A capacitor stores energy in an electric field between two conductive plates separated by an insulator (the dielectric). When voltage is applied, electrons pile up on one plate and repel electrons from the other, creating a charge imbalance. It does not store electrons like a bucket holds water; it stores potential energy via electrostatic tension. When the circuit demands it, that tension drives current back into the system.
If you are designing a power supply, filtering an audio signal, or decoupling a microcontroller, understanding how a capacitor works beyond the textbook definition is the difference between a stable board and a smoking prototype. This guide skips the abstract theory and focuses on the physics, markings, selection criteria, and failure modes you actually encounter at the workbench.
The Core Mechanism: Dielectric Polarization and the Math
The capacitance ($C$) of a parallel-plate capacitor is governed by a single physical equation:
Where: $\varepsilon_r$ is the relative permittivity (dielectric constant) of the insulator, $\varepsilon_0$ is the permittivity of free space, $A$ is the overlapping area of the plates, and $d$ is the distance between them.
To get more capacitance, you must increase the area ($A$), decrease the distance ($d$), or use a material with a higher dielectric constant ($\varepsilon_r$). This is exactly how a Multilayer Ceramic Chip Capacitor (MLCC) achieves 100µF in a tiny 1210 SMD package. Manufacturers stack hundreds of microscopic ceramic layers (increasing $A$), separated by dielectric layers only micrometers thick (decreasing $d$), using barium titanate which has a massive dielectric constant in the thousands.
However, physics demands a trade-off. High-$\varepsilon_r$ materials (Class II dielectrics like X7R and X5R) are non-linear. A 10µF X7R capacitor rated at 16V might only provide 2µF of actual capacitance when 12V DC is applied across it—a phenomenon known as DC bias derating. Class I dielectrics (C0G/NP0) have a low $\varepsilon_r$ (requiring larger physical sizes for the same capacitance) but offer perfect linearity and zero DC bias drop.
Decoding Capacitor Markings: What the Codes Actually Mean
Reading a resistor is straightforward, but capacitor markings are notoriously cryptic due to space constraints on the physical part. Here is how to decode them on the bench.
Through-Hole Ceramic Disc Codes
Ceramic disc capacitors use a three-digit code expressed in picofarads (pF). The first two digits are the significant figures, and the third digit is the multiplier (number of zeros).
- 104: 10 followed by four zeros = 100,000 pF = 100 nF = 0.1 µF.
- 223: 22 followed by three zeros = 22,000 pF = 22 nF.
- 471: 47 followed by one zero = 470 pF.
Tolerance and Temperature Letters
A letter trailing the numeric code indicates tolerance or temperature coefficient. According to standard component references:
- J: ±5% tolerance
- K: ±10% tolerance
- M: ±20% tolerance
- Z: +80% / -20% (common for bulk electrolytic decoupling where minimum capacitance matters, but maximum is fine)
SMD MLCC Codes
Most surface-mount MLCCs have no markings at all. You must rely on the reel packaging or measure them with an LCR meter. If they are marked, they follow the same 3-digit pF system, but the physical size (e.g., 0402, 0603, 0805) dictates the maximum available capacitance and voltage rating.
Selection Matrix: Which Capacitor Type for Which Job?
Choosing the wrong dielectric is a primary cause of circuit failure. Use this matrix to match the component to the application. For a deeper dive into dielectric behaviors, Analog Devices provides excellent application notes on capacitor selection.
| Type / Dielectric | Construction | Tolerance & Tempco | Typical Application | Bench Gotchas |
|---|---|---|---|---|
| MLCC (C0G/NP0) | Class I Ceramic | ±5%, 0 ±30ppm/°C | RF filters, precision analog timing, PLL loops | Low max capacitance (~10nF in 0805). Physically large for high values. |
| MLCC (X7R/X5R) | Class II Ceramic | ±15% to ±22%, High drift | Digital decoupling, bulk bypass, general I/O | Severe DC bias derating. Piezoelectric "singing" under AC ripple. |
| Aluminum Electrolytic | Liquid electrolyte, etched foil | ±20%, High temp drift | Power supply bulk filtering, high-energy storage | High ESR. Dries out over time. Polarized (explodes if reversed). |
| Polymer / OS-CON | Solid conductive polymer | ±20%, Stable ESR over temp | Switching regulator output, high-ripple current | Lower max voltage ratings (usually <35V). More expensive. |
| Film (Polypropylene) | Metallized plastic film | ±1% to ±5%, Very stable | Audio crossovers, AC line filtering, snubbers | Massive physical footprint. Not suitable for tight SMD layouts. |
Bench Scenario: The 100µF Decoupling Disaster
Theory is clean; the bench is messy. Here is a real-world scenario that illustrates why capacitor selection goes far beyond just matching the microfarad rating.
