Strictly speaking, electrons do not flow through the dielectric of a capacitor. The dielectric is an insulator. However, in an AC circuit, current appears to flow through the component due to displacement current—the continuous charging and discharging of the plates that creates a changing electric field across the gap. In a DC circuit, current only flows transiently until the capacitor charges to the source voltage, after which it blocks DC entirely (minus a microamp-level leakage current). Understanding this distinction is the difference between designing a stable power supply and watching your prototype vent electrolyte across the workbench.
The Physics: Displacement Current vs. Conduction Current
To visualize why AC 'flows' through a capacitor without electrons crossing the gap, use the water analogy: imagine a pipe blocked by a flexible rubber membrane. Water (electrons) cannot pass through the rubber (dielectric). But if you push and pull the water back and forth rapidly (AC), the membrane flexes, and water moves in and out of the pipe on both sides. An observer looking at the pipe as a whole would say water is 'flowing' through it, even though no single water molecule crossed the membrane.
In electrical terms, this is governed by Maxwell's equations. Conduction current is the physical movement of charge carriers. Displacement current is the rate of change of the electric flux density. When you apply an AC voltage, the continuous $dV/dt$ (change in voltage over time) forces charge to pile up on one plate and deplete from the other, creating a measurable current in the external circuit:
Current (I) equals Capacitance (C) multiplied by the rate of voltage change. If $dV/dt$ is zero (steady DC), current is zero.
The only time actual electrons cross the dielectric is during leakage current (a non-ideal trait of all real capacitors) or during a catastrophic dielectric breakdown when the voltage exceeds the part's rating, turning the insulator into a conductor and usually destroying the component.
Bench Scenario: The Snubber Capacitor Failure
Abstract theory only gets you so far. Let us look at a real-world failure involving a 120V AC relay coil switching an inductive motor load, which perfectly illustrates how displacement current behaves during transients.
The Setup
An automated workshop rig used a 120VAC electromechanical relay to switch a 1/2 HP induction motor. To suppress the inductive voltage spike when the relay opened, the builder placed a 0.1µF, 275VAC X2 safety capacitor directly across the relay contacts.
The Numbers
- Steady State: At 60Hz, the capacitive reactance ($X_c = 1 / (2\pi fC)$) of 0.1µF is roughly 26.5kΩ. The steady-state AC leakage through the cap is about 4.5mA. Harmless.
- The Transient: When the relay contacts open, the motor's collapsing magnetic field creates a high-frequency voltage ring. Let us say the ring frequency hits 1MHz with a 400V peak. At 1MHz, the $X_c$ of that 0.1µF capacitor drops to just 1.59Ω.
The Outcome and What Went Wrong
With an impedance of 1.59Ω and a 400V spike, the instantaneous displacement current surging in and out of the capacitor exceeded 250 Amps. While brief, this massive $dV/dt$ and high peak current caused localized heating in the capacitor's internal metallization. Over three months, the metallization vaporized in microscopic pockets, dropping the capacitance to 0.02µF. The suppressed spikes returned, pitting the relay contacts until they welded shut, destroying the motor controller.
The Fix: A capacitor alone across an inductive switch just shifts the resonant frequency. The correct approach is an RC snubber. Replacing the lone capacitor with a series network of a 0.047µF X2 capacitor and a 100Ω, 2W carbon composition resistor damped the high-frequency ring, limited the peak displacement current, and protected both the capacitor and the relay contacts. For deeper math on snubber design, the Cornell Dubilier technical library provides excellent application notes on dampening inductive spikes.
Capacitor Types: Which Dielectric for Which Job?
Choosing the right capacitor is not just about matching the microfarad value. The dielectric material dictates temperature stability, microphonics, and DC bias behavior. Here is the selection framework for the most common bench components.
| Type / Dielectric | Construction & Traits | Tolerance & Tempco | Typical Use Case | When to Avoid |
|---|---|---|---|---|
| MLCC (C0G/NP0) | Ceramic. Linear, no piezoelectric effect, zero DC bias capacitance loss. | ±5% (J) ±30ppm/°C |
RF filters, precision timing, high-Q resonant circuits, audio signal paths. | High capacitance values (>10nF) in small SMD packages; physically too large/expensive. |
| MLCC (X7R/X5R) | Ceramic. High volumetric efficiency. Exhibits piezoelectric ringing and severe DC bias drop. | ±10% (K) / ±20% (M) ±15% over temp |
Power supply bypass/decoupling, bulk filtering, non-critical coupling. | Precision analog, audio paths (causes distortion), timing oscillators. |
| Aluminum Electrolytic | Wound foil with liquid/solid electrolyte. High ESR, polarized, dries out over time. | ±20% (M) High temp variance |
Bulk energy storage, low-frequency power supply smoothing, audio coupling. | High-frequency switching (use polymer/film instead), high-ripple current without rated life. |
| Film (Polypropylene) | Metallized plastic film. Self-healing, extremely low ESR/ESL, non-polar, physically large. | ±5% to ±10% Very stable |
AC line filtering (X/Y safety), high-current snubbers, motor run, high-end audio. | Space-constrained PCBs, high-density SMD designs. |
The DC Bias Trap: If you place a 16V, 10µF X7R MLCC in a 12V circuit, do not expect 10µF. Due to the ferroelectric nature of the Barium Titanate dielectric, a 12V DC bias can reduce the actual capacitance by up to 70%. Always derate X7R/X5R MLCCs by at least 50% for voltage, or check the manufacturer's DC bias curves.
