capacitor bypass can effectively reduce the impact of high-frequency noise on signals and improve the circuit's anti-interference capability.
Capacitor bypass is commonly used in power supply filtering to suppress power supply noise and improve the power supply stability of equipment. Power supply filtering circuits typically use low-pass filters composed of capacitors and inductors to filter high-frequency noise, ensuring normal equipment operation. In analog circuits, capacitor bypass is mainly used to attenuate high-frequency signals and maintain signal integrity. When signal circuits need to suppress high-frequency noise, capacitor bypass can be connected in series at the input, output, or signal path to reduce or eliminate the interference of high-frequency noise on the signal.
In digital circuits, capacitor bypass is primarily used to suppress high-frequency noise in the power supply and maintain circuit stability. In digital circuit chips, capacitor bypass typically consists of a low-pass filter circuit composed of a capacitor and a resistor to eliminate high-frequency noise in the power supply.
In amplifier design, capacitor bypass is commonly used to reduce noise signals and improve signal purity. Inserting capacitors in the input and output stages of an amplifier can reduce the impact of power supply noise on the signal and improve amplifier performance.
