First-Order Step Response: Rise Time, Settling Time

Why rise time and settling time matter in control system design

Rise time and settling time tell you how fast a control loop reacts and how long it takes to stop moving, two numbers that decide whether your embedded controller meets a real-time deadline or not. A motor controller that needs to reach target speed within 50 ms, or a temperature loop that must stabilize within 2 °C before a process step begins, lives or dies by these two timing specifications. Without closed-form expressions for them, you're stuck running simulations every time a component value changes — slow, error-prone, and hard to reason about when you're picking a sensor bandwidth or a filter cutoff at 2 a.m. the night before a design review.

Speed of response vs. stability trade-off

A faster rise time generally means a system that is more aggressive, more sensitive to noise, and — in anything beyond first order — closer to instability. First-order systems are the forgiving case: they cannot overshoot and cannot oscillate, so the only trade-off is speed versus noise sensitivity, not speed versus stability. That simplicity is exactly why first-order models are the starting point for every control systems course and every back-of-the-envelope embedded design estimate. Once you add poles — an actuator lag, a sensor filter, a second thermal mass — you reintroduce the overshoot-and-ringing trade-offs covered in Performance Specs: %OS, Ts, Tr, Tp, Ess (Steady-State Error).

Real-world examples: motor startup, temperature regulation, sensor filtering

A DC motor's speed loop, an RC thermal sensor, and a single-pole analog anti-aliasing filter are all well-approximated as first-order systems over their useful frequency range. A brushed DC motor driven by a current-limited supply ramps up to speed with a time constant set by its electrical and mechanical time constants combined; an NTC thermistor embedded in a potted enclosure tracks ambient temperature with a thermal time constant that can run from seconds to tens of minutes depending on thermal mass; an RC low-pass filter on an ADC input settles to a new input voltage with a time constant of R×C. In each case, the same two formulas — one for rise time, one for settling time — tell you how long to wait before trusting a reading or declaring a move complete.

How a first-order system responds to a step input

A first-order system's step response is a single decaying exponential that climbs monotonically from its initial value to its final value, with no overshoot. That shape comes directly from the system having exactly one energy-storage element — one capacitor, one thermal mass, one rotating inertia with dominant viscous damping — and therefore exactly one pole in its transfer function.

Standard first-order transfer function form

The canonical form is G(s) = K / (τs + 1), where K is the DC gain (output per unit input at steady state) and τ is the time constant in seconds. Both parameters are defined and derived from physical RC, RL, and thermal circuits in First-Order System: Time Constant τ, DC Gain K. For a unit step input U(s) = 1/s, the output in the Laplace domain is Y(s) = K / [s(τs + 1)].

Time-domain solution and exponential approach

Taking the inverse Laplace transform gives y(t) = K(1 − e^(−t/τ)) for t ≥ 0. This is the single most useful equation in introductory control systems: it says the output starts at zero, climbs steeply at first, and asymptotically approaches the final value K without ever overshooting it. You can confirm the final value independently using the Final Value Theorem: lim(t→∞) y(t) = lim(s→0) sY(s), which gives lim(s→0) s · K/[s(τs+1)] = K — matching the time-domain result and providing a useful cross-check when you've derived Y(s) from a more complex circuit or block diagram.

Role of time constant τ in shaping the response curve

τ is the time at which the response reaches 63.2% of its final value, and it scales the entire curve: double τ and every time-to-reach-a-given-percentage doubles with it. Physically, τ = RC for an RC circuit, τ = L/R for an RL circuit, and τ = J/b for a rotational system with inertia J and viscous damping coefficient b. Everything that follows — rise time, settling time — is just a fixed multiple of this one number.

Defining and calculating rise time for first-order systems

Rise time Tr for a first-order system is the time required for the output to go from 10% to 90% of its final value, and it equals approximately 2.2τ. There's no overshoot to worry about, so unlike second-order systems, 10%–90% rise time is the only rise-time definition you need — there's no ambiguity about which percentage points to use.

10%–90% rise time definition and formula

Starting from y(t) = K(1 − e^(−t/τ)), solve for the time at which y(t)/K = 0.1 and y(t)/K = 0.9. The first gives t₁ = −τ·ln(0.9) ≈ 0.105τ, and the second gives t₂ = −τ·ln(0.1) ≈ 2.303τ. Rise time is the difference:

Tr = t2 - t1 = τ[ln(0.1⁻¹) - ln(0.9⁻¹)] = τ ln(9) ≈ 2.197 τ

So for a time constant of τ = 50 ms, rise time is roughly 2.2 × 50 ms = 110 ms.

Relationship between rise time and time constant

Rise time is always about 2.2 times the time constant for a first-order system, a ratio that holds regardless of K or the input amplitude, because rise time comes from the shape of the normalized exponential curve, not its scale. This proportionality is why specifying τ and specifying Tr are functionally interchangeable for a first-order system — pick whichever is easier to measure in your setup.

