Replacing a branch with a source without changing the rest of the circuit
If, in a circuit that has already been solved, a branch has a voltage \(V\) across it and carries a current \(I\), that branch can be replaced by:
and all the other voltages and currents in the circuit stay the same, provided the new circuit has a unique solution.
It looks obvious, but it is a very useful tool: it lets you split a circuit into pieces, replace an awkward element (nonlinear, a motor, a diode) with something simple, and it is the building block behind the proofs of the Thévenin and Norton and compensation theorems.
We start from the Millman's theorem circuit: two batteries of 12 V and 6 V, with internal resistances of 2 Ω and 3 Ω, feed a 6 Ω load. Solved, the node voltage is \(V_a = 8\) V and the currents are \(I_1 = 2\) A (battery 1), \(I_2 = -0.667\) A (battery 2 is being charged) and \(I_3 = 1.333\) A (load).
The battery currents do not change, and the new source carries the same current that flowed through the load.
The same 8 V is obtained, and with it the same currents in battery 1 and in the load.
A Thévenin or Norton equivalent behaves like the original circuit for any load. Substitution is only valid at that operating point: if something in the circuit changes afterwards (another load, another battery voltage), the source that was put in no longer represents the original branch and the result is wrong.
In the example, if battery 1 went up to 13 V, with the real load the node voltage would rise to 8.5 V; with the 8 V source in its place it would still be 8 V.
Unlike superposition, the substitution theorem does not require linearity: it only uses Kirchhoff's laws. If a diode, a varistor or a lamp is known to work at a given voltage and current, it can be replaced by a source of that value and the rest of the circuit, which is now linear, can be solved. This is what is done, for example, when a conducting diode is approximated by a source of about 0.7 V; the exact substitution uses the real voltage at the operating point.
The unique-solution condition fails when the substitution creates:
With the resistor, 2 A flows. If it is replaced by a 10 V source, the loop law holds (10 − 10 = 0) for any current: the solution is no longer unique and the circuit does not say what it is. On the other hand, replacing the resistor with a 2 A current source is valid.
Related: compensation theorem · Tellegen's theorem · source conversion