LIMA
With God's help
עברית

Signal Transformations

Imagine you are editing a short video in an editing program on a timeline. Shifting is simply dragging the clip forward or backward along the timeline so that it starts later (a delay) or earlier. Time reversal is applying a "reverse" effect, in which the clip plays from the end to the beginning. Scaling is changing the playback speed: double speed \((\times 2)\) squeezes the clip into half of its original time, while slow motion \((\times 0.5)\) stretches it. When you combine them all, the order is critical – just as you first place the clip in the right spot on the timeline and only then change its speed.

The Theory in Brief

Signal transformation deals with changing the independent variable (the time \(t\)), and it works in exactly the same way for continuous and discrete signals. There are three basic operations:

  • Time shift: \(x(t) \to x(t - t_0)\), where for \(t_0 > 0\) the signal moves to the right (a delay) and for \(t_0 < 0\) it moves to the left (an advance).
  • Time reversal: \(x(t) \to x(-t)\), an operation that mirrors the signal about the vertical axis \((t=0)\).
  • Time scaling: \(x(t) \to x(at)\), which compresses the signal when \(|a| > 1\), and expands (stretches) it when \(|a| < 1\).

Combining the three operations gives a linear transformation of the form \(x(t) \to x(at + b)\). To sketch the new signal without falling into mistakes, you follow a fixed order: first you perform the shift \((x(t) \to x(t+b))\), and only then you apply the scaling and the reversal to the variable \(t\) alone \((x(t+b) \to x(at+b))\). For example, to get \(x\left(\frac{3}{2}t + \frac{3}{2}\right)\), we first shift the signal \(\frac{3}{2}\) units to the left to \(x\left(t+\frac{3}{2}\right)\), and only then compress it by a factor of \(\frac{3}{2}\).

Time shiftx(t − t₀) Time reversalx(−t) Time scalingx(at) Shift first, then scalex(at + b)

How to Read the Lab

The original signal \(x(t)\)

A dashed, semi-transparent neon-turquoise line (a "ghost trace"). It never moves, so it is always clear where we started.

The transformed signal \(y(t) = x(at + b)\)

A solid, thick line glowing in magenta-orange. It responds immediately to every slider movement, with a small elastic transition.

The \(y\) axis \((t = 0)\)

A soft white line with a mirror icon at its top: the anchor about which the reversal and the compression happen.

x(at+b)

The dynamic equation

Above the graph: \(y(t) = x(\color{#FF5E7E}{a}\cdot t + \color{#38BDF8}{b})\). The numbers are colored like the sliders and change while you drag.

Shift \(b\) and the direction arrow

The slider moves between \(-8\) and \(8\). While you drag, a glowing arrow appears below the \(t\) axis: "Shift left" for \(b > 0\) and "Shift right" for \(b < 0\).

Scaling \(|a|\)

The slider moves between \(0.25\) and \(4\), with magnetic stops at \(0.5\), \(1\), \(1.5\) and \(2\). The tag next to it switches between "Expansion (stretching)" \((|a| < 1)\) and "Compression" \((|a| > 1)\).

Time reversal \((\pm a)\)

Flips the sign of \(a\). The signal turns over about the \(y\) axis like a page, so you see the right and left sides swap places.

Quick examples

Four buttons that restore \(x(t-4)\), \(x(-t)\), \(x(0.5t)\) and \(x\left(\frac{3}{2}t+\frac{3}{2}\right)\) on the selected signal with one click.

12

Step-by-step demo

The signal resets to \(x(t)\), slides to \(x(t+b)\) (step 1), pauses with a floating explanation, and then stretches or compresses (and flips if \(a < 0\)) to \(x(at+b)\) (step 2). A dotted orange line keeps \(x(t+b)\) for comparison.

\(t\) coordinates on the axis

During the player, dashed lines drop every key point to the \(t\) axis, and orange labels show its \(t\) coordinate in real time, for example while it is divided by \(a\).

Key-Point X-ray

Glowing dots on the start, the break and the end of the signal. Hovering with the mouse (or touching) a point of the new signal draws a dashed line to its twin in the original signal and pops up a label: at which \(t\) the signal reads which value of \(x\).

Speed and pause

The slider slows the player and the animations down to \(\times 0.1\) or speeds them up to \(\times 2\), and ⏸ freezes everything. It changes only the pace of the display.

