Intuition The one core idea
A rocket steers by tilting its engine, but the machine that tilts it is not instant — it lags and it hesitates. This whole topic is about measuring that lag with two numbers (how fast the tilter can move, and how long it waits before starting) and checking those numbers don't quietly push your rocket into a wobble.
Before you can read the parent note, you need the vocabulary. Below is every symbol and idea it uses, built from nothing, each one leaning on the one before it. Read top to bottom.
Picture the bottom of a rocket. The engine nozzle can pivot on a hinge (a gimbal ), like a garden hose you swivel by hand.
Definition Deflection angle
δ (delta)
δ is the actual angle the nozzle is tilted away from straight-down, measured in radians or degrees. Straight-down (no steering) is δ = 0 .
Definition Commanded angle
δ c (delta-commanded)
δ c is the angle the flight computer asks for . The little "c" subscript just means "commanded" — it is a request , not yet a reality.
The entire topic lives in the gap between the orange "asked" line and the blue "delivered" line in that figure. When they match, steering is perfect. When they lag, you lose control margin.
Intuition Why we need two symbols
If commanded and actual were always identical we'd need only one symbol. Reality forces two, because a physical machine can never instantly become the number a computer typed.
Why bother tilting the nozzle at all? To turn the rocket .
T
T is the pushing force of the engine, in newtons (N ). Normally it points straight down the rocket's spine, pushing it up.
M (also called moment)
A twisting effort that rotates the rocket about its balance point (centre of mass). Force alone pushes; force applied off to the side twists. Torque is measured in newton-metres (N⋅m ).
When the nozzle tilts by δ , a sideways slice of the thrust appears, and because that sideways push acts at the far end of the rocket (a long lever arm), it produces a turning torque. This is exactly the geometry built in Thrust Vector Control geometry & torque .
Intuition The lever picture
Push the bottom of a broom sideways and the whole broom pivots. The nozzle is the bottom of the broom; T sin δ is the sideways push; the distance to the centre of mass is the lever.
Definition Servo (actuator)
The motor and mechanism that physically rotates the nozzle to try to reach δ c . Think of it as an arm with a limited speed and a reaction time.
The servo has to fight against three things while it moves — and each one is a symbol you must know.
J
How hard the nozzle is to get moving/stopping (its rotational heaviness), in kg⋅m 2 . Big heavy nozzle = big J = sluggish.
c
A resistance proportional to speed — like moving your hand through water. It fights δ ˙ (how fast the angle is changing) and is measured in N⋅m⋅s .
k and gain K
k is a spring-like pull back toward centre (structure resists bending); K is the strength of the motor's correction — how hard it pushes to fix the error δ c − δ . Together k + K act like one combined spring.
Intuition The mass–spring–damper trio
Every physical thing that moves and settles is heaviness (J ) + a spring (k + K ) + a drag (c ). Memorise this trio — it is the whole reason the servo is a second-order system , the topic of Second-order systems — natural frequency & damping .
Definition Newton's dot notation
One dot = rate of change per second . δ ˙ ("delta-dot") is how fast the angle is turning (angular velocity). Two dots δ ¨ ("delta-double-dot") is how fast that is changing (angular acceleration).
Intuition Why dots and not new letters
The dot is just shorthand for a derivative — the slope of the angle-vs-time graph. We need them because Newton's law is about acceleration (δ ¨ ), damping fights velocity (δ ˙ ), and stiffness fights position (δ ). Three levels, three notations.
The parent note's governing equation
J δ ¨ + c δ ˙ + ( k + K ) δ = K δ c
now reads in plain English: (heaviness × acceleration) + (drag × velocity) + (spring × position) = (motor strength × requested angle). Every symbol is now earned.
Definition The Laplace variable
s
A trick variable that ==replaces "take a derivative" with "multiply by s "==. Differentiation is hard; multiplication is easy. So we swap the whole differential equation into the "s -world" where it becomes ordinary algebra.
Rule of thumb inside the s -world (with everything starting at rest):
δ ˙ → s δ ( s ) , δ ¨ → s 2 δ ( s )
Intuition Why physicists love
s
In the time-world you'd solve a differential equation for every different command. In the s -world you get ONE ratio — output over input — that describes the servo forever. That ratio is called a transfer function.
Definition Transfer function
G ( s )
The ratio of what comes out to what goes in , written as a function of s :
G ( s ) = δ c ( s ) δ ( s )
Feed it a command, it tells you the response. This is the language of the whole loop in Rigid-body attitude control loop (autopilot) .
Any mass–spring–damper is fully described by just two numbers.
Definition Natural frequency
ω n (omega-n)
How fast the servo naturally oscillates if you poke it, in radians per second. Stiffer spring or lighter nozzle → higher ω n → faster servo. Formula: ω n = ( k + K ) / J .
ζ (zeta)
A pure number (no units) describing how the wobble dies out. ζ < 1 = bouncy/ringing; ζ = 1 = the fastest settle with no bounce; ζ > 1 = slow and sluggish.
