Voltage Dependent Capacitance in Multilayer Ceramic Capacitors

Abstract  The value printed on a high dielectric constant MLCC is not necessarily the capacitance available in circuit. DC bias and AC excitation change the polarization state of the ferroelectric dielectric, so effective capacitance depends on operating conditions. Ripple, filtering, and loop stability calculations should use the biased value rather than the catalog headline.

1 Nominal Capacitance Is a Test Result

Data sheet capacitance is measured under a defined frequency, AC level, and temperature. In service, the part sees DC voltage, ripple, and temperature at the same time. Those conditions differ from the catalog test, and high dielectric constant materials such as X5R, X7R, and Y5V can respond strongly to the difference.

Voltage rating defines an electrical stress boundary; it does not specify the capacitance remaining at a particular bias. Parts with the same nominal capacitance and rating may have different bias curves because dielectric composition, package size, and internal layer geometry differ.

2 DC Bias Reduces Incremental Capacitance

Barium titanate ferroelectrics contain domains whose polarization changes with applied field. As the field increases and polarization approaches saturation, the incremental dielectric constant falls. The small-signal capacitance measured around that operating point therefore decreases. C0G dielectrics, film capacitors, and some polymer technologies show much less of this effect.

Murata illustrates the point with a 100 microfarad 6.3 volt MLCC at 1.8 volts DC. The example X5R part loses about 10 percent, while the Y5V part loses about 40 percent. These are product-specific examples, not universal derating factors, but they show why a nominal-only calculation can be misleading.

3 AC Excitation Changes the Measured Value

Ferroelectric nonlinearity and hysteresis also create an AC voltage characteristic. Changing the excitation amplitude changes the average slope of the D E relationship, and therefore the apparent capacitance. In the source example, a 22 microfarad 6.3 volt part measured at 120 hertz and 0.2 volts RMS shows roughly 10 percent reduction for X5R and 20 percent for Y5V.

Capacitance data should always be tied to test frequency and AC level. A bridge measurement made with a very small signal may not represent a switching regulator carrying appreciable ripple.

4 Effective Capacitance Changes System Behavior

Loss of capacitance increases ripple, moves filter poles, and reduces charge available during a load step. Temperature dependence, aging, tolerance, and bias can stack in the same direction. For a regulator whose stability depends on an allowed output-capacitance range, an optimistic value can become a control-loop problem.

Simply choosing a larger nominal value is not always the best correction. A larger case, higher voltage rating, or more stable dielectric may retain more capacitance at the operating point. Parallel parts or another capacitor technology can also help, but ESR, ESL, and resonances must be reassessed.

5 Select from Operating Point Data

Define maximum DC bias, ripple amplitude, temperature, and the minimum capacitance required by the circuit. Read effective capacitance from manufacturer curves or a component modeling tool, then include negative tolerance, temperature shift, and aging.

On hardware, measure impedance or small-signal capacitance while the real DC bias is applied, and confirm the result with load-step testing. Record a minimum effective-capacitance requirement in the BOM so that a future substitute cannot pass on nominal value alone.

Design Review Matrix