The readouts

\(a\) and \(b\), the shift direction, where the signal starts and ends on the \(t\) axis, the slope of the sloped segment, where each key point moved, and the point marked by the X-ray.

To the lab ↓
Loading the lab...
x(t)
y(t)
1

Experiments

  1. The minus-and-plus paradox (time shift). Choose the ramp-plateau signal and make sure that \(a=1\). Drag the shift slider \(b\) to \(-4\) to get \(x(t-4)\), and notice where the vector arrow points. Move the mouse (Key-Point X-ray) over the new start point at \(t=0\) and over the break point at \(t=4\). Then drag \(b\) to \(+4\) and watch the change.
    First intuition says that minus belongs to the left, but on the time axis exactly the opposite happens: for \(x(t-t_0)\) with \(t_0 > 0\) the signal moves to the right (a delay), and for \(t_0 < 0\) it moves to the left (an advance). The reason is simple and clear when you look through the X-ray: to get in the new signal the event that originally happened at \(t=0\), the expression inside the parentheses \((t-4)\) has to vanish – and that happens only later, at \(t=4\).
  2. The law of stretching and compression (time scaling). Reset the shift \((b=0)\) on the ramp-plateau signal. Drag the scaling slider \(a\) to \(0.5\) and notice the new range of the signal (from \(-8\) to \(8\)). Now drag \(a\) to \(2\). Pay attention to the length of the whole signal, and especially to what happens to the point that lies exactly on the \(y\) axis \((t=0)\).
    Scaling \(x(at)\) behaves like a playback-speed button: when \(|a| < 1\) time runs slower and so the signal is expanded (stretched), and when \(|a| > 1\) time runs faster and so the signal is compressed. Mathematically, every time coordinate is simply divided by \(a\). That is why the point at \(t=0\) also always stays pinned in place like an anchor – because zero divided by any number is still zero.
  3. The mirror effect (time reversal). Reset the simulation to \(a=1, b=0\) on the ramp-plateau signal. Press the "Time reversal" button to get \(x(-t)\). Follow the mirroring animation about the \(y\) axis and use the Key-Point X-ray to check where the points \(t=-4\) and \(t=4\) went.
    The operation \(x(-t)\) swaps the sign of the time axis and creates a perfect mirror image of the signal about the vertical axis \((t=0)\). Everything that happened in the past \((t<0)\) moves to the future \((t>0)\) and vice versa – so the rising slope to the left of the axis becomes a falling slope to the right of the axis, and the flat part moves to the negative side. Here too, the value at \(t=0\) is the fixed point that does not move.
  4. The order-of-operations trap (linear transformation). Switch to the trapezoid signal. Set \(a = 1.5\) \(\left(\frac{3}{2}\right)\) and \(b = 1.5\) to get \(x\left(\frac{3}{2}t + \frac{3}{2}\right)\), and press the "Step-by-step demo" button. Follow the two steps: first the slide to \(x\left(t+\frac{3}{2}\right)\) (between \(-1.5\) and \(1.5\)), and only then the compression to \(x\left(\frac{3}{2}t+\frac{3}{2}\right)\) (between \(-1\) and \(1\)). Check the three new corners with the X-ray.
    When combining transformations \(x(at+b)\) the order of operations is critical: first you shift \((x(t+b))\), and only then you scale \((x(at+b))\). The reason is that the scaling acts only on the variable \(t\) (and not on the constant \(b\)). In practice, the compression by \(a\) divides the whole time axis – including the shift we already made – and so the start point at \(-1.5\) shrinks to \(-1\).
  5. The combined challenge – shift, reversal and compression together. Stay on the trapezoid signal. Set \(b = 3\), press "Time reversal" and set the scaling to \(a = -2\), so that you get the equation \(x(-2t + 3)\). Before you run the animation, try to guess on which side of the \(y\) axis the signal will land (positive or negative). Now press "Step-by-step demo" and check yourselves.
    This is the most common "landmine" in exams, and the simulation exposes it immediately: even though \(b=+3\) pushes the signal to the left in the first step (to the range \([-3, 0]\)), the second step of reversal and compression \((a=-2)\) mirrors it about \(t=0\) and compresses it by a factor of \(2\), so that it actually lands on the positive side \(([0, 1.5])\). As long as you stick to the iron rule – shift first, and only at the end scale and reverse – no combination of signs will be able to confuse you.