A bypass capacitor (also called a decoupling capacitor) is a capacitor placed between a power-supply rail and ground, physically close to a load device, to suppress high-frequency noise on the supply and provide a local reservoir of charge during transient current demands. The terms "bypass" and "decoupling" are often used interchangeably in practice, though some engineers reserve "bypass" for RF applications and "decoupling" for digital logic supply filtering, and the distinction can carry more weight in RF or power-integrity contexts.
In practice
In digital embedded designs, bypass capacitors are placed as close as possible to each VDD/VCC pin of every IC, with the capacitor's ground terminal connecting directly to the IC's ground pin or nearby ground plane via the shortest possible path. A common starting point is a 100 nF ceramic capacitor (X5R or X7R dielectric, typically in 0402 or 0603 package) per supply pin for high-frequency decoupling, though some ICs require multiple capacitors per pin, different values, or specific types per their datasheet. This is supplemented by a larger bulk capacitor (1 uF to 10 uF) per device or per power domain to handle lower-frequency transients. Mixing capacitor values takes advantage of their differing self-resonant frequencies, so together they present low impedance across a broader frequency range. The blog post "Two Capacitors Are Better Than One" explores this parallel combination effect in detail.
The physical placement and routing of bypass capacitors matter as much as their values. A capacitor placed 5 mm away with long trace stubs adds parasitic inductance that degrades its effectiveness at high frequencies, though the magnitude of this effect depends on trace geometry, layer stackup, and operating frequency. The ideal layout connects the capacitor's pads directly to the IC supply pin and to a via into the ground plane with no intervening trace length. In high-speed or high-current designs, these parasitics can turn a bypass capacitor into a resonator rather than a filter.
For power electronics and motor-drive applications, bypass and DC link capacitors face additional stresses. Electrolytic capacitors are commonly used for bulk energy storage in H-bridge and DC-DC converter designs, but they have significant equivalent series inductance (ESL) that makes them ineffective at suppressing fast switching transients. Ceramic capacitors in parallel are typically required to handle high-frequency ripple, a topic covered in depth in "Lost Secrets of the H-Bridge, Part IV: DC Link Decoupling and Why Electrolytic Capacitors Are Not Enough" and the related H-Bridge series.
A less obvious application is bypassing the analog supply and reference pins of ADCs and DACs. Here, the goal is not just transient response but also preventing digital switching noise from coupling into sensitive analog circuitry. Ferrite beads or small series resistors are sometimes combined with bypass capacitors to form an LC or RC low-pass filter on the analog supply pin. The blog post "The Other Kind of Bypass Capacitor" addresses bypass strategies that go beyond the standard digital decoupling use case.
Discussed on EmbeddedRelated
Frequently asked
What value bypass capacitor should I use?
For general digital IC decoupling, 100 nF (0.1 uF) ceramic in an X5R or X7R dielectric is the most common starting point, placed on every VDD pin. Add a 1 uF to 10 uF ceramic or electrolytic per power domain for bulk charge storage. For precision analog pins (
ADC VREF, AVDD), values from 1 nF to 10 uF are common depending on the datasheet recommendation. Always check the IC datasheet first, as some devices specify exact
capacitor types and values.
Why does capacitor placement matter so much?
Every millimeter of PCB trace between the
capacitor and the IC's supply pin adds parasitic inductance, typically on the order of 0.5 nH to 1 nH per mm. At high frequencies (tens to hundreds of MHz), this inductance dominates and the capacitor loses its effectiveness. Keeping the capacitor immediately adjacent to the supply pin, with short direct connections to both the power rail and the ground plane, minimizes this parasitic inductance.
What is the difference between a bypass capacitor and a decoupling capacitor?
In practice, the terms are used interchangeably by most embedded engineers. Historically, 'bypass' referred to shunting high-frequency
RF signals around a component to ground, while 'decoupling' referred to isolating one part of a circuit from power-supply noise generated by another. Modern usage blurs this distinction, and the same 100 nF ceramic
capacitor on a VDD pin may be called either.
Why use two capacitors of different values in parallel instead of one larger capacitor?
Every
capacitor has a self-resonant frequency (SRF) above which it behaves inductively rather than capacitively. A 100 nF ceramic might have an SRF around 30-50 MHz in a typical package as a rough example, while a 10 uF ceramic or electrolytic resonates at a much lower frequency; actual SRF values vary widely with package, dielectric, voltage rating, and mounting inductance. Placing both in parallel means the smaller capacitor handles high-frequency transients and the larger one handles lower-frequency bulk current demand, resulting in lower supply impedance across a wider frequency band. The blog post 'Two Capacitors Are Better Than One' analyzes this effect quantitatively.
Why aren't electrolytic capacitors alone sufficient for bypassing fast-switching circuits like H-bridges?
Electrolytic
capacitors have relatively high equivalent series inductance (ESL), often in the range of 10 nH to 50 nH or more for many common package styles, though actual values vary by part and construction. For fast power-switch transitions in motor drivers or DC-DC converters, the di/dt during switching can be large enough that the ESL of the electrolytic dominates its impedance, causing significant voltage spikes. Small ceramic capacitors placed close to the switch nodes are typically required in parallel to absorb these fast transients. This is discussed in 'Lost Secrets of the H-Bridge, Part IV: DC Link Decoupling and Why Electrolytic Capacitors Are Not Enough.'
Differentiators vs similar concepts
The term "
decoupling capacitor" is effectively synonymous in most embedded contexts. "Bulk
capacitor" refers specifically to a large-value capacitor (often electrolytic, tens to thousands of uF) whose primary role is energy storage and low-frequency supply stabilization rather than high-frequency noise suppression. A "filter capacitor" is a broader term that includes bypass capacitors but also describes capacitors used in deliberate LC or RC filter networks. "DC link capacitor" is a term specific to power electronics, referring to the bulk capacitor across the DC bus in inverters and motor drives, which handles both energy storage and ripple current absorption.
