
Circuit Theory and Network Analysis
Reduce a bridge by star-delta, watch a phasor diagram close, switch an RLC circuit and see why the capacitor voltage cannot jump
28 chapters, 28 interactives, few words. Every chapter is mapped to the Anna University, VTU, SPPU, Mumbai University and KTU units it covers: elements and sources, Kirchhoff's laws, network reduction, mesh and nodal analysis, the network theorems, phasors and impedance, AC and three-phase power, coupled circuits, series and parallel resonance, first- and second-order transients, the Laplace transform in circuits, network functions, graph theory, two-port networks, passive filters and network synthesis. Op-amp and transistor circuits are in the Analog Electronics book; Laplace theory on its own is in the Signals and Systems book.
- AUEC3251, EE3251Circuit Analysis (ECE); Electric Circuit Analysis (EEE)
- VTUBEC304, BEE302Network Analysis (ECE); Electric Circuit Analysis (EEE)
- SPPU203147Network Analysis
- MUECC304, EEC302Network Theory (EXTC); Electrical Circuit Analysis (Electrical)
- KTUPCEET302Circuits and Networks
By Skill-Lync · Updated 19 Nov 2026 · 28 chapters · 28 interactives · free, no sign-up
Circuit elements and sources: R, L, C, ideal and practical, independent and dependent
Which syllabus covers thisAU EC3251 Unit I; EE3251 Unit IVTU BEC304 Module 1; BEE302 Module 1SPPU 203147 Unit 1MU ECC304 Module 1KTU not in syllabus
How do R, L and C each relate voltage and current?
A resistor obeys v = Ri at every instant and turns all the energy it receives into heat. An inductor only shows a voltage while its current changes: v = L di/dt. A capacitor only passes current while its voltage changes: i = C dv/dt. Drive a 1 A peak, 50 Hz sine through 100 mH and the coil shows 31.4 V peak, a quarter cycle ahead of the current; it stores up to ½Li² = 50 mJ and hands it all back, so its average power is zero. A 100 µF capacitor stores ½Cv² instead, up to 50.7 mJ here (all values illustrative).
What is the difference between an ideal and a practical source?
An ideal voltage source holds its voltage whatever current you draw; an ideal current source pushes its current whatever voltage that takes. Real sources sag, so a practical voltage source is drawn as an ideal one with an internal resistance in series, and a practical current source as an ideal one with a resistance in parallel. A 12 V source with 5 Ω inside, loaded by 10 Ω, gives 8 V, not 12 V. Its current-source twin (2.4 A with 5 Ω in parallel) gives the same 8 V, but wastes 12.8 W inside instead of 3.2 W: the two are equal only at the terminals.
What are dependent sources, and where do they come from?
A dependent (controlled) source is set by a voltage or current somewhere else in the circuit, and is drawn as a diamond. There are four kinds. A voltage-controlled voltage source (VCVS) is an op-amp's output, gain times the input difference. A voltage-controlled current source (VCCS) is a transistor in its small-signal model: collector current = gm × base-emitter voltage. A current-controlled current source (CCCS) is the BJT's iC = β·iB. A current-controlled voltage source (CCVS) is a transresistance amplifier that turns a photodiode's current into a voltage. They make transistor circuits solvable with the same mesh and nodal methods.
What do linear, bilateral, lumped and time-invariant mean for a network?
Linear: double every source and every voltage and current doubles; a resistor is linear, a diode is not. Bilateral: the element behaves the same for current in either direction; R, L and C are bilateral, a diode is unilateral. Lumped: the circuit is much smaller than the wavelength, so a wire has one voltage along it. At 50 Hz the wavelength is about 6,000 km, so a room's wiring is lumped; at an FM frequency of 100 MHz it is 3 m, so a long cable is distributed. Time-invariant: element values do not change with time. Active elements (sources) can deliver net energy; passive ones (R, L, C) cannot.
A 1 A peak triangular current at 50 Hz flows through a 100 mH inductor. What does the inductor voltage look like?
v = L di/dt, and a triangle has a constant slope on each ramp: 4 × 1 A × 50 Hz = 200 A/s. So v = 0.1 H × 200 A/s = 20 V, held flat on the rising ramp and flipped to −20 V on the falling one.
Worked examples
A triangular current through a coil · L = 47 mH, triangular current 0.4 A peak, f = 100 Hz
- The current ramps from −0.4 A to +0.4 A in half a period (5 ms): slope = 4·Im·f = 4 × 0.4 × 100 = 160 A/s
- v = L di/dt = 0.047 × 160 = 7.52 V while rising, −7.52 V while falling: a square wave
- Peak stored energy ½LIm² = ½ × 0.047 × 0.4² = 3.76 mJ, reached at each current peak
- From zero current to the peak the source delivers ∫v·i dt = 3.76 mJ; on the way back down it is returned, so the average power is 0
Answer: A ±7.52 V square wave; up to 3.76 mJ stored, zero average power
A practical source and its transformed twin · 9 V source with 1.5 Ω internal resistance, load RL = 6 Ω (illustrative)
- I = 9/(1.5 + 6) = 1.2 A, so V = 1.2 × 6 = 7.2 V at the terminals
- Load power = 7.2 × 1.2 = 8.64 W; loss inside = 1.2² × 1.5 = 2.16 W
- Transform: Is = 9/1.5 = 6 A with 1.5 Ω in parallel. V = 6 × (1.5 || 6) = 6 × 1.2 = 7.2 V, the same
- But the parallel 1.5 Ω now dissipates 7.2²/1.5 = 34.56 W: the twins agree at the terminals, not inside
- Short-circuit current 6 A and open-circuit voltage 9 V are the two ends of the same V-I line
Answer: 7.2 V and 1.2 A at the load for both models; internal loss 2.16 W against 34.56 W
A transistor as a dependent source · BJT biased at IC = 1 mA, 300 K, small-signal input 5 mV peak, collector load 4.7 kΩ
- Thermal voltage VT = kT/q = 25.85 mV, so gm = IC/VT = 38.68 mS
- VCCS: ic = gm·vbe = 38.68 mS × 5 mV = 0.1934 mA peak
- That current through 4.7 kΩ gives 0.909 V peak at the collector, inverted
- Gain = −gm·RL = −181.8
Answer: A VCCS of gm = 38.7 mS turns 5 mV into 0.91 V, a gain of about −182
Ohm's law, Kirchhoff's laws, series-parallel circuits and dividers
Which syllabus covers thisAU EC3251 Unit I; EE3251 Unit IVTU BEE302 Module 1SPPU 203147 Unit 1MU not in syllabusKTU not in syllabus
What does Kirchhoff's current law say, and why must it hold?
The currents entering any node add up to the currents leaving it: the sum, counting entering as + and leaving as −, is zero. Charge cannot pile up in a junction of wires. In the ladder below (12 V, R1 = 1 kΩ, R2 = 3.3 kΩ, R3 = 1.5 kΩ, R4 = 4.7 kΩ, R5 = 2.2 kΩ, all illustrative), 4.667 mA arrives at node B through R1 and leaves as 2.222 mA down R2 plus 2.445 mA along R3. Ohm's law, I = V/R for each resistor, supplies the individual currents.
What does Kirchhoff's voltage law say?
Walk around any closed loop and the voltage rises equal the drops: their signed sum is zero. A charge that returns to where it started has gained and lost the same energy. In the ladder, loop 1 rises 12 V through the source and drops 4.667 V across R1 and 7.333 V across R2. The outer loop works too: 12 = 4.667 + 3.668 + 3.665. Tap any node or loop in the figure and the sum stays at zero, whatever the sliders.
How do you reduce a series-parallel network to one resistance?
Resistors carrying the same current are in series and add. Resistors across the same two nodes are in parallel and combine as R1R2/(R1 + R2), always less than the smaller one. Work from the far end of the ladder: R4 || R5 = 1.4986 kΩ, plus R3 gives 2.9986 kΩ, in parallel with R2 gives 1.571 kΩ, plus R1 gives Req = 2.571 kΩ. The source current is 12 V/Req = 4.667 mA.
