Thevenin's Theorem Through a Black-Box Castle: A Fairy-Tale Guide to Vth and Rth, with Derivations and Three Pitfalls

Thévenin's theorem is the first major hurdle in every circuits course. Everyone can recite the statement—"any linear two-terminal network can be replaced by a voltage source in series with a resistor"—but on a real board, many people still don't know where to measure those two values. When I teach this to newcomers, I made up a fairy tale; after telling it, I have them work through the math by hand, and they rarely forget it afterward. This time I'm writing down the story and the formulas together.

The Story: The Black-Box Castle in Misty Valley

Deep in the forest stands a black-box castle in Misty Valley. How many waterwheels, millstones, or stones are inside, nobody can say for sure—all anyone knows is that the outer wall has two brass studs (what we call the two terminals).

A little fox named Ali carries a bucket to the castle every day to fetch water. It doesn't care what the castle looks like inside; it only cares about two things: If I hook up my waterwheel, how much flow will I get? Will it be dragged down?

The old craftsman Grandpa Brass-Stud says: no need to go inside. I'll teach you three spells out here at the gate, and you can fold the entire castle into "one small water pump + one rough pipe."

Spell One: Connect Nothing, First Measure the Water Level at the Gate

Block the outlet (open circuit), and poke a measuring stick that draws no water into the brass studs—that's an ideal voltmeter.

$$V_{th} = V_{oc}$$

This is not an approximation; it's the definition: the value of the Thévenin voltage source is simply the open-circuit voltage at the port. No matter how much is inside the castle, the pressure difference at the gate when nobody is drawing water is exactly that.

Spell Two: Put All the Things That "Spin by Themselves" to Sleep

The castle contains two kinds of self-moving magic: constant-pressure machines (voltage sources) and constant-flow channels (current sources). To measure how rough the pipe is, you must first make them stop:

• The pressure machine no longer produces a pressure difference → it becomes an empty pipe running straight through: the voltage source is shorted;
• The constant-flow channel no longer delivers water → it becomes a fully closed valve: the current source is opened.

Then measure the equivalent resistance looking into the two brass studs—that is the Thévenin internal resistance:

$$R_{th} = R_{eq}\Big|_{\text{voltage sources shorted, current sources opened}}$$

There's a second version that doesn't require disabling the sources—you just have to dare to short the outlet: measure the short-circuit current $I_{sc}$, then

$$R_{th} = \frac{V_{oc}}{I_{sc}}$$

The two versions must agree—and later I'll explain why.

Spell Three: Fold the Entire Castle into a Pump Plus a Pipe

$$\boxed{\;\text{any linear two-terminal network} \;\equiv\; V_{th} \text{ in series with } R_{th}\;}$$

After connecting the load $R_L$ (Ali's waterwheel):

$$I_L = \frac{V_{th}}{R_{th}+R_L},\qquad V_L = V_{th}\cdot\frac{R_L}{R_{th}+R_L}$$

The voltage-divider expression makes it clear: $R_{th}$ is the castle's own roughness, and $R_L$ is your waterwheel. To get more water ($V_L$ close to $V_{th}$), either make $R_L \gg R_{th}$, or find a way to lower the castle's $R_{th}$. From now on, "how much the voltage sags under load" is just one division.

Trial: Hand Calculation on the Voltage-Divider Castle

The first castle Ali meets is simple: a $12\,\mathrm{V}$ pressure machine inside, in series with $R_1 = 6\,\mathrm{k\Omega}$, with $R_2 = 3\,\mathrm{k\Omega}$ in parallel to ground, and the two brass studs taken across $R_2$.

Step one, open-circuit voltage (nothing connected to $R_2$):

$$V_{th} = 12 \times \frac{3\,\mathrm{k}}{6\,\mathrm{k}+3\,\mathrm{k}} = 4\ \mathrm{V}$$

Step two, measure the internal resistance after lights-out (the $12\,\mathrm{V}$ source is shorted, so $R_1$ and $R_2$ become parallel):

$$R_{th} = 6\,\mathrm{k} \parallel 3\,\mathrm{k} = \frac{6\times3}{6+3}\,\mathrm{k} = 2\ \mathrm{k\Omega}$$

Step three, connect the $R_L = 2\,\mathrm{k\Omega}$ waterwheel:

$$V_L = 4 \times \frac{2}{2+2} = 2\ \mathrm{V},\qquad I_L = \frac{4}{2+2}\,\mathrm{mA} = 1\ \mathrm{mA}$$

Now compute directly on the original circuit to verify equivalence. The $2\,\mathrm{k}$ load in parallel with $R_2$ gives $1.2\,\mathrm{k}$, the total current is $12/(6+1.2) = 1.667\,\mathrm{mA}$, and the load gets $1.667 \times 1.2 = 2\ \mathrm{V}$. Exactly the same—and the equivalent circuit has only two components.

Note an engineering habit: for a bias divider network like $R_2$, the source impedance is exactly $R_1 \parallel R_2$. When choosing ADC divider resistors, the "maximum source impedance" in the datasheet is exactly this value—the online voltage divider calculator shows it directly.

Why It Must Hold: The Port V-I Is a Straight Line

A fairy tale needs logic. Let $V$ be the port voltage and $I$ the current flowing out of the port. Since the network contains only linear elements (resistors, controlled sources), the superposition theorem tells us:

$$V = \underbrace{V_{oc}}_{\text{sources acting alone}} \;\;-\;\;\underbrace{I\,R_{th}}_{\text{external current acting alone}}$$

In other words, the port's $V$ and $I$ always lie on the same straight line: the vertical intercept is $V_{oc}$ ($I=0$, open circuit), the horizontal intercept is $I_{sc}$ ($V=0$, short circuit), and the slope is $-R_{th}$.

