When you search for a capacitor size calculator, you are usually trying to solve one specific problem: keeping a voltage rail from collapsing when the load spikes or the primary supply drops out. Online calculators will happily spit out a number, but if you do not understand the underlying physics, the assumptions baked into the formula, and the parasitic realities of physical components, that number will lead you straight to a brownout on the bench.
This guide strips away the black box. We will derive the core DC discharge formula, track the units through two distinct solved problems, and walk through a real-world ESP32 design failure where the math was perfect, but the physical component choice was wrong.
The Core Capacitor Sizing Formula and Symbol Definitions
The fundamental equation used by almost every basic capacitor size calculator for DC hold-up and constant-current discharge is derived from the definition of capacitance ($C = Q/V$) and current ($I = dQ/dt$). Assuming a constant current draw, the relationship simplifies to a linear voltage drop over time.
$$C = \frac{I \cdot \Delta t}{\Delta V}$$
Every symbol in this equation represents a strict physical parameter. If you misidentify the voltage delta or the time window, the calculator output is useless. Here is the exact definition of each variable as it applies to power supply design.
| Symbol | Parameter | SI Unit | Practical Design Definition |
|---|---|---|---|
| $C$ | Capacitance | Farads (F) | The total energy storage required to bridge the gap. |
| $I$ | Discharge Current | Amperes (A) | The constant load current drawn by the circuit during the hold-up window. |
| $\Delta t$ | Time Delta | Seconds (s) | The duration the capacitor must support the load without primary power. |
| $\Delta V$ | Voltage Drop | Volts (V) | The maximum allowable voltage sag (Initial Voltage minus Minimum Operating Voltage). |
Rearranged Forms and Realistic Magnitudes
A good engineer does not just solve for $C$. Depending on the design constraint, you often need to know how long a given capacitor will last, or what the voltage droop will be. Here are the rearranged forms of the master formula:
- Solve for Current ($I$): $I = \frac{C \cdot \Delta V}{\Delta t}$ (Useful for sizing a load based on an existing backup cap).
- Solve for Time ($\Delta t$): $\Delta t = \frac{C \cdot \Delta V}{I}$ (Useful for calculating RTC or memory backup duration).
- Solve for Voltage Drop ($\Delta V$): $\Delta V = \frac{I \cdot \Delta t}{C}$ (Useful for predicting ripple or droop on a known rail).
What does a realistic answer magnitude look like? If your calculator outputs $1 \times 10^{-12}$ F, you are in the RF/tuning domain (pF). If it outputs $1 \times 10^{-6}$ F, you are looking at standard decoupling (µF). If the output is $0.01$ F to $5.0$ F, you have crossed into supercapacitor territory for long-term hold-up. If your hold-up calculation yields a result in the hundreds of Farads for a standard PCB, you have likely made a unit conversion error or your load current is too high for capacitive backup.
Solved Problems: Tracking Units from Microamps to Farads
The most common way a capacitor size calculator fails the user is through hidden unit conversions. The formula only works if you feed it base SI units: Amperes, Seconds, and Volts. Let us track the units explicitly through two distinct bench scenarios.
Problem 1: LDO Dropout Hold-Up for a Sensor Hub
Scenario: You have a 5V rail feeding a 3.3V LDO. The sensor hub draws a constant 150 mA. When the main 5V supply is cut, you need the LDO to maintain regulation for at least 10 ms to allow the microcontroller to save state to EEPROM. The LDO dropout voltage is 0.2V, meaning the input must stay above 3.5V.
- Identify $\Delta V$: The rail starts at 5.0V and can drop to 3.5V. $\Delta V = 5.0 - 3.5 = 1.5 \text{ V}$.
- Convert $I$ to Amperes: $150 \text{ mA} = 0.15 \text{ A}$.
- Convert $\Delta t$ to Seconds: $10 \text{ ms} = 0.01 \text{ s}$.
- Apply Formula: $$C = \frac{0.15 \text{ A} \cdot 0.01 \text{ s}}{1.5 \text{ V}}$$ $$C = \frac{0.0015 \text{ A}\cdot\text{s}}{1.5 \text{ V}} = 0.001 \text{ F}$$
- Convert to standard units: $0.001 \text{ F} = 1000 \text{ \mu F}$.
Bench Note: You would select a standard 1000 µF or 1500 µF electrolytic capacitor rated for at least 10V (applying the 50% voltage derating rule for electrolytic longevity).
Problem 2: Real-Time Clock (RTC) Supercapacitor Backup
Scenario: An RTC chip draws 2 µA from a 3.3V backup line when main power fails. The RTC stops functioning if the voltage drops below 1.2V. You need the circuit to survive a 30-day power outage.
- Identify $\Delta V$: $3.3 \text{ V} - 1.2 \text{ V} = 2.1 \text{ V}$.
- Convert $I$ to Amperes: $2 \text{ \mu A} = 2 \times 10^{-6} \text{ A}$.
- Convert $\Delta t$ to Seconds: $30 \text{ days} \times 24 \text{ hr} \times 3600 \text{ s} = 2,592,000 \text{ s}$.