The Setup: Designing a 5V, 3A buck converter using a TI TPS5430 to drive a high-torque servo motor. The datasheet recommends 100µF of output capacitance to maintain loop stability and handle transient loads.
The Numbers: I grabbed a standard, cheap 100µF 16V radial aluminum electrolytic capacitor from the bench bin. The switching frequency was 500 kHz. The servo drew pulsed currents of 2A peak.
The Outcome: Upon power-up, the output voltage rang like a bell. The oscilloscope showed a massive 1.5V peak-to-peak ripple on the 5V rail. The servo jittered violently, and the TPS5430 IC became too hot to touch within 30 seconds.
What Went Wrong: I ignored Equivalent Series Resistance (ESR) and ripple current ratings. The generic electrolytic cap had an ESR of roughly 0.6Ω. When the servo pulled a 2A transient step, Ohm's law ($V = I \times R$) dictated that the ESR alone would cause a $2A \times 0.6\Omega = 1.2V$ instantaneous voltage drop. Furthermore, the high ESR caused massive internal heating ($I^2R$ losses) from the 500 kHz switching ripple, pushing the cap toward thermal failure.
The Fix: I replaced the generic electrolytic with a 100µF Panasonic OS-CON solid polymer capacitor (ESR ~0.025Ω) and placed a 10µF X7R MLCC in parallel directly at the IC pins to handle the high-frequency switching edges. The ripple dropped to 18mV, and the IC ran cool. The lesson: always check the ESR and ripple current specs for switching power supplies.
Failure Modes: Spotting a Dead Cap Before It Kills Your Board
Capacitors fail in distinct ways depending on their chemistry. Recognizing these visual and electrical symptoms will save you hours of debugging.
1. Aluminum Electrolytic: Venting and Drying
- Visual Symptom: The scored "K" or cross on the top of the can is bulged upward. You may see crusty brown or black residue leaking from the vent or the bottom rubber bung.
- Electrical Symptom: Capacitance drops by 30-50%, and ESR skyrockets (measured with an ESR meter). The circuit experiences brownouts or high ripple.
- Cause: Over-voltage, reverse polarity, or simply old age (the liquid electrolyte evaporates through the rubber seal over 10-15 years).
2. MLCC (Ceramic): Flex Cracking
- Visual Symptom: Often invisible to the naked eye. Sometimes you can see a microscopic hairline crack near the PCB pad under 10x magnification.
- Electrical Symptom: Dead short. The power supply trips, or the board draws massive current and burns the PCB trace.
- Cause: Mechanical stress. If the PCB is bent during assembly, depaneling, or connector insertion, the rigid ceramic cracks. The crack bridges the internal alternating plate layers, creating a short.
3. Tantalum: Thermal Runaway
- Visual Symptom: A charred, blackened lump on the board, often with scorch marks on the surrounding FR4 fiberglass. Sometimes accompanied by literal flames and acrid smoke during the failure event.
- Electrical Symptom: Hard dead short.
- Cause: Tantalum capacitors are highly intolerant of surge currents and voltage spikes. If a transient exceeds their rating, the dielectric breaks down, resistance drops, current increases, and the part catches fire. Always derate tantalum voltage by 50%.
The Substitution Rules: When the Exact Part is Missing
You are repairing a board or building a prototype, and the BOM calls for a specific capacitor you do not have. Substituting capacitors is safe if you follow these four rules:
- Voltage can go up, never down. A 50V capacitor can safely replace a 25V capacitor. However, beware of the Minimum ESR Trap in older Low Dropout Regulators (LDOs). Some legacy LDOs (like the LM1117) require a minimum output ESR (often >0.1Ω) to maintain internal loop stability. If you substitute a high-ESR electrolytic with a near-zero ESR MLCC, the LDO will oscillate at high frequencies.
- Capacitance can go up for bulk filtering, but never for timing. In a power supply filter, swapping 100µF for 220µF is usually fine (check physical clearance and inrush current limits). But in a 555 timer oscillator, an RC filter, or a compensation network, changing the capacitance alters the time constant ($\tau = R \times C$) and will break the circuit's intended frequency response.
- Respect the Temperature Coefficient in analog paths. Never substitute a C0G/NP0 capacitor with an X7R in an active analog filter or audio coupling path. X7R capacitors exhibit a "microphonic effect" (they act as piezoelectric microphones, picking up mechanical vibrations) and their capacitance shifts with applied voltage, which will introduce severe harmonic distortion into audio signals.
- Watch the physical footprint and lead spacing. A 1000µF 35V electrolytic might have a 10mm diameter and 5mm lead pitch, while your PCB pads are drilled for an 8mm diameter and 3.5mm pitch. Forcing the leads will stress the internal connections and cause a premature open-circuit failure.
Understanding how a capacitor works means looking past the printed microfarad value. By evaluating ESR, dielectric class, DC bias derating, and mechanical vulnerabilities, you transition from simply swapping parts to engineering reliable, robust circuits.
