The relation between capacitance and voltage dictates that while an ideal capacitor's physical capacitance is fixed by its geometry, the effective capacitance of real-world components—especially Class II ceramics—decreases significantly as applied DC voltage approaches the component's rated limit. This phenomenon changes your circuit's actual filter cutoff frequencies, reduces decoupling effectiveness on power rails, and alters energy storage in pulse-load applications. Beginners commonly confuse a capacitor's Working Voltage DC (WVDC) rating—the absolute maximum voltage before dielectric breakdown—with the much lower voltage threshold where capacitance actually begins to drop off.
The Physics: Charge, Voltage, and the Dielectric Trap
At the bench level, the fundamental equation is C = Q / V. Capacitance (C) is the ratio of stored charge (Q) to the applied voltage (V). In an ideal component, C is a constant determined strictly by plate area, distance between plates, and the dielectric material's permittivity. If you double the voltage, you double the stored charge, and the capacitance remains unchanged.
To visualize this, use the water analogy exactly once: think of a capacitor as a flexible rubber membrane stretched across a water pipe. Voltage is the water pressure pushing against the membrane, and capacitance is how much the membrane stretches to hold a volume of water (charge). In an ideal world, the membrane stretches linearly forever. But in real Class II ceramic dielectrics (like the Barium Titanate used in X7R and X5R capacitors), the molecular dipoles inside the dielectric align with the electric field. Once most dipoles are aligned at higher voltages, the 'membrane' stiffens. It cannot stretch as much for the same increase in pressure, meaning the effective capacitance drops. This is known as the Voltage Coefficient of Capacitance (VCC) or DC bias derating.
Worked Numeric Example: The 10µF X7R Illusion
Let's look at a common mistake on the workbench. You are designing the output filter for a 12V-to-5V buck converter. The datasheet calls for 10µF of output capacitance to maintain loop stability and keep ripple under 15mV.
You select a standard 10µF, 16V, 1206 X7R MLCC (e.g., Murata GRM31CR71C106KA12, costing about $0.12 in bulk). You solder it on, power up the 5V rail, and measure the output ripple. It's 45mV, and the converter is oscillating slightly. Why?
- Nameplate Value: 10µF at 0V DC bias.
- Operating Voltage: 5V DC.
- DC Bias Derating: At 5V (which is ~31% of the 16V rating), the X7R dielectric's effective capacitance drops by roughly 40%.
- Effective Value in Circuit: 6µF.
Your LC filter's resonant frequency just shifted higher because your actual capacitance is 6µF, not 10µF. The phase margin collapsed, causing the oscillation. To fix this, you must select a capacitor based on its effective capacitance at your operating voltage, which usually means stepping up the voltage rating or the physical package size.
Where You Meet This in Practice
Understanding the relation between capacitance and voltage is not just academic; it dictates component selection across several core disciplines:
1. Power Rail Decoupling
Modern FPGAs and SoCs have core rails running at 0.8V and I/O rails at 3.3V. A 0.8V rail barely stresses a 6.3V rated capacitor, so you get nearly 100% of the nameplate capacitance. On the 3.3V rail, that same 6.3V cap loses 20-30% of its value. Engineers often use 10V or 16V rated caps on 3.3V rails purely to avoid DC bias derating, not because they expect voltage spikes.
2. Audio AC Coupling
When routing audio signals through a series capacitor to block DC, the op-amp's common-mode voltage acts as a DC bias across the capacitor. If you use a high-value X5R ceramic for a low-frequency high-pass filter, the DC bias will shrink the capacitance, shifting your -3dB cutoff frequency higher and rolling off your bass response.
3. Energy Storage and Pulse Loads
The energy stored in a capacitor is E = 0.5 × C × V². If you are building a camera flash circuit or a GSM modem pulse-load buffer, you calculate energy assuming C is constant. If C drops by 30% at your peak charging voltage, your available energy drops proportionally, leading to brownouts during high-current transmit bursts.
Decision Tree: Picking the Right Capacitor for Your Voltage
Use this decision path to select the correct component based on your circuit's DC voltage and precision requirements. Do not rely on nameplate values for power filtering.
| Circuit Condition | Required Action | Concrete Part Pick |
|---|---|---|
| Signal path < 1V, requires exact timing/filtering | Use Class I (C0G/NP0). Ignore DC bias derating entirely. | KEMET C0805C104F5GACTU (0.1µF, 50V, 1% tol) |
| Power rail < 5V, high ripple current | Use X5R/X7R but apply a 3x voltage overhead rule to keep derating under 10%. | Murata GRM21BR71E106MA73 (10µF, 25V, 0805 X7R) |
| Power rail > 12V, needs high C and low ESR | Abandon MLCCs. Use Polymer Tantalum which has zero DC bias derating. | KEMET T520D107M016ATE015 (100µF, 16V, Polymer) |
| Space-constrained 3.3V rail, needs 10µF | Use two 0805 10µF 10V caps in parallel rather than one 1206, to distribute field density. | 2x Samsung CL21A106KAYNNNE (10µF, 25V, 0805) |
Common Confusions and FAQ
Does putting capacitors in series increase the voltage rating without losing capacitance?
Putting two identical 10µF, 16V capacitors in series gives you a 32V rating, but the total capacitance drops to 5µF. Furthermore, unless you use balancing resistors, the DC voltage will not split evenly due to differences in leakage current, meaning one capacitor might still suffer severe DC bias derating while the other doesn't. For higher voltage handling, it is almost always better to buy a single capacitor with a higher native WVDC rating.
Why do smaller physical packages derate worse than larger ones?
A 10µF 16V capacitor in an 0402 package will lose capacitance much faster under DC bias than a 10µF 16V capacitor in a 1206 package. To fit the same capacitance into a smaller volume, manufacturers must use thinner dielectric layers or higher-permittivity (and more volatile) Barium Titanate formulations. Thinner dielectrics experience a higher electric field strength (V/m) at the same applied voltage, accelerating dipole saturation. Always check the manufacturer's DC bias curves for the specific package size, not just the capacitance and voltage rating.
Do electrolytic and film capacitors suffer from DC bias derating?
No. Aluminum electrolytic, tantalum, and film capacitors (like polyester or polypropylene) do not exhibit the ferroelectric dipole saturation seen in Class II ceramics. Their capacitance remains stable regardless of applied DC voltage. However, they have their own voltage-related quirks: electrolytics require a minimum DC bias to maintain their oxide layer, and film capacitors are physically massive compared to MLCCs for the same microfarad value.
For deeper analysis of specific part numbers, always consult Murata's official MLCC design and selection resources or reference the material science breakdown of dielectric classes in Wikipedia's technical documentation on ceramic capacitors. Never assume the number printed on the tape and reel is the value your circuit will actually see at runtime.