Decoding the Markings and Safe Substitution
Through-hole and large SMD capacitors use standardized codes. Misreading these is a primary cause of bench failures.
How to Read the 3-Digit Code
A typical film or ceramic capacitor might be stamped with 104K 250V.
- First two digits (10): The significant figures.
- Third digit (4): The multiplier (number of zeros), in picofarads (pF). So, 10 followed by four zeros = 100,000 pF.
- Conversion: 100,000 pF = 100 nF = 0.1 µF.
- Letter (K): Tolerance. J = ±5%, K = ±10%, M = ±20%, Z = +80%/-20%.
- Voltage (250V): Maximum DC or AC working voltage (check datasheet for AC vs DC specific ratings, especially for safety caps).
Rules for Safe Substitution
When the exact BOM part is out of stock, follow these substitution rules to avoid redesigning the board:
- Upgrading Voltage: You can always substitute a higher voltage rating (e.g., using a 50V cap instead of a 25V cap). Watch the physical footprint—higher voltage caps are larger and may not fit the PCB pads. Also, higher voltage electrolytics sometimes have higher ESR.
- Swapping Dielectrics: Never substitute an X7R ceramic for a C0G in a timing circuit, PLL filter, or precision ADC reference. The X7R's capacitance variance with temperature and voltage will detune the circuit. You can substitute C0G for X7R in bulk bypass, though it is overkill and expensive.
- Safety Capacitors (X/Y): Never substitute a standard high-voltage film or ceramic capacitor for an X1/X2 or Y1/Y2 safety-rated capacitor across the AC mains. Safety caps are designed to fail open and are tested against specific surge standards (like IEC 60384-14). Standard caps may fail short, causing a fire. For detailed safety classifications, refer to the Vishay safety capacitor application guide.
Failure Modes: What to Look For on the Bench
Capacitors fail in distinct ways depending on their chemistry and the abuse they endure. Before troubleshooting, always discharge large capacitors with a high-wattage resistor. A dead short across a charged 400V electrolytic with a screwdriver will vaporize the screwdriver tip and damage your eyes.
Visual and Electrical Symptoms by Type
| Capacitor Type | Common Failure Mode | Visual Symptoms | Electrical Measurement (ESR/Cap) |
|---|---|---|---|
| Aluminum Electrolytic | Electrolyte boil-off / Venting | Bulging top vent, crusty brown residue on the leads or PCB, popped pressure relief cross. | Capacitance drops significantly; ESR spikes well above datasheet limits (e.g., >5Ω on a 1000µF cap). |
| MLCC (Ceramic SMD) | Flex Cracking / Short Circuit | Microscopic hairline crack near the terminal edge (requires magnification). Often looks fine to the naked eye. | Dead short (near 0Ω). Caused by PCB bending during depaneling or connector insertion. See TI application note on MLCC flex cracks. |
| Film (Metallized PP) | Metallization Clearing / Open | Slight swelling of the epoxy casing, faint smell of ozone or burnt plastic. Sometimes no visual cue. | Capacitance drops in steps (e.g., from 0.1µF to 0.06µF) as internal series sections clear out from overvoltage transients. |
| Tantalum | Thermal Runaway / Short | Violent failure. Charred black mark on the casing, blown crater, smoke, and a distinct acrid chemical smell. | Dead short. Tantalums fail short and ignite if subjected to reverse polarity or excessive ripple current without adequate series resistance. |
When a circuit behaves erratically—especially switching power supplies that whine or fail to start under load—the electrolytic capacitors on the secondary side are the prime suspects. An ESR meter is mandatory here. A multimeter's capacitance function might read a failing 1000µF capacitor as 950µF (within 20% tolerance), hiding the fact that its ESR has climbed from 0.05Ω to 4.0Ω, rendering it useless for high-frequency ripple filtering.
Understanding that current does not physically cross the dielectric, but rather drives the displacement field, changes how you view circuit design. It forces you to consider $dV/dt$, dielectric absorption, and the physical limits of the materials inside that tiny component on your board. Treat capacitors as complex, frequency-dependent electrochemical devices rather than ideal math variables, and your bench success rate will climb immediately.