When can you use the voltage-divider and current-divider shortcuts?
Two resistors in series share a voltage in proportion to their resistance; two in parallel share a current in inverse proportion. Both assume nothing else draws or injects current between them. In the ladder, B divides 12 V between R1 and the 1.571 kΩ beyond it: VB = 7.333 V. Add a 3 mA source into C and the shortcut still says 7.333 V while the true VB is 8.249 V. A load across a divider is the same trap.
In the default ladder, R3 is raised from 1.5 kΩ to 10 kΩ. What happens to the current in R2?
Less current is diverted down the R3 branch, so the resistance beyond R1 rises, node B's voltage climbs from 7.33 V to 8.63 V, and I2 = VB/R2 rises from 2.222 to 2.616 mA, even though the source current falls.
Worked examples
Solving the ladder by reduction · 12 V; R1 = 1 kΩ, R2 = 3.3 kΩ, R3 = 1.5 kΩ, R4 = 4.7 kΩ, R5 = 2.2 kΩ
- R4 || R5 = 4.7 × 2.2/6.9 = 1.4986 kΩ; plus R3: 2.9986 kΩ
- In parallel with R2: Rx = 1.571 kΩ; Req = 1 + 1.571 = 2.571 kΩ
- I = 12/2.571 kΩ = 4.6674 mA; VB = I·Rx = 7.3326 V
- I2 = VB/R2 = 2.222 mA, I3 = VB/(R3 + R4 || R5) = 2.4454 mA (KCL: they add to 4.6674)
- VC = I3 × 1.4986 kΩ = 3.6645 V; I4 = 0.7797 mA, I5 = 1.6657 mA
Answer: Req = 2.571 kΩ, I = 4.667 mA, VB = 7.333 V, VC = 3.665 V
A divider that gets loaded · 9 V across two 10 kΩ resistors; then a 10 kΩ, and separately a 100 kΩ, load across the lower one
- Unloaded: Vout = 9 × 10/(10 + 10) = 4.5 V
- 10 kΩ load: lower leg = 10 || 10 = 5 kΩ, Vout = 9 × 5/15 = 3 V
- 100 kΩ load: lower leg = 10 || 100 = 9.091 kΩ, Vout = 4.286 V
- Rule of thumb: keep the load at least ten times the divider resistance
Answer: 4.5 V unloaded, 3 V with 10 kΩ, 4.29 V with 100 kΩ
The ladder with a current source added · The same ladder with Is = 3 mA injected into node C
- Divider shortcuts no longer apply: write KCL at B and C (A is held at 12 V)
- B: (VB − 12)/1k + VB/3.3k + (VB − VC)/1.5k = 0; C: (VC − VB)/1.5k + VC/4.7k + VC/2.2k = 3 mA
- Cramer's rule: VB = 8.2487 V, VC = 6.3713 V
- KCL at C: 1.252 mA (R3) + 3 mA = 1.356 + 2.896 mA (R4, R5)
- The voltage-divider formula would still give 7.3326 V, which is wrong now
Answer: VB = 8.249 V, VC = 6.371 V; the source now delivers 3.751 mA
Network reduction: source transformation, source shifting and star-delta conversion
Which syllabus covers thisAU EC3251 Unit II; EE3251 Unit IIVTU BEC304 Module 1; BEE302 Module 1SPPU 203147 Unit 1MU EEC302 Module 1KTU PCEET302 Module 1
What is source transformation, and what does it not preserve?
A voltage source Vs with a series resistance R can be swapped for a current source Vs/R with the same R in parallel, and back. Every load sees the same voltage and current, because both have the same open-circuit voltage and short-circuit current. It lets you merge parallel voltage sources or series current sources. What it does not preserve is the inside: 10 V with 100 Ω feeding 220 Ω gives 31.25 mA either way, but the internal resistor dissipates 97.7 mW in one form and 472.7 mW in the other.
What is source shifting?
Sometimes a source sits where no transformation applies: a voltage source with no series resistor, or a current source with no parallel one. Source shifting moves it. A voltage source in the wire into a node can be pushed through the node into every branch leaving it, one copy in each: a 6 V source feeding three branches becomes three 6 V sources, one per branch, and every KVL loop is unchanged. A current source between two nodes can be replaced by equal sources across each element of any path joining those nodes, which keeps KCL at every node. Each copy then has a resistor to transform with.
How do you convert between delta (Δ) and star (Y)?
Three resistors in a triangle (Δ) and three meeting at a centre (Y) can look identical from their three terminals. Δ to Y: each arm is the product of the two Δ resistors at that terminal over the sum of all three. Y to Δ: each Δ resistor is the sum of the pairwise products over the opposite arm. For the bridge in the figure (R1 = 100, R2 = 200, R5 = 100 Ω) the Δ becomes a star of 50, 25 and 50 Ω. Three equal resistors convert simply: Y = Δ/3.
Why can a bridge not be reduced by series and parallel, and what happens when it is balanced?
In a bridge, the cross arm R5 joins the middles of two dividers, so no two resistors share a current or both nodes. One star-delta conversion breaks the deadlock. The bridge is balanced when R1·R4 = R2·R3: both dividers then split the source in the same ratio, the ends of R5 sit at the same voltage, and no current flows in it. With R1 = 100, R2 = 200, R3 = 150 Ω, balance needs R4 = 300 Ω; then R5 can be removed and Req = (100 + 150) || (200 + 300) = 166.7 Ω, whatever R5 is.
The bridge is balanced (R1 = 100, R2 = 200, R3 = 150, R4 = 300 Ω) and Req is 166.7 Ω. You change R5 from 100 Ω to 10 Ω. What is Req now?
At balance both ends of R5 are at the same voltage, so it carries no current whatever its value. Removing it, shorting it or changing it leaves every other current, and Req, exactly as they were.
Worked examples
Reducing the bridge by Δ → Y · 10 V source; R1 = 100 Ω (A-B), R2 = 200 Ω (A-C), R3 = 150 Ω (B-D), R4 = 220 Ω (C-D), R5 = 100 Ω (B-C)
- Δ A-B-C: Σ = 100 + 100 + 200 = 400 Ω
- Ra = R1·R2/Σ = 50 Ω, Rb = R1·R5/Σ = 25 Ω, Rc = R2·R5/Σ = 50 Ω
- Series: Rb + R3 = 175 Ω, Rc + R4 = 270 Ω
- Parallel: 175 || 270 = 106.18 Ω; Req = Ra + 106.18 = 156.18 Ω
- Source current I = 10/156.18 = 64.03 mA
Answer: Req = 156.2 Ω, I = 64 mA
The same bridge by Y → Δ · The same bridge; convert the star at B (R1 to A, R3 to D, R5 to C)
- P = R1R3 + R3R5 + R5R1 = 40000 Ω²
- RAD = P/R5 = 400 Ω, RAC = P/R3 = 266.67 Ω, RCD = P/R1 = 400 Ω
- RAC || R2 = 114.29 Ω and RCD || R4 = 141.94 Ω, in series: 256.22 Ω
- Req = RAD || 256.22 = 156.18 Ω
- Check by nodal analysis: VB = 5.827 V, VC = 5.54 V, so R5 carries 2.88 mA (not balanced: 100 × 220 ≠ 200 × 150)
Answer: Req = 156.2 Ω, the same as Δ → Y
Merging two practical sources · 12 V with 2 Ω and 6 V with 3 Ω, connected in parallel, feeding a 4 Ω load (illustrative)
- Transform each: 12/2 = 6 A with 2 Ω, and 6/3 = 2 A with 3 Ω, both in parallel
- Add the currents and the conductances: 8 A with 2 || 3 = 1.2 Ω
- Transform back: 8 × 1.2 = 9.6 V in series with 1.2 Ω
- Load current = 9.6/(1.2 + 4) = 1.846 A; load voltage 7.385 V (nodal check: 7.385 V)
Answer: 1.85 A in the 4 Ω load, at 7.38 V
Mesh and nodal analysis: supermesh, supernode and dependent sources
Which syllabus covers thisAU EC3251 Unit I; EE3251 Unit IVTU BEC304 Module 1; BEE302 Module 1SPPU 203147 Unit 1MU ECC304 Module 1; EEC302 Module 1KTU PCEET302 Module 1
How do you write mesh equations by inspection?
