Thrust vector control — single-gimbal, dual-gimbal; TVC angles: Visual walkthrough

Step 0 — Fix a coordinate system and a sign convention first

WHAT. Before any picture of forces, we nail down which way is "positive," so no arrow or sign is ever ambiguous later.

WHY. A torque of " $+ 5$ " is meaningless until we say what $+$ means. Every sign in this page (the deflection $δ$ , the torque $M$ ) is measured against the frame we fix right here, once and for all.

Definition Our 2D frame and sign rules (used everywhere below)

$+ x$ points backward, out the tail (the direction the exhaust leaves). So the CoM is at $x = 0$ and the gimbal is at $x = + ℓ$ ... but by long-standing convention we place the CoM at the origin and write the gimbal position as $r = (- ℓ, 0)$ measured from CoM to gimbal along the nose direction. To keep one consistent picture, we take $+ x$ pointing toward the nose (forward), so the gimbal, being behind the CoM, sits at $x = - ℓ$ .
$+ y$ points up (in the plane of the page).
$+ δ$ = the engine tilted so its sideways thrust points in $+ y$ (deflects the nozzle "down-page"), which we will see produces a nose-up rotation. $- δ$ is the mirror image → nose-down.
$+ M$ = counter-clockwise spin in the page (the standard math convention, right-hand rule out of the page).

Keep these four rules in mind; every later sign is just bookkeeping against them.

Step 1 — Draw the rocket and name every distance

WHAT. We start with the barest possible picture: a rocket, a dot for its balance point, and a dot for where the engine is bolted on.

WHY. Before any formula, we must agree on what pushes and where it pushes from. Every symbol that appears later must first be a thing you can point at in this picture.

PICTURE. Look at the figure. Two special points:

The balance point of the rocket — the single spot where it would sit level if you tried to balance it on a fingertip. This balance point is the center of mass (CoM); from here on we use only the name "CoM," and it is the pivot about which the whole rocket rotates.
The gimbal point — the mechanical hinge where the engine is attached, sitting a distance $ℓ$ behind the CoM. In our frame (Step 0) that puts it at $x = - ℓ$ .

Figure — Thrust vector control — single-gimbal, dual-gimbal; TVC angles

At this moment the engine points straight back, so the push goes straight forward through the CoM. A push aimed straight at the pivot cannot spin anything — think of pushing a door exactly on its hinge. So right now: zero steering. Good. That is our baseline.

Step 2 — Tilt the engine and give the tilt a name

WHAT. Rotate the thrust arrow by a small angle away from the rocket's axis.

WHY. A push aimed straight at the pivot does nothing (Step 1). To turn, we must aim the push slightly to the side so it has a "sideways grab." The amount of tilt is the one knob the autopilot turns, so it deserves a symbol.

PICTURE. In the figure the thrust arrow (red) is swung off the dashed axis line. The wedge between the dashed axis and the red arrow is the tilt.

Step 3 — Split the tilted push into two honest pieces

WHAT. Break the single slanted red arrow into two arrows at right angles: one along the axis, one across it.

WHY. A slanted force is hard to reason about directly. The trick used everywhere in physics: replace one awkward arrow with two easy arrows that add back to it. We pick the two directions that mean something physically — "forward" (does useful pushing) and "sideways" (does the steering).

Why trig, and why cosine/sine specifically? The red arrow, the forward piece, and the sideways piece form a right triangle whose longest side (hypotenuse) has length $T$ . Trigonometry is exactly the tool that converts an angle plus a hypotenuse into the lengths of the two legs. That is the question we are asking, so that is the tool we use.

PICTURE. The figure shows the right triangle. The angle $δ$ sits at the gimbal.

The leg next to the angle (adjacent) is the forward piece. $cos δ = \frac{adjacent}{hypotenuse} = \frac{forward}{T}$ , so forward $= T cos δ$ .
The leg across from the angle (opposite) is the sideways piece. $sin δ = \frac{opposite}{hypotenuse} = \frac{sideways}{T}$ , so sideways $= T sin δ$ .

Step 4 — Turn the sideways push into a twist

WHAT. Ask what that sideways force does to the whole rocket, given it acts a distance $ℓ$ from the CoM.

WHY. A sideways force applied away from the pivot doesn't just slide the rocket — it rotates it. The strength of that rotation-tendency is called torque, and torque is what changes the rocket's heading. We want the heading to change, so torque is the quantity we are hunting. (Foundations: Torque and Moment Arm.)

Why the "force × distance" rule? Push a door near the hinge and little happens; push at the far edge and it swings easily — same force, more turning. Torque captures exactly this: it is the sideways force multiplied by how far out it acts, the moment arm $ℓ$ (the same black bar labelled in the Step 1 and Step 4 figures).

PICTURE. The figure shows only the sideways component $T sin δ$ acting at the gimbal, a distance $ℓ$ from the CoM (the labelled moment arm), curling the rocket around.