- Apply Formula: $$C = \frac{(2 \times 10^{-6} \text{ A}) \cdot (2,592,000 \text{ s})}{2.1 \text{ V}}$$ $$C = \frac{5.184 \text{ A}\cdot\text{s}}{2.1 \text{ V}} \approx 2.468 \text{ F}$$
Bench Note: A standard 2.5F or 3.0F, 5.5V supercapacitor (like a Cornell Dubilier DSK series) is the correct physical choice here, as standard electrolytics max out in the milli-Farad range.
When the Formula Applies (and When It Fails)
The $C = \frac{I \cdot \Delta t}{\Delta V}$ equation is a powerful tool, but it is built on strict assumptions. According to fundamental circuit theory outlined by resources like All About Circuits, this formula assumes an ideal capacitor discharging at a constant current.
Assumptions and Limits
- Constant Current Load: If your load is resistive (like a heating element or a simple LED string), the current drops as the voltage drops ($I = V/R$). The voltage decay will be an exponential curve ($V(t) = V_0 e^{-t/RC}$), not a linear slope. Using the linear formula for a purely resistive load will underestimate the required capacitance.
- Zero ESR and ESL: The formula assumes the capacitor has no Equivalent Series Resistance (ESR) or Inductance (ESL). In reality, ESR causes an instantaneous voltage drop the millisecond the load is applied.
- No Leakage Current: Supercapacitors and large electrolytics have significant internal leakage. If $\Delta t$ is measured in days or weeks, you must subtract the capacitor's leakage current from your available load budget.
Unit Mistakes That Break the Math
The most frequent error when using a manual capacitor size calculator is mixing prefixes. If you input $I$ in milliamps (mA) and $C$ comes out in Farads, your answer is off by a factor of 1,000. Always convert to base SI units (Amps, Volts, Seconds) before multiplying, then convert the final Farad value to microfarads (µF) or millifarads (mF) for component sourcing.
Real-World Scenario: The ESP32-CAM Brownout Failure
Let us look at a scenario where the math was executed perfectly, but the physical implementation failed. This highlights the gap between theoretical calculator outputs and bench realities.
The Setup
An engineer was designing a 5V to 3.3V buck converter to power an ESP32-CAM module. The module sits in deep sleep but wakes up every 10 seconds to transmit a burst of WiFi data. During the TX burst, the current draw spikes to 800 mA for roughly 2 milliseconds. The buck converter's transient response was too slow to handle the 2ms spike, causing the 3.3V rail to dip below the ESP32's 2.7V brownout threshold.
The Numbers
The engineer used the capacitor size calculator to find the bulk capacitance needed to handle the 2ms transient while the buck controller spooled up.
- $I = 0.8 \text{ A}$
- $\Delta t = 0.002 \text{ s}$
- Allowable droop $\Delta V = 0.4 \text{ V}$ (keeping the rail above 2.9V for safety margin).
$$C = \frac{0.8 \cdot 0.002}{0.4} = 0.004 \text{ F} = 4000 \text{ \mu F}$$The engineer placed a single, high-quality 4700 µF, 10V aluminum electrolytic capacitor on the 3.3V output rail. The math provided a comfortable 17% margin.
The Outcome and What Went Wrong
On the bench, the ESP32-CAM still brownout-reset during every WiFi transmission. The engineer hooked up an oscilloscope and observed a massive 0.8V instantaneous drop at the exact microsecond the TX pin went high, followed by a slow linear decay.
The calculator formula only accounts for the capacitive discharge ($\Delta V$ over time). It completely ignores the resistive instantaneous drop governed by Ohm's Law: $V_{drop} = I_{spike} \cdot ESR$.
The 4700 µF electrolytic capacitor had an ESR of roughly 40 mΩ (0.04 Ω). When the 800 mA (0.8 A) spike hit, the instantaneous voltage drop was:
$$V_{ESR} = 0.8 \text{ A} \cdot 0.04 \text{ \Omega} = 0.032 \text{ V}$$
Wait, 32 mV is negligible. Why did the scope show an 800 mV drop? Because at the high-frequency edge of a digital TX burst (where $di/dt$ is massive), the capacitor's Equivalent Series Inductance (ESL) and the PCB trace inductance dominate. The electrolytic capacitor, with its internal foil roll, acts like an inductor at high frequencies, effectively becoming an open circuit to the fast transient spike.
The Fix
The engineer kept the 4700 µF electrolytic to handle the bulk energy storage (the $\Delta t$ requirement calculated by the formula) but added an array of four 22 µF, 10V X7R MLCC (Multi-Layer Ceramic Capacitors) in parallel directly under the ESP32 VCC pins. MLCCs have near-zero ESR and virtually no ESL. The ceramics supplied the instantaneous high-frequency current spike, while the electrolytic supplied the sustained 2ms hold-up energy. The brownouts vanished.
The Takeaway: A capacitor size calculator gives you the minimum bulk energy storage required. It does not size your high-frequency decoupling network. Always pair large calculated bulk capacitors with physically small, low-ESL ceramic capacitors placed as close to the load IC as possible. For deeper insights into capacitor parasitics and impedance curves, refer to application notes from major manufacturers like SparkFun's capacitor guides or manufacturer datasheets before finalizing your BOM.