Give every window of a planar circuit a clockwise mesh current; there are b − n + 1 of them (branches minus nodes plus one). For each mesh, the diagonal entry is the total resistance around it, each off-diagonal entry is minus the resistance shared with the neighbour, and the right side is the source voltage rise around the mesh. With R1 = 2.2, R2 = 4.7, R3 = 1 kΩ and sources of 10 V and 5 V, the matrix is [6.9 −4.7; −4.7 5.7] and the currents come out 1.943 and 0.725 mA. A passive circuit always gives a symmetric matrix.
How do you write nodal equations by inspection?
Pick a reference node; the other n − 1 node voltages are the unknowns. For each node, the diagonal entry is the total conductance touching it, each off-diagonal entry is minus the conductance shared with the other node, and the right side is the source current flowing in. With 5 mA into node 1, 1 mA out of node 2 and the same three resistors, the node voltages are 7.658 V and 0.519 V. Choose whichever method has fewer unknowns: the two-mesh circuit above needs only one nodal equation.
What are a supermesh and a supernode?
A current source shared by two meshes fixes their difference, I2 − I1 = Is, but its voltage is unknown, so you cannot write KVL around either mesh alone. Write that constraint, then one KVL around the loop that avoids the source: the supermesh. With a 2 mA source in the middle leg, I1 = −0.177 mA (negative: it really flows the other way) and I2 = 1.823 mA. A voltage source between two non-reference nodes is the dual: V2 − V1 = Vs, plus one KCL for both nodes together. With 3 V, V1 = 0.927 V and V2 = 3.927 V.
How do dependent sources change mesh and nodal analysis?
Write the dependent source as if it were independent, then express its controlling voltage or current in terms of the unknowns and move that term to the left side. Replace the 5 V source in the mesh circuit with a current-controlled voltage source of 2 kΩ × I1: mesh 2's row becomes [−2.7 5.7], since the −4.7 picks up +2. The matrix is no longer symmetric, which is the sign that a controlled source is present. Now I1 = 2.14 mA, I2 = 1.014 mA and the dependent source shows 4.28 V.
A circuit has 6 nodes, one of them the reference, and two voltage sources each connected from a node to the reference. How many nodal equations do you need?
There are 6 − 1 = 5 node voltages, but each grounded voltage source fixes one of them outright. That leaves 5 − 2 = 3 unknowns, so 3 KCL equations.
Worked examples
Two meshes, and the same circuit by nodal · 10 V source, R1 = 2.2 kΩ, shared R2 = 4.7 kΩ, R3 = 1 kΩ, 5 V source (+ at the top), clockwise mesh currents
- Mesh 1: (2.2 + 4.7)I1 − 4.7·I2 = 10; mesh 2: −4.7·I1 + (4.7 + 1)I2 = −5 (the 5 V source opposes I2)
- Determinant = 6.9 × 5.7 − 4.7² = 17.24
- I1 = (10 × 5.7 − 4.7 × 5)/17.24 = 1.9432 mA; I2 = (6.9 × (−5) + 4.7 × 10)/17.24 = 0.7251 mA
- Nodal needs one unknown, the middle node: V = (10/2.2 + 5/1)/(1/2.2 + 1/4.7 + 1/1) = 5.7251 V
- Check: V/R2 = 1.2181 mA = I1 − I2
Answer: I1 = 1.943 mA, I2 = 0.725 mA; R2 carries 1.218 mA down
A supermesh · 10 V source, R1 = 2.2 kΩ (top left), a 2 mA source pointing up in the middle leg, R2 = 4.7 kΩ (top right), R3 = 1 kΩ (right leg)
- The middle leg carries I2 − I1 upward, so I2 − I1 = 2 mA
- KVL round the outer loop, which skips the source: 10 = 2.2·I1 + (4.7 + 1)·I2
- Substitute I2 = I1 + 2: 10 = 7.9·I1 + 11.4, so I1 = −0.1772 mA
- I2 = 1.8228 mA. Negative I1: 0.177 mA really flows right to left through R1 and into the + terminal of the 10 V source, so the 2 mA source is pushing current back through it and the 10 V source absorbs 1.77 mW
Answer: I1 = −0.177 mA, I2 = 1.823 mA
A current-controlled voltage source · The two-mesh circuit with the 5 V source replaced by a CCVS of 2 kΩ × I1 (+ at the top)
- Mesh 2 KVL: −1·I2 − 2·I1 − 4.7(I2 − I1) = 0, so (−4.7 + 2)·I1 + 5.7·I2 = 0
- Matrix: [6.9 −4.7; −2.7 5.7][I1; I2] = [10; 0], no longer symmetric
- Determinant = 6.9 × 5.7 − 4.7 × 2.7 = 26.64
- I1 = 10 × 5.7/26.64 = 2.1396 mA, I2 = 2.7 × 10/26.64 = 1.0135 mA
- The dependent source shows 2 kΩ × 2.1396 mA = 4.279 V
Answer: I1 = 2.14 mA, I2 = 1.014 mA, CCVS = 4.28 V
Linearity and the superposition theorem
Which syllabus covers thisAU EC3251 Unit II; EE3251 Unit IIVTU BEC304 Module 2; BEE302 Module 2SPPU 203147 Unit 2MU ECC304 Module 1; EEC302 Module 1KTU PCEET302 Module 1
What makes a circuit linear, and why does that matter?
A circuit of resistors, inductors, capacitors, linear dependent sources and independent sources is linear: every current and voltage is a weighted sum of the source values. Two rules follow. Homogeneity: double a source and every response it causes doubles. Additivity: the response to two sources together is the sum of the responses to each. In the interactive network, raising Vs from 12 V to 24 V raises its share of the R2 current from 3.09 mA to 6.17 mA, exactly twice. A diode, or a lamp whose resistance rises as it heats, breaks both rules.
How do you apply superposition?
Keep one independent source on, switch every other independent source off, and find the current you want. Repeat for each source, then add the answers with their signs. In the network above (Vs = 12 V, Is = 10 mA, R2 = 2.2 kΩ, illustrative), Vs alone drives I′ = 3.09 mA down R2 and Is alone drives I″ = 1.77 mA, so with both on I = 4.86 mA, exactly what a full solve gives. Each sub-problem is a plain series-parallel circuit, so no simultaneous equations are needed.
What does switching a source off mean, and what about dependent sources?
An ideal voltage source switched off is 0 V whatever its current: a short circuit. An ideal current source switched off is 0 A whatever its voltage: an open circuit. A practical source leaves its internal resistance behind. Dependent sources are never switched off, because their value is set by a voltage or current inside the circuit, and that changes from one sub-problem to the next. Leave them in every sub-circuit and solve each one fully. In Example 3, a 3·i1 source stays on in both halves: 6.25 V + 1.5 V = 7.75 V at node A.
Why can you not add powers by superposition?
Power is I²R, which is not linear in I. With R2 = 2.2 kΩ, (I′ + I″)²R2 = 51.86 mW, but I′²R2 + I″²R2 = 27.84 mW. The missing 24.02 mW is the cross term 2·I′·I″·R2. Add the currents or voltages first, then square. The cross term only vanishes when one share is zero, or, for sinusoids of different frequencies, when you average over time: then average powers do add, which is how the power in harmonics is counted.
Source A alone drives 3 mA through a 1 kΩ resistor; source B alone drives 4 mA through it in the same direction. What power does the resistor take with both on?