$∣ M ∣ = ℓ T sin δ$

This is the parent's exact result (a magnitude), now assembled brick by brick.

Step 5 — Restore the sign: the same twist via the cross product

WHAT. Redo Step 4 with the formal cross-product bookkeeping, so the direction (sign) of the torque is honest and matches the Step 0 convention.

WHY. Step 4 gave the size $∣ M ∣ = ℓ T sin δ$ . But a twist also has a direction — clockwise or counter-clockwise. The cross product $r \times F$ is the tool that returns both size and spin direction from the position and the force. We need direction so the autopilot knows which way the nose will swing. (Machinery: Rigid Body Rotational Dynamics.)

PICTURE. The figure places the position vector $r = (- ℓ, 0)$ (from CoM back to gimbal, matching Step 0: gimbal at $x = - ℓ$ ) and the force $F$ from Step 3, and shows the curl direction the cross product predicts.

Step 6 — Small angles: why steering is almost free

WHAT. Zoom in on tiny $δ$ and compare how fast the turning grows versus how fast the forward thrust is lost.

WHY. Real gimbals move only a few degrees. In that tiny window, curves look like straight lines, and we can replace $sin$ and $cos$ with simpler expressions. This is where the trade-off ("cheap steering") reveals itself. (Tool: Small-Angle Approximation.)

PICTURE. The figure overlays three curves near $δ = 0$ : the straight line $δ$ , the curve $sin δ$ (they hug each other), and $1 - cos δ$ (which stays flat then rises slowly).

Step 7 — Two axes at once (dual-gimbal) and the circular limit

WHAT. Let the engine tilt in two perpendicular directions — up/down (pitch, angle $δ_{p}$ ) and left/right (yaw, angle $δ_{y}$ ) — and ask what limit the mechanical hinge imposes.

WHY. One tilt plane steers in one plane only (pitch or yaw). A single engine that tilts in two planes can steer in both. But the hinge can only lean so far from center in any direction, so we must find the combined tilt.

Why a square root? The two tilts are perpendicular pieces of one physical lean, exactly like the two legs of Step 3's triangle. For small tilts the true lean is the hypotenuse of the $(δ_{p}, δ_{y})$ triangle — Pythagoras — so it is $δ_{p}^{2} + δ_{y}^{2}$ , not $δ_{p} + δ_{y}$ .

PICTURE. Looking straight up the engine bell: the tip of the nozzle can reach anywhere inside a circle of radius $δ_{max}$ . A command $(δ_{p}, δ_{y})$ is a point; its distance from center is the total tilt.

The one-picture summary

WHAT. One figure that carries the entire chain: tilt → sideways component → lever → torque, plus the linear-torque / quadratic-loss trade.

Reading it left to right: the engine tilts by $δ$ → the push splits into forward $T cos δ$ (through the CoM, no turn) and sideways $T sin δ$ → that sideways push, a distance $ℓ$ (the moment arm) from the CoM, makes a torque of magnitude $ℓ T sin δ$ (signed $- ℓ T sin δ$ in our frame) → for small tilts in radians, $∣ M ∣ \approx ℓ T δ$ (big turn) while only $\frac{1}{2} δ^{2}$ of thrust is lost (tiny cost).

Recall Feynman retelling — say it in plain words

Imagine pushing a shopping trolley from behind. If you push dead center, it rolls straight. If you angle your push a hair to the left, the trolley still rolls forward almost as fast — but it also starts to turn, because your push now grabs it off to one side. A rocket engine does the same: it swings a few degrees, so most of its shove still goes forward but a slice goes sideways. That sideways slice, acting way down at the tail (far from the balance point), twists the whole rocket around — like pushing a door at its edge instead of its hinge. Which way it turns depends on which way you tilt: tilt one way, nose up; the other way, nose down. Tilt a little, turn a lot; tilt a little, lose almost no speed. Do it in two directions at once and you can point the nose anywhere — except spin it like a top, because pushing straight down the middle can't make it roll. That is thrust vector control.

Recall Rebuild the boxed formula from memory

Start with tilt $δ$ ::: split thrust into $T cos δ$ (forward) and $T sin δ$ (sideways) Which piece turns the rocket? ::: the sideways piece $T sin δ$ Multiply it by what to get torque? ::: the moment arm $ℓ$ from CoM to gimbal Exact torque magnitude? ::: $∣ M ∣ = ℓ T sin δ$ Signed torque in our frame (Step 0)? ::: $M = - ℓ T sin δ$ — positive $δ$ gives clockwise (nose-down) Small-angle version and why? ::: $∣ M ∣ \approx ℓ T δ$ , since $sin δ \approx δ$ for $δ$ in radians Two-axis combined limit (small angles)? ::: $δ_{p}^{2} + δ_{y}^{2} \leq δ_{max}$ , the circular reach limit