Intuition Radians per second, and why
ω (omega) is an angular frequency: full circle = 2 π radians, so ω = 2 π f where f is cycles per second. We use radians because the maths of oscillation (sin , cos ) is cleanest there. A "rad/s" is just "how many radians of a sine wave sweep by each second."
Definition The imaginary unit
j
j = − 1 . Engineers write j instead of i (to avoid clashing with electric current). Setting s = j ω asks: "how does the servo respond to a pure sine wave wiggling at frequency ω ?"
∣ G ( j ω ) ∣
The size of the output wiggle divided by the input wiggle at frequency ω . If it's 1 , the servo tracks perfectly. If it's 0.5 , the output is half-sized — the servo is falling behind.
ω B
The frequency where the magnitude drops to 1/ 2 ≈ 0.707 of its slow-speed value — the fastest wiggle the servo can still faithfully follow . Ask for faster than ω B and the nozzle just can't keep up.
1/ 2 and not, say, half
At 1/ 2 of amplitude the power has dropped to exactly half. That's a natural, universal cutoff engineers agreed on (called the "− 3 dB point"). It's a convention, but a consistent one across all of Bode plots & phase margin .
∠ G
How much later (in the wiggle cycle) the output peaks compared to the input , measured as an angle. A full cycle is 360° ; being a quarter-cycle late is − 90° of phase.
τ (tau)
A pure wait , in seconds. The servo hears the command, then does nothing for τ seconds (sampling, computing, valve travel), then responds. See Digital control — sampling & computational delay .
Definition The delay operator
e − s τ
The exact s -world stamp for "shift everything later by τ seconds." Its size is always 1 (it changes nothing about amplitude) but its phase is − ω τ : faster wiggles suffer proportionally more lateness .
Intuition Why delay is sneaky
Because ∣ e − j ω τ ∣ = 1 , a delay never shrinks your signal — you can't see it on a size plot. It only steals phase, quietly, and phase is exactly what keeps a control loop stable (Bode plots & phase margin ). That's why the parent note calls it "silent."
The spare phase you have before disaster . A loop goes unstable when total phase lag hits − 180° at the crossover frequency ω c . However many degrees you sit above that danger line is your margin. The delay eats into it: Δ ϕ = − ω c τ .
Definition Padé approximation
A way to ==replace the awkward e − s τ with a simple fraction== so ordinary root-locus tools work. Details in Padé approximation of transport delay ; here just know it exists because e − s τ is "transcendental" (not a plain polynomial).
Definition Structural bending mode
A rocket is a long tube; it can flex and vibrate like a plucked ruler . Each vibration has its own frequency. If the servo's bandwidth reaches those frequencies, the nozzle starts feeding the wobble. The cure is a notch filter .
Intuition The ceiling on bandwidth
Section 6 said "higher ω n = faster = good." Section 9 adds the catch: too high and you excite the flexing tube. So the servo must be fast — but capped below the first bending frequency.
Nozzle tilt angles delta and delta-c
Inertia J damping c stiffness k plus K
Dot notation rates of change
Laplace s and transfer function
omega-n and zeta the two numbers
j-omega magnitude and bandwidth
Bending modes cap the bandwidth
Cover the right side and answer aloud; reveal to check.
What does δ mean and what is δ c ? δ = the nozzle's actual tilt angle; δ c = the angle the computer commands . The whole topic is their gap.
Why is the servo a second-order system? It's a mass–spring–damper: inertia J , stiffness k + K , and damping c — three physical resistances give a s 2 equation.
What does the dot mean in δ ˙ and δ ¨ ? One dot = rate of change (angular velocity); two dots = acceleration. Shorthand for derivatives.
What is s and why use it? The Laplace variable; it turns "take a derivative" into "multiply by s ," converting a differential equation into simple algebra.
In plain words, what is a transfer function G ( s ) ? The ratio of output to input in the s -world — one formula that describes the servo's response to any command.
What does ω n tell you physically? How fast the servo naturally oscillates — its speed.
ω n = ( k + K ) / J .
What does ζ tell you, and what value is the sweet spot? How the wobble settles;
ζ ≈ 0.707 (i.e.
1/ 2 ) is the balance of fast and non-ringing.
Why do we set s = j ω ? To test the servo with a pure sine wave of frequency ω and read off its magnitude and phase.
Define bandwidth ω B in one sentence. The frequency where the response magnitude falls to
1/ 2 of its slow value — the fastest wiggle the servo can still follow.
Why is a pure delay e − s τ dangerous but invisible? Its magnitude is always 1 (no size change) but it adds phase − ω τ , quietly eating the phase margin that keeps the loop stable.
Why can't servo bandwidth be pushed arbitrarily high? Too high and it excites the rocket's structural bending modes, turning TVC into an oscillator; it must stay below the first bending frequency.