Currents superpose, powers do not: I = 3 + 4 = 7 mA, so P = (7 mA)² × 1 kΩ = 49 mW. Adding 9 mW and 16 mW misses the cross term 2 × 3 mA × 4 mA × 1 kΩ = 24 mW.
Worked examples
Two sources sharing a load · V1 = 15 V with R1 = 3 Ω and V2 = 9 V with R2 = 6 Ω, both feeding a common node; load R3 = 4 Ω from that node to ground (illustrative)
- V1 alone (V2 shorted): R = 3 + (6 || 4) = 5.4 Ω, I = 15/5.4 = 2.7778 A; R3 share = × 6/10 = 1.6667 A
- V2 alone (V1 shorted): R = 6 + (3 || 4) = 7.7143 Ω, I = 9/7.7143 = 1.1667 A; R3 share = × 3/7 = 0.5 A
- Both on: I3 = 1.6667 + 0.5 = 2.1667 A
- Check by nodal analysis: V = (15/3 + 9/6)/(1/3 + 1/6 + 1/4) = 8.6667 V, I3 = V/4 = 2.1667 A
- Power in R3 = 2.1667² × 4 = 18.78 W, not 11.11 + 1 = 12.11 W
Answer: I3 = 2.167 A, P3 = 18.8 W
A voltage source and a current source · Vs = 20 V in series with R1 = 4 Ω feeding node A; R2 = 6 Ω from A to ground; a 2 A current source pushing current into A (illustrative)
- Vs alone (current source open): I2′ = 20/(4 + 6) = 2 A
- Is alone (Vs shorted): 2 A splits between R2 and R1 to ground: I2″ = 2 × 4/(4 + 6) = 0.8 A
- Both: I2 = 2 + 0.8 = 2.8 A
- Check by nodal: VA = (20/4 + 2)/(1/4 + 1/6) = 16.8 V, I2 = 16.8/6 = 2.8 A
Answer: I2 = 2.8 A (2 A from the voltage source, 0.8 A from the current source)
Superposition with a dependent source left on · Vs = 10 V – R1 = 2 Ω – node A; R2 = 4 Ω from A to ground; R3 = 2 Ω from A to a CCVS of 3·i1 volts (i1 = current in R1 towards A); a 3 A source into A (illustrative)
- KCL at A: (Vs − vA)/2 + Is = vA/4 + (vA − 3·i1)/2, with i1 = (Vs − vA)/2; the CCVS stays on in every sub-circuit
- Vs alone (Is open): 5 − 0.5vA = 1.5vA − 7.5, so vA′ = 6.25 V (i1′ = 1.875 A)
- Is alone (Vs shorted, i1 = −vA/2): 3 − 0.5vA = 1.5vA, so vA″ = 1.5 V (i1″ = −0.75 A)
- Sum: vA = 6.25 + 1.5 = 7.75 V; a direct solve of both equations gives 7.75 V and i1 = 1.125 A
- Current in R2 = 7.75/4 = 1.9375 A
Answer: vA = 7.75 V, I(R2) = 1.938 A
Thevenin and Norton equivalents, including dependent sources
Which syllabus covers thisAU EC3251 Unit II; EE3251 Unit IIVTU BEC304 Module 2; BEE302 Module 2SPPU 203147 Unit 2MU ECC304 Module 1; EEC302 Module 1KTU PCEET302 Module 1
What does Thevenin's theorem say, and how do you find Vth and Rth?
Any linear network, seen from two terminals a-b, behaves like one voltage source Vth in series with one resistance Rth. Vth is the open-circuit voltage at a-b. Rth is Vth divided by the short-circuit current, or, when there are only independent sources, the resistance seen at a-b with every source switched off. For 12 V, a 1 kΩ / 2.2 kΩ divider and R3 = 1 kΩ (illustrative), Vth = 12 × 2.2/3.2 = 8.25 V and Rth = (1 k || 2.2 k) + 1 k = 1.6875 kΩ. Helmholtz derived it in 1853, Thévenin in 1883.
How is the Norton equivalent related to Thevenin's?
Norton's form is a current source IN = Isc in parallel with the same Rth. Source transformation turns one into the other: IN = Vth/Rth, here 8.25/1687.5 = 4.889 mA. Both give the same terminal behaviour at every load: with RL = 2.2 kΩ, the Thevenin form gives 8.25/(1.6875 k + 2.2 k) = 2.122 mA, and the Norton current divider gives 4.889 mA × 1.6875/(1.6875 + 2.2) = 2.122 mA. Hans Ferdinand Mayer and Edward Norton published it independently in 1926.
How do dependent sources change the method, and can Rth be negative?
A dependent source cannot be switched off, so find Rth as Voc/Isc, or, if the network has no independent source at all (Voc = Isc = 0), drive a-b with a test current and read the voltage. In the interactive network a VCCS gm·v2 feeds terminal a, giving Rth = (Rp + R3)/(1 − gm·Rp) with Rp = 687.5 Ω. Below gm = 1.45 mS, Rth is positive. Above it the dependent source pushes out more current than the resistors draw, and Rth turns negative: gm = 3 mS and R3 = 1 kΩ give −1.59 kΩ. Negative resistance is what keeps an oscillator running.
How can you find a real source's Vth and Rth by measurement?
Measure the terminal voltage twice. Open circuit gives Vth directly. Then add a known load and measure again: the drop divided by the load current is Rth. An illustrative bench supply reads 12.6 V open and 12 V with a 10 Ω load, so I = 1.2 A and Rth = (12.6 − 12)/1.2 = 0.5 Ω. Measuring Isc directly would mean shorting it, about 25.2 A here, which can damage the source, so the two-load method is the safer route.
A network contains resistors and a dependent source but no independent source. What are its Thevenin voltage and Norton current?
With nothing to excite it, every voltage and current is zero, so Voc = Isc = 0 and Voc/Isc is 0/0. Apply a test source (say 1 mA into a) and Rth = V/1 mA. The network is just a resistance, possibly a negative one.
Worked examples
Thevenin equivalent of an unbalanced bridge · A 10 V source across two dividers: left arm 100 Ω (top) and 150 Ω (bottom) meeting at a, right arm 100 Ω and 200 Ω meeting at b; a 50 Ω galvanometer joins a and b (illustrative)
- Open circuit: va = 10 × 150/250 = 6 V, vb = 10 × 200/300 = 6.6667 V, so Vth = va − vb = −0.6667 V
- Sources off (10 V shorted): Rth = (100 || 150) + (100 || 200) = 60 + 66.667 = 126.667 Ω
- Galvanometer current: Ig = −0.6667/(126.667 + 50) = −3.774 mA (from b to a)
- Check: a nodal solve of the whole bridge with the 50 Ω in place gives −3.774 mA
- Balance needs equal ratios, 100/150 = 100/R: the bottom-right arm would have to be 150 Ω for zero current
Answer: Vth = −0.667 V, Rth = 126.7 Ω, Ig = 3.77 mA
Norton equivalent of a current-source network · A 5 A source in parallel with 10 Ω, followed by a 5 Ω series resistor, with 15 Ω across the terminals a-b (illustrative); load 2.5 Ω
- Transform 5 A || 10 Ω into 50 V in series with 10 Ω; add the 5 Ω: 50 V behind 15 Ω
- Open circuit: Vth = 50 × 15/(15 + 15) = 25 V
- Sources off: Rth = 15 || 15 = 7.5 Ω
- Short a-b: the 15 Ω across a-b is bypassed, so IN = 50/15 = 3.3333 A (= Vth/Rth)
- Load 2.5 Ω: IL = 3.3333 × 7.5/(7.5 + 2.5) = 2.5 A
Answer: IN = 3.333 A in parallel with 7.5 Ω; IL = 2.5 A
Rth with a dependent source · Vs = 10 V, R1 = R2 = 2 kΩ (divider, v2 across R2), R3 = 1.5 kΩ from v2 to a; a VCCS of 0.5 mS × v2 pushes current into a (illustrative)
- Open circuit: KCL at a gives (va − v2)/R3 = gm·v2, and at v2: (v2 − 10)/2k + v2/2k − gm·v2 = 0, so v2 = 10 V and Vth = v2(1 + gm·R3) = 17.5 V
- Short circuit: v2 = 5 mA/(0.5 + 0.5 + 0.667) mS = 3 V, Isc = v2(1/R3 + gm) = 3.5 mA
- Rth = Vth/Isc = 17.5/3.5 mA = 5 kΩ
- Test source on the dead network (Vs = 0, 1 mA into a): va = 5 V, so Rth = 5 kΩ, the same
- Closed form Rth = (Rp + R3)/(1 − gm·Rp) with Rp = 1 kΩ; at gm = 1.5 mS it gives (1k + 1.5k)/(1 − 1.5) = −5 kΩ: negative
Answer: Vth = 17.5 V, Rth = 5 kΩ (and −5 kΩ at gm = 1.5 mS)
Maximum power transfer: DC, AC and the conjugate match
Which syllabus covers thisAU EC3251 Unit II; EE3251 Unit IIVTU BEC304 Module 2; BEE302 Module 2SPPU 203147 Unit 2MU ECC304 Module 1; EEC302 Module 1KTU PCEET302 Module 1
Which load takes the most power from a DC source?