And the branch "voltage source $V_{th}$ in series with resistor $R_{th}$" gives exactly the same V-I relationship, $V = V_{th} - I R_{th}$. Two points determine a line—so as long as the two ports' $V$–$I$ lines coincide, a person outside the gate can never tell which is the real castle and which is the folded version. That is the entire content of Thévenin's theorem, and it explains why $R_{th} = V_{oc}/I_{sc}$ must equal the value from the lights-out method.

As a bonus we get the Norton version: write that line in current form,

$$I = I_{sc} - \frac{V}{R_{th}},\qquad I_N = I_{sc} = \frac{V_{th}}{R_{th}}$$

which is a current source $I_N$ in parallel with $R_{th}$. Thévenin and Norton are two ways of drawing the same line.

The Three Easiest Pitfalls to Step Into

Pitfall One: You Can Only Compute at the Port, Not Internal Power

This is the most common one in exams and in the field. Look again at the divider castle above with a $2\,\mathrm{k\Omega}$ load:

	Source Output	Delivered to Load	Internal Loss
Real circuit	$12\,\mathrm{V} \times 1.667\,\mathrm{mA} = 20\,\mathrm{mW}$	$2\,\mathrm{V}\times1\,\mathrm{mA}=2\,\mathrm{mW}$	$R_1$ 16.7 mW + $R_2$ 3 mW = 19.7 mW
Thévenin equivalent	$4\,\mathrm{V} \times 1\,\mathrm{mA} = 4\,\mathrm{mW}$	$2\,\mathrm{mW}$	2 mW across $R_{th}$

The port voltage and port current are exactly the same, but the internal loss differs by nearly ten times. The equivalent circuit is only a stand-in "as seen from outside the gate"; it compresses all losses inside the castle into a single $R_{th}$. Using it to calculate efficiency, resistor power dissipation, or temperature rise is simply wrong.

Pitfall Two: Nonlinear Elements Can't Be Applied Directly

Diodes, transistors, saturated op-amps, and inductors with hysteresis all have curved port V-I characteristics that no pair of straight lines can fit. The remedy is small-signal linearization around the operating point: find the slope near a given DC operating point to get $r_{th} = \Delta V/\Delta I$, valid only near that point. That is exactly where the transistor's $r_e$ and the diode's dynamic resistance come from.

Pitfall Three: $R_{th}$ Is Not Necessarily a Positive Resistance

When the network contains controlled sources, the $R_{th}$ computed after lights-out can absolutely be negative—negative resistance means the port voltage rises as the current increases, which is precisely the foundation of oscillators, hysteresis comparators, and negative-resistance converters. Also, $R_{th}$ generally varies with temperature and operating point; a nominal 2 kΩ is only the value under one set of conditions.

The Fairy Tale's Real-World Versions on a Board

• Sensor / signal source output impedance: that's its $R_{th}$. If the next stage's input impedance isn't large enough, $V_L = V_{th}R_L/(R_{th}+R_L)$ immediately takes a hit—that's the entire reason "high-impedance signals need a buffer".

• Maximum power transfer: optimizing over $R_L$, $P_L$ is maximized when $R_L = R_{th}$,

  $$P_{L,\max} = \frac{V_{th}^2}{4R_{th}}$$

  At that point the efficiency is only 50% (half the power burns in $R_{th}$). So RF matching pursues $R_L = R_{th}$, while power-supply design pursues $R_L \gg R_{th}$—the two goals are exactly opposite.

• Wheatstone bridge: when unbalanced, the Thévenin voltage at the output is the voltage difference between the bridge-arm midpoints, and $R_{th}$ is the equivalent resistance of the two half-bridges in parallel, $R_1\parallel R_3 + R_2\parallel R_x$—the next-stage instrumentation amplifier's high input-impedance requirement exists precisely to suppress the divider error from this $R_{th}$. To practice, change the arm values in the online Wheatstone bridge calculator and watch how these two quantities change.

• ADC sampling: at the sampling instant, a SAR ADC must charge its internal sampling capacitor; if the source impedance is too high, it can't fully charge and the reading comes out low. The "maximum source impedance" limit in the datasheet is essentially an RC settling problem between $R_{th}$ and the sampling time.

Small Exercises for Yourself

1. Replace the $12\,\mathrm{V}$ in the trial with $5\,\mathrm{V}$, set $R_1 = 10\,\mathrm{k}$ and $R_2 = 10\,\mathrm{k}$, find $V_{th}$ and $R_{th}$, then connect a $10\,\mathrm{k}$ load to verify. (Answers: $2.5\,\mathrm{V}$, $5\,\mathrm{k}$, $1.67\,\mathrm{V}$)

2. Add a $1\,\mathrm{mA}$ constant-current source in parallel from the divider to ground, and run the lights-out method again—get a feel for what "open-circuiting a current source" really means.

3. Think it through: why does the internal loss computed from the equivalent circuit come out too small? Hint: in the real circuit, the open-circuited $R_2$ has been leaking current all along.

The three spells are all you need to remember: open-circuit for voltage, lights-out for internal resistance, connect a load for the divider. However complex the castle is inside, the world outside the gate is always "one pump in series with one rough pipe."