Capacitor Charging Questions Engineers Actually Ask
Every electronics engineer has been there: you're staring at a charging curve on an oscilloscope, your RC network isn't behaving the way you expected, and someone on the team casually drops "just wait five tau" like that explains everything. It doesn't — not until you actually understand what's happening. These are the questions I get asked most often, answered without the textbook padding.
What exactly is a time constant, and why does it matter?
The time constant τ (tau) for a resistor-capacitor circuit is simply R × C — resistance in ohms multiplied by capacitance in farads. The result is in seconds. That's the entire formula. What it tells you is how quickly the capacitor charges through the resistor.
Here's the physical intuition: a larger resistance slows the current flowing into the capacitor, so it fills up more slowly. A larger capacitor has more "room" to fill, so it also takes longer. Multiply them together and you get a number that captures both effects in a single value.
At exactly one time constant after you apply voltage, the capacitor has charged to about 63.2% of the supply voltage. Not 50%, not 100% — 63.2%. That specific percentage comes from the math of exponential charging: the voltage follows V(t) = Vsupply × (1 − e−t/τ), and plugging in t = τ gives you 1 − 1/e ≈ 0.632.
Why do people say "five tau" means fully charged? That seems arbitrary.
It's not arbitrary — it's a practical engineering cutoff based on how the math behaves.
At 5τ, the capacitor sits at 99.33% of supply voltage. The remaining 0.67% is so small that for almost every real-world application, you can treat the cap as fully charged. The charging curve is asymptotic — it approaches the supply voltage but technically never reaches it. So "fully charged" is always an approximation; the question is just how precise you need to be.
Here's the breakdown across multiples of τ:
- 1τ: 63.2% charged
- 2τ: 86.5% charged
- 3τ: 95.0% charged
- 4τ: 98.2% charged
- 5τ: 99.3% charged
Notice how the gains shrink rapidly. Going from 3τ to 4τ buys you 3.2 percentage points. Going from 4τ to 5τ buys you only 1.1 points. Beyond 5τ the returns are truly diminishing — you'd be waiting significant extra time for fractions of a percent. That's why the industry settled on 5τ as the "good enough" threshold, and it held up through decades of real circuit design.
If you're building something safety-critical or precision analog, you might wait 7τ (99.91%). If you're doing a quick timing pulse in a microcontroller reset circuit, 3τ might be plenty. The 5τ rule is a sensible default, not a physical law.
I need a delay circuit that holds a signal low for about 200ms before going high. How do I size the capacitor and resistor?
This is one of the most common practical applications of RC timing, and the sizing process is straightforward once you define your trigger threshold.
The typical approach: you're feeding the RC network's output into a comparator or a logic gate with a defined input threshold. Let's say you're using a CMOS gate that switches high when its input crosses 50% of VCC. You need to find when the capacitor voltage hits that 50% mark.
From the charging equation:
t = −τ × ln(1 − Vthreshold/Vsupply)
For a 50% threshold: t = −τ × ln(0.5) = τ × 0.693
So the time to reach 50% is roughly 0.693τ. You want that to equal 200ms:
0.693 × R × C = 0.200 seconds
R × C = 0.289 seconds
Now you have freedom to pick one value and solve for the other. Common practice is to constrain R based on drive current availability and leakage concerns. If you pick R = 100kΩ:
C = 0.289 / 100,000 = 2.89μF
Reach for the nearest standard value — a 3.3μF cap would give you about 229ms delay, close enough for most applications. If precision matters, use a 2.7μF cap in series with a small trimmer resistor to tune the actual delay in-circuit.
One practical note: resistors above about 1MΩ become problematic in delay circuits because input leakage current on comparator or gate inputs starts distorting the waveform. Keeping R under 470kΩ is usually wise unless you know your comparator has sub-nA input bias current.
Why doesn't my RC circuit behave the way the formula predicts?
Several culprits, in rough order of how often I've seen them in practice:
Capacitor ESR and ESL. Electrolytic capacitors in particular have significant equivalent series resistance. At the start of charging, when current is highest, the ESR creates a voltage drop that makes the cap appear to charge faster than it should. This is especially visible in very low-resistance circuits.
Source impedance. If you're treating your supply as ideal (zero ohms) but it's actually a signal from a microcontroller I/O pin, that pin has maybe 50–200Ω of output impedance. Add that to your external R before you calculate τ.
Capacitor tolerance. Electrolytic caps are commonly rated ±20%. Your 10μF cap could legally be 8μF or 12μF straight from the factory. This alone can shift your delay by 20% in either direction. If you need tight timing, use a film capacitor or NP0/C0G ceramic — those hold much better tolerances (1–5%).
Load on the RC output. If you're reading the capacitor voltage through a circuit that draws current — even a few microamps — that load acts like a parallel resistance and changes your effective RC. A high-impedance buffer (op-amp voltage follower) between the cap and whatever it's driving solves this cleanly.
Temperature effects. Electrolytic capacitance drifts noticeably with temperature, especially below 0°C. If your product operates outdoors or in a vehicle, factor this in or switch to a different capacitor chemistry.
Can I use an online RC time constant calculator, or do I need to do this by hand?
Calculators are genuinely useful for getting a first answer quickly — especially for the solve-for-C or solve-for-R direction when you're picking standard component values. A good RC calculator will let you plug in your desired delay and threshold percentage, then spit out the τ you need and suggest standard R/C pairs.
That said, calculators only solve the idealized model. They won't account for your source impedance, capacitor tolerance, comparator hysteresis, or the ESR of that electrolytic you pulled from your parts bin. So the workflow that actually works is: use a calculator to get your target component values, then build it, probe it with a scope, and tweak from there. The formula gives you the ballpark; the scope gives you the truth.
How does the discharge calculation differ from charging?
When a charged capacitor discharges through a resistor toward 0V, the voltage follows: V(t) = Vinitial × e−t/τ
The time constant τ is still R × C, and the same multiples apply — at 5τ, the capacitor is at 0.67% of its initial voltage, effectively discharged. The shape is a mirror exponential falling from full voltage rather than rising toward it.
What changes in practice: the resistance value might be different. In many circuits, the charging path and discharging path use separate resistors — for instance, a diode steers the cap to charge through R1 and discharge through R2. A classic example is an asymmetric astable oscillator where the on-time and off-time have different durations. Knowing which resistor is in the path at each phase is the key to predicting both times correctly.
One last thing: what's the biggest mistake engineers make with RC timing circuits?
Assuming the capacitor starts at 0V. If your circuit can be powered off and back on rapidly, or if some previous state left the cap partially charged, your delay will be shorter than expected. A bleed resistor (usually 10× your timing resistor value) across the capacitor ensures it fully discharges when power goes away. It's a minor component cost that prevents an infuriating intermittent bug that only shows up in field conditions.
The RC time constant is one of those foundational concepts that seems trivially simple until you have to rely on it in a real product. Getting the numbers right, understanding the tolerances, and accounting for real-world parasitics is what separates a circuit that works on the bench from one that works reliably in production.