Reduce the source to Vth and Rth. The load power is P = Vth²·RL/(Rth + RL)². Too small an RL and the voltage collapses; too large and the current does. The peak is at RL = Rth. With the interactive's illustrative 10 V and 10 Ω, Pmax = 2.5 W, and a 25 Ω load gets 2.04 W, 82 % of it. The peak is flat: any RL from 5.2 to 19.2 Ω keeps at least 90 %.
Why are power systems not run at maximum power transfer?
At the match the load and Rth carry the same current and have the same resistance, so they take equal power: efficiency η = RL/(Rth + RL) = 50 %. Moritz von Jacobi published the result around 1840, and it is also called Jacobi's law. A supply that wasted half its energy would be absurd, so power circuits use a load far larger than the source resistance: with Rth = 0.1 Ω and RL = 10 Ω, η = 99 %. Matching suits signal paths such as an antenna feeding a receiver, where the signal is tiny and efficiency hardly matters.
What is the conjugate match for AC?
With Zth = Rth + jXth, the load power is |Vth|²·RL/((Rth + RL)² + (Xth + XL)²), with Vth as an RMS phasor. The reactance term can only hurt, so first make XL = −Xth, then RL = Rth: ZL = Zth*, the complex conjugate. Pmax = |Vth|²/(4Rth), as in DC. For 10 V RMS and Zth = 10 + j20 Ω (illustrative), ZL = 10 − j20 Ω takes 2.5 W, while a plain 25 Ω load takes only 1.54 W because the source reactance is left uncancelled.
What if the load can only be a resistor?
If XL is fixed (zero for a pure resistor), only RL can move. Setting dP/dRL = 0 gives RL = |Zth + jXL|, so a purely resistive load should equal the magnitude |Zth|. For Zth = 10 + j20 Ω that is 22.4 Ω, which takes 1.545 W, 62 % of the conjugate-match 2.5 W. The formula is Pmax = |Vth|²/(2(Rth + |Zth|)). This is why a capacitor or a matching network is added when the full power is wanted.
A 10 V source has Rth = 10 Ω. Which load resistor takes more power: 5 Ω or 20 Ω?
P/Pmax = 4r/(1 + r)² with r = RL/Rth, which is unchanged when r becomes 1/r. Both 0.5 and 2 give 8/9: 2.222 W each, against 2.5 W at 10 Ω.
Worked examples
Best load for a resistive network · 24 V source, 6 Ω and 12 Ω divider, 4 Ω series resistor to the load terminals (illustrative)
- Thevenin: Vth = 24 × 12/18 = 16 V, Rth = (6 || 12) + 4 = 8 Ω
- Best load RL = 8 Ω: Pmax = 16²/(4 × 8) = 8 W
- RL = 4 Ω: P = 16² × 4/12² = 7.111 W; RL = 16 Ω: P = 16² × 16/24² = 7.111 W, the same
- Inside the real network at the match: the 24 V source delivers 48 W, so only 16.7 % reaches the load (the Thevenin 50 % is not the real source's efficiency)
Answer: RL = 8 Ω, Pmax = 8 W
Conjugate match on 50 Hz · Vth = 50 V RMS, Zth = 20 + j15 Ω (inductive), 50 Hz (illustrative)
- Best load: ZL = Zth* = 20 − j15 Ω
- The −j15 Ω is a capacitor: C = 1/(2π × 50 × 15) = 212.2 µF
- Loop impedance = 40 Ω (resistive), I = 50/40 = 1.25 A, P = 1.25² × 20 = 31.25 W
- Check by a two-variable golden-section search on P(RL, XL): RL = 20 Ω, XL = −15 Ω
Answer: ZL = 20 − j15 Ω (20 Ω with 212 µF), P = 31.25 W
Resistor-only load on the same source · The source above (50 V RMS, 20 + j15 Ω) but the load must be a pure resistor
- RL = |Zth| = √(20² + 15²) = 25 Ω
- |Z| loop = |45 + j15| = 47.434 Ω, I = 50/47.434 = 1.0541 A
- P = I² × 25 = 27.778 W, 88.9 % of the conjugate match
- For comparison a 20 Ω resistor (equal to Rth) gives 27.397 W: the DC rule is not the best here
Answer: RL = 25 Ω, P = 27.78 W
Reciprocity, Millman, Tellegen, substitution and duality
Which syllabus covers thisAU EC3251 Unit II; EE3251 Unit IIVTU BEC304 Module 2; BEE302 Module 1SPPU 203147 Units 1, 2MU EEC302 Modules 1, 3KTU PCEET302 Module 1
What does Millman's theorem give for sources in parallel?
When several branches, each a voltage source Vk in series with Rk, meet at one pair of nodes, the common voltage is the conductance-weighted average V = Σ(Vk/Rk)/Σ(1/Rk); a plain resistor counts as a branch with Vk = 0. With the interactive's illustrative 12 V/1 kΩ, 6 V/2.2 kΩ, 5 V/3.3 kΩ and a 4.7 kΩ load, V = 8.243 V. Think of a balance: each source is a weight of size 1/Rk at position Vk, and V is the balance point. Jacob Millman published it in 1940.
What does the reciprocity theorem say, and when does it fail?
In a linear network of R, L, C and coupled coils with no dependent sources, an ideal voltage source at port 1 and an ideal ammeter at port 2 can swap places without changing the reading. In the interactive T network (1 kΩ, 2.2 kΩ, 3.3 kΩ, illustrative) 12 V gives 2.069 mA both ways. As two-port parameters, z12 = z21. Add the 0.5 mS dependent source and the readings become 5.483 mA and 1.034 mA: an amplifier is not reciprocal, which is why it has a forward direction.
What is Tellegen's theorem?
Take every branch with its voltage v and current i, using the passive sign convention. Then Σ v·i = 0: the power the sources deliver equals what everything else absorbs. In the Millman circuit the sources deliver 34.05 mW and the resistors absorb the same; the 6 V source is actually being charged (6.12 mW in). The proof uses only KCL and KVL, so it holds for nonlinear elements too, and even with v from one network and i from another network of the same graph. Bernard Tellegen published it in 1952.
What does the substitution theorem allow?
If you know the voltage v across a branch and the current i through it, you may replace that branch by an independent voltage source of value v, or a current source of value i, and nothing else in the network changes. In the Millman circuit the 4.7 kΩ load sits at 8.243 V with 1.754 mA, so a 8.243 V source in its place leaves every other branch current the same. It is used to cut a network into pieces, and to handle a nonlinear element once its operating point is known.