Physical interpretation: how fast does output climb?

Rise time answers the question "once the input changes, how long before the output is clearly moving toward and almost at its new value?" It's the number a motion-control engineer quotes when asked "how fast does the axis get up to speed," and the number a thermal engineer quotes when asked "how long until the sensor is reading real temperature instead of its own self-heating transient."

Defining and calculating settling time for first-order systems

Settling time Ts is the time for the output to enter and stay within a specified tolerance band — commonly 2% or 5% — around its final value, and for a first-order system it equals −τ·ln(tolerance). For the 2% band, Ts ≈ 4τ; for the 5% band, Ts ≈ 3τ.

2% and 5% band definitions and formulas

Setting 1 − e^(−t/τ) = 1 − 0.02 and solving for t gives t = −τ·ln(0.02) ≈ 3.912τ, rounded in practice to 4τ. The 5% band gives t = −τ·ln(0.05) ≈ 2.996τ, rounded to 3τ. These two numbers — 3τ and 4τ — are worth memorizing outright; they turn a single time-constant measurement into an instant settling-time estimate without touching a calculator.

How settling time depends on time constant and tolerance

Tighter tolerance bands demand proportionally longer settling times because the exponential tail decays slowly — getting from 95% to 98% of the way there costs nearly as much time as getting from 63% to 95% did. This is why an embedded spec that asks for "settling within 1%" rather than "within 5%" is demanding roughly 1.5× longer wait time for the same physical system, not a marginally tighter number.

Asymptotic behavior: why first-order systems never truly settle

A first-order exponential response never mathematically reaches its final value — e^(−t/τ) only approaches zero as t → ∞, it never equals zero at any finite time. Settling time is therefore always a practical, tolerance-based definition, not an exact arrival time; this is also why you must always quote the tolerance band (2%, 5%, or otherwise) alongside any settling-time number, since "settling time" alone is meaningless.

How to extract rise time and settling time from real measurements

Measuring Tr and Ts from real hardware means applying a clean step input, logging the output at high enough sample rate to resolve the fast part of the transient, and comparing against the measured steady-state value rather than a theoretical one.

Step test procedure and data collection

Apply a step change to the input — a commanded setpoint change for a motor or thermal loop, a switched voltage for an RC filter — and record the output at a sample rate at least 10–20× faster than the expected 1/τ bandwidth. On an oscilloscope, trigger on the step edge, set the timebase so the full settling transient (at least 5τ) fits on screen, and use the scope's built-in rise-time or cursor measurement if available; otherwise, place manual cursors at the 10% and 90% amplitude levels and read the time delta directly.

Identifying steady-state value and tolerance bands

Determine the steady-state value by averaging the output over a window well after the expected settling time, not by assuming it matches a commanded setpoint — actuator nonlinearity, steady-state error, or sensor offset can all make the real final value differ from the target. Once you have that steady-state value, compute the 2% or 5% tolerance band around it and find the last time the trace crosses outside that band; that crossing point is your measured Ts.

Extracting τ from experimental step response

The fastest way to get τ from a trace is to read the time at which the output reaches 63.2% of its total step change, since that time equals τ by definition. A more robust method, especially with noisy data, is to compute ln(1 − y(t)/K) and fit a straight line to it — the slope of that line is −1/τ, and fitting over many samples averages out noise far better than reading a single point off a scope screen.

Using rise time and settling time to design or tune a system

Choosing τ to hit a speed requirement is a matter of inverting the rise-time or settling-time formula: if a specification calls for Ts ≤ 100 ms at 2%, you need τ ≤ 100 ms / 3.91 ≈ 25.6 ms, which then tells you what RC, L/R, or control-loop bandwidth to design for.

Selecting time constant to meet speed requirements

Work backward from whichever spec is binding — rise time for "how fast does it start moving" requirements, settling time for "how long before I can trust the reading" requirements — and solve for the maximum allowable τ. If both are specified, take the smaller resulting τ, since that's the more stringent constraint.

Trade-offs between fast response and noise sensitivity

Shrinking τ to satisfy a tight settling-time spec always widens the system's bandwidth, and wider bandwidth passes more high-frequency noise straight through to the output. An RC input filter with τ = 1 ms responds to an ADC sample fast but also lets through more switching noise than one with τ = 10 ms; there is no way to get the speed without accepting the noise penalty, short of adding more sophisticated filtering (which typically adds more poles and moves you out of first-order territory entirely).

Impact of actuator and sensor delays on observed metrics

Real systems include actuator and sensor dynamics that add extra lag on top of the dominant time constant, making measured Tr and Ts worse than the single-pole formulas predict. A motor's current loop, PWM switching delay, or an analog sensor's own filtering all stack additional small time constants or pure transport delay onto the dominant τ; if measured settling time consistently runs 20–30% longer than 4τ predicts, suspect one of these secondary lags rather than assuming your first-order model itself is wrong.