# Don’t Bypass Your Capacitor!

Fig 1. This top side of a two-layer printed-circuit board was milled to compare the footprints of different power solutions. Notice that the bypass capacitor is placed as close as possible to each chip (LDO, switching regulator, and power module).

Fig 2. A basic capacitor structure is shown on the left, and the equations generally define dielectric constant, voltage, and current in such a capacitor.

Fig 3. The plot illustrates the impedance of an actual capacitor (non-ideal).

Fig 4. Here, impedance of an actual capacitor (non-ideal) is given for different surface-mount packages.

Dielectric Constant Of Materials

Among everyday electronic components, few are as misperceived as the bypass capacitor. Generally, designers understand that circuits, systems, and individual ICs need to be bypassed. Too often, though, they assume that simply tossing one or two capacitors into designs will handle any unwanted noise. They might even take the next step to find a couple of good ones based on voltage and capacitance, and then it’s off to a long lunch.

Well, not so fast. While capacitors are able to resolve noise issues, they can be vulnerable to parasitic resistance and inductance, with performance affected by temperature, voltage, mechanical effects, and other factors. The wrong choice or positioning may actually create added noise, power losses, or unstable circuit behavior. Thus, it’s important to consider how and when to apply different types of bypass capacitors and account for parameters beyond capacitance and voltage rating.

Methods for choosing bypass capacitors typically follow a traditional path, rather than selecting them with an eye toward circuit optimization. All design aspects must be analyzed, based on a fundamental understanding of equivalent circuits, dielectrics, and capacitor options.

In addition to the fundamentals discussed here, one should prepare for unknown and unwanted noise. Many times, a circuit in an ideal configuration will not work efficiently because noise coupled into the circuit from the power supply or other close-proximity ICs on the board. This can occur because wires and board connections act like antennas, causing power-supply levels to change with current draw.

Also, lots of high-frequency noise can displace the dc level, maybe 10 mV p-p with a switched supply. Subsequently, regular spikes in excess of 50 mV may occur due to load transients. Many designers mistakenly assume that power supplies are stable and have a constant dc voltage.

These unwanted perturbations (if not controlled) can couple directly into the circuit and cause instability or damage. In this case, the bypass capacitor is a first line of defense. It eliminates voltage droops on the power supply by storing electric charge to be released upon the occurrence of a voltage spike. Moreover, it accomplishes this task at a wide range of frequencies by creating a low-impedance path to ground for the supply.

**The Big Choice **

Now that we’ve defined the role and environment for a capacitor, it’s time to get down to details. Next on the agenda are size requirements, its placement for maximum effect, and selection of the capacitor that best optimizes the circuit or system design. It’s also important to choose the right package based on prior decisions about size, type, and placement.

Start by defining placement. A bypass capacitor should be located as close as possible to the power-supply pin of each chip (Fig. 1). Any extra distance translates into additional series inductance, which lowers the self-resonant frequency (useful bandwidth) of the bypass capacitor.

**Capacitor Basics**

Capacitor size, type, and package choice are at the heart of the immediate discussion. But first, let’s cover some capacitor basics.

Generally, a capacitor is two conductive plates separated by a dielectric material. As charge collects on the plates, an electric field builds across the dielectric. The amount of charge needed to create a certain potential between the plates is called capacitance, which is measured in farads. Capacitance also can be measured by the dimensions of the plates and quality of the dielectric (Eq. 1 in Fig. 2).

Capacitance increases as the area of the plates grows in size, since more charge can be stored as the potential is created. The distance between the plates dictates the attraction between charges stored on them. For instance, the more distance between plates diminishes the interaction, which decreases the capacitance (Eq. 2 in Fig. 2).

The last of the basic equations involves current. By definition, current is the movement of charge (Eq. 3 in Fig. 2). Therefore, charge will move only when the voltage (potential between the plates) is changing. If the voltage is constant, the charge that’s forming must also be constant—thus, there’s no current flow.

Another factor of capacitance is dielectric quality. The dielectric—the material between the two conductors forming a capacitor—has a high impedance and does not allow significant current to flow from one plate to the other. Different materials used as a dielectric have varying amounts of temperature stability, breakdown voltages, and loss coefficients. The materials in the table are accompanied by their dielectric constant (epsilon), the coefficient that directly relates to a structure’s capacitance through Equation 1 in Figure 2.

Several kinds of capacitors meet the requirements outlined earlier. Ceramic capacitors are the most common due to their low cost, wide range of values, and solid performance. Tantalum, Oscon, and aluminum electrolytic capacitors are all polarized, specifically to be used as bypass capacitors.

Tantalum found its niche in low-voltage systems. Aluminum electrolytic capacitors are often used in low-to medium-frequency systems. However, they’re infrequently found in switching circuits because they hold the charge too well, which doesn’t suit them for the rapid cycling of production testing.

Oscon is a special capacitor type developed to provide low parasitics, wide frequency range, and full temperature range. Basically, it offers the best quality available—at the highest price tag. If you have the budget, these capacitors will ensure quality bypass for any circuit.

Mica and plastic-film capacitors are included for completeness. Their primary use involves filter design rather than bypass.

**The Package**

Choose the package after selecting dielectric material, dielectric quality, temperature range, acceptable leakage, and voltage range requirements. Typically, package size is chosen by “what was used last time,” what is big enough to solder by hand, or what is small enough to fit in the box.

Remember that the equivalent circuit model will change with different packages. The main issue revolves around equivalent series inductance (ESL). Obviously, a capacitor structure is constant as long as the capacitance value is constant. If that same capacitor is available in a variety of packages, then the connections must change between the plates and the outer dimensions of the package. This appears as additional series resistance and series inductance. The smaller the package, the smaller the series parasitics.

**Sizing Bypass Capacitors**

Capacitors are usually sized by convention or typical values of 1 μF and 0.1 μF. Simply speaking, the larger value handles the lower frequencies and high-current issues, while the smaller value handles higher frequencies. The need for multiple capacitors comes from the parasitics associated with real capacitors.

Figure 3 provides a representation of capacitor impedance. The axes are left blank so the values can be scaled to fit any capacitor. The left half of the curve shows the traditional (and ideal) capacitor response—as frequency increases, the impedance of the capacitor decreases. This is desirable, since bypass capacitors deliver a low impedance (effectively a short) to ac signals on the power line.

The negative slope of the line is constant, but the lateral placement of the line depends on the capacitor’s size. For example, a larger capacitor would shift the left half of the curve lower in frequency (farther to the left).

Any inductance in the capacitor package will cause a positive slope, as seen in the right half of the plot. In this region of frequencies, the inductance cancels and then dominates the low impedance provided by the capacitor.

Because the impedance value relates to the size and construction of the bypass capacitor, the frequency response is also related (Fig. 4). It’s important, then, to check the datasheet to ensure the package for your bypass capacitor enables the low impedance necessary for your system’s frequencies. The ESLs are in the range of hundreds of picohenries. Their rising effect on impedance only emerges when system frequencies exceed 100 MHz.

The criteria above cover the basics for selecting the appropriate bypass capacitor. One key factor to remember: a single capacitor isn’t going to be sufficient for many wideband systems. Instead, paralleling multiple capacitors becomes the solution. Such paralleling also reduces the paralleled ESL and ESR.