What is duality in circuits?
Swap every quantity for its dual and a true statement stays true: KVL and KCL, mesh and node, series and parallel, R and G, L and C, voltage source and current source, short and open, Thevenin and Norton. A series RLC loop driven by 12 V (R = 4 Ω, L = 10 mH, C = 100 µF) has a dual parallel circuit driven by 12 A with G = 4 S (R = 0.25 Ω), C = 10 mF and L = 100 µH. Its node equation has the same numbers as the original's mesh equation. Only planar networks, drawable without crossings, have duals.
In a passive resistive network, 10 V at port 1 drives 3 mA through an ideal ammeter at port 2. You swap them: 10 V at port 2, the ammeter at port 1. What does it read?
Reciprocity: for any linear network without dependent sources, the transfer from port 1 to port 2 equals the transfer from 2 to 1, so the ammeter reads 3 mA again in the same sense. A transistor stage inside would break this.
Worked examples
Three batteries in parallel with a load (Millman) · Batteries 12 V/0.5 Ω, 12.6 V/0.3 Ω and 11.8 V/0.4 Ω (EMF/internal resistance, illustrative) in parallel, feeding 5 Ω
- ΣG·V = 12/0.5 + 12.6/0.3 + 11.8/0.4 = 24 + 42 + 29.5 = 95.5 A
- ΣG = 2 + 3.333 + 2.5 + 0.2 = 8.0333 S
- V = 95.5/8.0333 = 11.888 V
- Battery currents (EMF − V)/R: 0.224 A, 2.373 A, −0.22 A (negative: the 11.8 V battery is being charged)
- Check: 0.224 + 2.373 + (−0.22) = 2.378 A = load current 11.888/5
Answer: V = 11.888 V; load 2.378 A; the weakest battery takes 0.22 A
Checking reciprocity on a T network · T network: 2 Ω from port 1 to the centre node, 3 Ω from the centre to port 2, 6 Ω from the centre to ground; a 10 V source and an ideal ammeter (illustrative)
- Source at 1, ammeter at 2: R = 2 + (3 || 6) = 4 Ω, I = 2.5 A; ammeter share 6/(3 + 6): 1.6667 A
- Source at 2, ammeter at 1: R = 3 + (2 || 6) = 4.5 Ω, I = 2.2222 A; ammeter share 6/(2 + 6): 1.6667 A
- Both equal 10 × 6/(2·3 + 2·6 + 3·6) = 1.6667 A: the transfer resistance is 10/1.6667 = 6 Ω each way
Answer: 1.667 A in both arrangements
Tellegen's theorem, one network and two · 18 V source, 3 Ω in series, then 6 Ω in parallel with 3 Ω (illustrative); second network: same graph, the 3 Ω shunt replaced by 6 Ω
- Network 1: R = 3 + 2 = 5 Ω, I = 3.6 A; v(R1) = 10.8 V, v(shunts) = 7.2 V; i = 1.2 A and 2.4 A
- Powers (passive sign): source −64.8 W, R1 38.88 W, 6 Ω 8.64 W, 3 Ω 17.28 W; sum = 0
- Network 2: I = 18/6 = 3 A, branch currents 3, 1.5, 1.5 A
- Mixed sum Σ v1·i2 = 18 × (−3) + 10.8 × 3 + 7.2 × 1.5 + 7.2 × 1.5 = 0
Answer: Σ v·i = 0 W, and Σ v1·i2 = 0 even across the two networks (sources deliver 64.8 W in network 1)
Sinusoids, average and RMS values, and phasors
Which syllabus covers thisAU EC3251 Unit III; EE3251 Unit IVTU not in syllabusSPPU not in syllabusMU not in syllabusKTU not in syllabus
What three numbers describe a sinusoid?
v(t) = Vm sin(ωt + θ). Vm is the peak, ω = 2πf the angular frequency, θ the phase. Indian mains (nominal 230 V rms, 50 Hz) has Vm = 230 × √2 = 325.3 V, ω = 314.16 rad/s and a period of 20 ms. A phase angle is just a time shift: 30° of a 50 Hz wave is 30/360 of 20 ms = 1.667 ms. A wave with larger θ reaches its peak earlier, so it leads.
What do the RMS and average values mean?
The RMS value is the steady DC value that would heat the same resistor equally: square the wave, average over a period, take the root. For a sine it is Vm/√2, so 230 V mains is an RMS figure with a 325.3 V peak. A sine averages to zero over a full cycle, so its average value is taken over a half cycle (or of |v|): 2Vm/π = 207.1 V for the mains. Meters, ratings and heating all use RMS.
What are the form factor and the peak factor?
Form factor = RMS/average; peak (crest) factor = peak/RMS. Sine: 1.111 and 1.414. Half-wave rectified sine: 1.571 and 2. Square: 1 and 1. Triangle: 1.155 and 1.732. A cheap meter measures the average of |v| and multiplies by the sine's 1.111. On a 100 V-peak square wave it shows 111.1 V instead of 100 V (11.1 % high); on a triangle it is 3.8 % low. A true-RMS meter has no such error.
What is a phasor, and why does it make adding sines easy?
Picture an arrow of length Vm turning anticlockwise at ω, starting at angle θ. Its vertical projection traces Vm sin(ωt + θ). Two sines of the same frequency turn together, so their sum is the sum of the arrows, drawn tip to tail. Example: 10 sin ωt + 6 sin(ωt + 60°) = 14 sin(ωt + 21.8°), not 16. This book writes phasors as RMS values unless it says peak: mains is 230∠0° V.
Two 50 Hz voltages, each 100 V rms and 120° apart, are connected in series. What does an RMS voltmeter across both read?
Add the phasors: 100∠0° + 100∠120° = −50 + j86.6 + 100 = 50 + j86.6, which is 100∠60° V. Magnitudes add only when the waves are in phase.
Worked examples
Reading the mains as a sinusoid · 230 V rms, 50 Hz, v(t) = Vm sin(ωt − 45°). Find Vm, ω, v at t = 4 ms, and when v first crosses zero going up and first peaks
- Vm = 230 × √2 = 325.27 V; ω = 2π × 50 = 314.16 rad/s; T = 20 ms
- At t = 4 ms, ωt = 0.004/0.020 × 360° = 72°, so v = 325.27 × sin(72° − 45°) = 325.27 × sin 27° = 147.67 V
- Upward zero crossing: ωt − 45° = 0, ωt = 45°, t = 45/360 × 20 ms = 2.5 ms
- Positive peak: ωt − 45° = 90°, ωt = 135°, t = 7.5 ms
Answer: Vm = 325.3 V, ω = 314.16 rad/s, v(4 ms) = 147.7 V; zero crossing at 2.5 ms, peak at 7.5 ms
RMS of a stepped inverter-style waveform · An illustrative stepped wave: +280 V for 120° in the middle of each positive half cycle, −280 V likewise in the negative half, 0 V otherwise
- v² = 280² for 240° of every 360°: Vrms = 280 × √(240/360) = 228.62 V
- Average of |v| = 280 × 240/360 = 186.67 V
- Form factor = 228.62/186.67 = 1.2247; peak factor = 280/228.62 = 1.2247
- An average-responding meter shows 1.1107 × 186.67 = 207.33 V, 9.3 % below the true RMS
- A sine with the same RMS would have a 323.3 V peak
Answer: Vrms = 228.6 V (close to 230 V), but an average-responding meter reads 207.3 V
Adding two branch currents · Two branch currents of 50 Hz feeding one node: I1 = 6∠−30° A and I2 = 4∠45° A (RMS phasors). Find the total and write it as a time function
- I1 = 6 cos(−30°) + j6 sin(−30°) = 5.196 − j3 A
- I2 = 4 cos 45° + j4 sin 45° = 2.828 + j2.828 A
- I = I1 + I2 = 8.025 − j0.172 A = 8.026∠−1.22° A
- Peak = 8.026 × √2 = 11.35 A, so i(t) = 11.35 sin(314.16t − 1.22°) A
Answer: I = 8.03∠−1.22° A rms, not 10 A: the two currents are 75° apart
Impedance, admittance, phasor diagrams and AC mesh and nodal analysis
Which syllabus covers thisAU EC3251 Unit III; EE3251 Unit IVTU BEC304 Module 1; BEE302 Module 1SPPU not in syllabusMU EEC302 Module 1KTU PCEET302 Module 1
What is impedance?
Impedance Z = V/I with both as phasors, so it carries a size and an angle. A resistor is R. An inductor is jωL: at 50 Hz, 100 mH is j31.42 Ω, and at 500 Hz it is j314.2 Ω. A capacitor is 1/(jωC) = −j/(ωC): 47 µF is −j67.73 Ω at 50 Hz and −j6.77 Ω at 500 Hz. In series the impedances add, exactly like resistors at DC.
How is a series RLC phasor diagram drawn?
Take the current as reference, since it is common to all three. VR lies along I, VL points 90° ahead, VC 90° behind. Add them tip to tail and they close on the source V. With 100 Ω, 470 mH and 33 µF on 230 V, 50 Hz: I = 2.047 A, VR = 204.7 V, VL = 302.3 V, VC = 197.5 V. VL is larger than the supply: L and C partly cancel. V leads I by 27.1°. At 40.4 Hz the circuit would look purely resistive.
What is admittance, and why use it for parallel branches?
Admittance Y = 1/Z = G + jB, in siemens. Parallel branches share one voltage, so their admittances simply add. Put the same 100 Ω, 470 mH and 33 µF in parallel across 230 V: G = 0.01 S, B = ωC − 1/(ωL) = 0.00359 S. Then I = V·Y = 2.444 A, leading V by 19.8°. The branch currents IR = 2.3 A, IC = 2.38 A and IL = 1.56 A add tip to tail.
How do mesh and nodal analysis work for AC?
Exactly as for DC, with phasor sources and impedances. Write the equations by inspection, then solve them with complex arithmetic. A two-mesh circuit gives a 2 × 2 complex matrix; one unknown node gives one complex equation. Each answer comes out as a magnitude and an angle. Convert it back to a time function only at the end. Example 3 below solves one circuit both ways and gets the same 141∠−38.5° V.
A series RLC circuit is fed from 230 V. Can a voltmeter across the inductor alone read more than 230 V?
Voltages add as phasors. With 100 Ω, 470 mH, 33 µF at 50 Hz, VL = 302 V and VC = 197 V point opposite ways. Their difference, 105 V, adds at 90° to VR = 205 V and gives 230 V.
Worked examples
A coil on the mains · A coil with R = 12 Ω and L = 80 mH on 230 V, 50 Hz
- XL = 2π × 50 × 0.08 = 25.13 Ω, so Z = 12 + j25.13 Ω
- |Z| = √(12² + 25.13²) = 27.85 Ω, angle = atan(25.13/12) = 64.48°
- I = 230/27.85 = 8.258 A, lagging V by 64.5°
- VR = 99.1 V, VL = 207.6 V; check √(99.1² + 207.6²) = 230 V
Answer: I = 8.26 A at power factor 0.431 lagging
A series RLC circuit and its resonant frequency · R = 47 Ω, L = 220 mH, C = 68 µF in series on 230 V, 50 Hz
- XL = 314.16 × 0.22 = 69.12 Ω; XC = 1/(314.16 × 68 µF) = 46.81 Ω
- Z = 47 + j(69.12 − 46.81) = 47 + j22.3 Ω = 52.02∠25.39° Ω
- I = 230/52.02 = 4.421 A
- VR = 207.8 V, VL = 305.6 V, VC = 206.9 V; VL alone exceeds the supply
- Resonance: f0 = 1/(2π√(0.22 × 68 µF)) = 41.15 Hz
Answer: I = 4.42 A lagging by 25.4°; f0 = 41.1 Hz
One circuit by nodal and by mesh analysis · 230∠0° V source, series Z1 = 10 + j20 Ω to node A; from A to the return, Z2 = −j25 Ω in parallel with Z3 = 15 + j10 Ω
- Nodal at A: (VA − 230)/Z1 + VA/Z2 + VA/Z3 = 0, so VA = (230/Z1)/(1/Z1 + 1/Z2 + 1/Z3) = 140.98∠−38.49° V
- Mesh: (Z1 + Z2)J1 − Z2·J2 = 230 and −Z2·J1 + (Z2 + Z3)J2 = 0
- Determinant = 700 − j225 Ω²; J1 = 6.636∠−27.18° A, J2 = 7.82∠−72.18° A
- VA = J2 × Z3 = 140.98∠−38.49° V: the same as the nodal answer
Answer: VA = 141∠−38.5° V; source current 6.64∠−27.2° A; current in Z3 7.82∠−72.2° A
AC power: instantaneous, average, reactive, apparent and complex power
Which syllabus covers thisAU EC3251 Unit III; EE3251 Unit IVTU BEE302 Module 5SPPU not in syllabusMU not in syllabusKTU PCEET302 Module 2
What does instantaneous power p(t) look like on AC?
p = v·i at every instant. With v and i both at 50 Hz, the product pulses at 100 Hz around a steady average. That average is the real power P = VI cos φ, in watts. Take a 40 + j30 Ω load on 230 V: I = 4.6 A, φ = 36.87°, P = 846.4 W. Yet p(t) swings from 1904.4 W down to −211.6 W. For a short part of each cycle the load hands energy back to the supply.
What is reactive power?
An inductor or capacitor stores energy for a quarter cycle and returns it the next. For a pure inductance of j30 Ω on 230 V, I = 7.67 A and p(t) swings ±1763 W at 100 Hz, yet its average is zero. That swing amplitude, Q = VI sin φ, is the reactive power, in VAR. It does no net work but still needs current in every wire. By convention an inductor absorbs +Q and a capacitor −Q.
What are apparent and complex power?
Complex power S = V·I* (I* is the conjugate of the RMS current phasor) packs both into one number: S = P + jQ. Its size |S| = VI is the apparent power in VA. Cables, fuses and transformers are rated by it, because current heats them whatever the power factor. For 40 + j30 Ω on 230 V, S = 846.4 + j634.8 VA and |S| = 1058 VA. In any network, the sum of S delivered by the sources equals the sum absorbed by the elements.
Why does power factor matter?
Power factor = P/|S| = cos φ for sinusoids. The 40 + j30 Ω load takes 846.4 W at pf 0.8, drawing 4.6 A. At unity pf the same power would need only 3.68 A. Feeder loss goes as I², so pf 0.8 costs 56.2 % more loss in the wires. A capacitor across an inductive load supplies the Q locally (example 3). The Power Systems book covers correction in depth.
A pure inductor is connected to 230 V, 50 Hz mains. What does its instantaneous power p(t) do?
p = v·i = VI sin ωt × sin(ωt − 90°) × 2 = −VI sin 2ωt: it swings at twice the supply frequency between +VI and −VI, so the average (real) power is zero. The inductor only borrows energy.
Worked examples
From a meter reading to P, Q and an equivalent circuit · An illustrative inductive load takes 6 A from 230 V, 50 Hz at a power factor of 0.75 lagging
- S = 230 × 6 = 1380 VA; P = 1380 × 0.75 = 1035 W
- sin φ = √(1 − 0.75²) = 0.6614; Q = 1380 × 0.6614 = 912.8 VAR
- |Z| = 230/6 = 38.333 Ω; R = |Z| cos φ = 28.75 Ω; X = |Z| sin φ = 25.36 Ω
- Equivalent L = X/ω = 80.7 mH; check: I²R = 36 × 28.75 = 1035 W
Answer: P = 1035 W, Q = 913 VAR; series model 28.75 Ω with 80.7 mH
Two loads on one supply: complex power adds · On 230 V: load A takes 1.5 kW at pf 0.8 lagging; load B takes 600 W at pf 0.6 leading
- QA = 1500 × tan(acos 0.8) = 1125 VAR; QB = −600 × tan(acos 0.6) = −800 VAR
- S = (1500 + 600) + j(1125 − 800) = 2100 + j325 VA
- |S| = 2125 VA; pf = 2100/2125 = 0.9882 lagging
- Supply current = 2125/230 = 9.239 A, although the loads draw 8.152 A and 4.348 A
Answer: S = 2100 + j325 VA; I = 9.24 A, less than 8.15 + 4.35 = 12.5 A
Raising the power factor with a capacitor · The load of example 1 (1035 W at pf 0.75 lagging, 230 V, 50 Hz), corrected to pf 0.95 lagging
- tan φ1 = tan(acos 0.75) = 0.8819; tan φ2 = tan(acos 0.95) = 0.3287
- QC = P(tan φ1 − tan φ2) = 1035 × 0.5532 = 572.6 VAR
- C = QC/(ωV²) = 572.6/(314.16 × 230²) = 34.45 µF
- New current = 1035/(230 × 0.95) = 4.737 A, down from 6 A
Answer: About 34.5 µF; the line current falls from 6 A to 4.74 A while P stays 1035 W
Three-phase circuits: star and delta, balanced and unbalanced loads, neutral shift
Which syllabus covers thisAU EE3251 Unit VVTU BEE302 Module 5SPPU not in syllabusMU not in syllabusKTU PCEET302 Module 2
How are line and phase quantities related in star and delta?
A three-phase supply is three sinusoids 120° apart, in the sequence R-Y-B. In star, each phase sits between a line and the neutral. The nominal 400 V workshop supply has a phase voltage of 400/√3 = 230.9 V (230 V nominal). The line voltage VRY = VRN − VYN is √3 times larger and leads VRN by 30°. Line and phase currents are equal. In delta, each phase sits across two lines: phase voltage = line voltage, and line current = √3 × phase current, lagging it by 30°.
Why is a balanced load easy?
Three equal impedances draw three equal currents 120° apart, and those add to zero. So the neutral carries nothing and one phase tells the whole story. For 16 + j12 Ω per phase in star on 400 V: Iph = 230.9/20 = 11.55 A, P = √3 × 400 × 11.55 × 0.8 = 6.4 kW. Connect the same three impedances in delta and each sees 400 V: IL = 34.6 A and P = 19.2 kW, three times as much.
What happens with an unbalanced load and a neutral wire?
The neutral holds the load star point at the source neutral. So every load still sees its full phase voltage and draws Vph/Z, independently of the others. The three currents no longer cancel, and the difference returns in the neutral: IN = IR + IY + IB. With resistive loads of 20, 40 and 80 Ω the line currents are 11.55, 5.77 and 2.89 A. The neutral carries 7.64 A, more than two of the lines. Total power is 4.67 kW.
What is neutral shift, and why is a broken neutral dangerous?
Remove the neutral and the load star point N′ floats to wherever the currents sum to zero. Millman's theorem gives VN′N = Σ(Vk/Zk)/Σ(1/Zk). For the same 20, 40 and 80 Ω loads, VN′N = 87.3 V. The loads now see 151, 262 and 302 V instead of 231 V. The lightest load gets the highest voltage. That is why a broken neutral in a building can burn out equipment on its lightly loaded phase.
Three resistive loads, 30 Ω on R, 75 Ω on Y and 50 Ω on B, run from a 400 V star supply. The neutral wire breaks. Which load now sees the highest voltage?
Without the neutral, the star point shifts toward the heaviest load (30 Ω) and away from the lightest. The 75 Ω load on Y rises to 280 V and the 30 Ω load on R falls to 174 V.
Worked examples
The same load in star and in delta · Three impedances of 16 + j12 Ω on a 400 V, 50 Hz three-phase supply, first in star, then in delta
- |Z| = 20 Ω, cos φ = 16/20 = 0.8 lagging
- Star: Vph = 400/√3 = 230.94 V, IL = Iph = 230.94/20 = 11.547 A
- Star: P = √3 × 400 × 11.547 × 0.8 = 6400 W, Q = 4800 VAR
- Delta: Iph = 400/20 = 20 A, IL = √3 × 20 = 34.64 A
- Delta: P = 3 × 400 × 20 × 0.8 = 19200 W
Answer: Star 6.4 kW at 11.55 A per line; delta 19.2 kW at 34.64 A per line (3 times)
Unbalanced star load with the neutral connected · Resistive loads 30 Ω (R), 75 Ω (Y), 50 Ω (B) in star on 400 V with the neutral wire; VRN = 230.94∠0° V
- IR = 230.94/30 = 7.698∠0° A; IY = 230.94/75 = 3.079∠−120° A; IB = 230.94/50 = 4.619∠120° A
- IN = IR + IY + IB = 3.849 + j1.333 A
- |IN| = 4.073 A at 19.1°
- P = 230.94² × (1/30 + 1/75 + 1/50) = 3556 W
Answer: Each load sees 230.9 V; IN = 4.07∠19.1° A; P = 3.56 kW
The same load with the neutral broken · The loads of example 2, neutral wire open
- Millman: VN′N = (VRN/30 + VYN/75 + VBN/50)/(1/30 + 1/75 + 1/50)
- Numerator = 3.849 + j1.333 A (the 4-wire IN again); denominator = 0.06667 S
- VN′N = 61.1∠19.1° V
- Load voltages VRN′ = VRN − VN′N = 174.4 V, VYN′ = 280 V, VBN′ = 249.8 V
- Line currents 5.812, 3.733, 4.996 A (they now sum to zero); P = 3307 W
- Check by two mesh equations: IR = 5.812∠−6.6° A, the same
Answer: The star point shifts 61.1 V; the 75 Ω load rises to 280 V and the 30 Ω load falls to 174 V
Three-phase power measurement: the two-wattmeter method and power factor
Which syllabus covers thisAU EE3251 Unit VVTU BEE302 Module 5SPPU not in syllabusMU not in syllabusKTU PCEET302 Module 2
Why do two wattmeters measure all the power of a three-wire load?
With no neutral wire, the three line currents add to zero, so only two of them are free. Put current coils in two lines and return both pressure coils to the third line, and the two readings add up to the total power for any load, balanced or not (Blondel's theorem: n wires need n − 1 meters). In the figure, 400 V feeds 25 Ω at 37° lagging per phase: W1 = 3667 W, W2 = 1444 W, and W1 + W2 = 5111 W, exactly 3·I²·R.
What does each meter read on a balanced load?
Take phase sequence R-Y-B and a lagging angle φ. The B-line meter sees V_BY and I_B, which are (30° − φ) apart. The R-line meter sees V_RY and I_R, which are (30° + φ) apart. So W1 = VL·IL·cos(30° − φ) and W2 = VL·IL·cos(30° + φ). At unity pf they are equal. At φ = 60° (pf 0.5) W2 reads zero, and beyond that it goes negative: reverse its current-coil leads to get an upscale reading and record it as negative.
How do you get the power factor from the two readings?
Subtract and add the two formulas: W1 − W2 = VL·IL·sin φ and W1 + W2 = √3·VL·IL·cos φ. So tan φ = √3(W1 − W2)/(W1 + W2) and the reactive power is Q = √3(W1 − W2). Readings of 5.2 kW and 1.8 kW give φ = 40.1° and pf 0.765. Which meter reads higher depends on the phase sequence. Swap two supply lines and the 37° lagging load of the figure reads as 37° leading, so check the sequence first.
When do you need three wattmeters, or can one do?
A four-wire star needs three meters, because the neutral carries current and the three line currents no longer add to zero. Feed 20 Ω, j40 Ω and 30 − j10 Ω from 400 V with the neutral connected. The neutral carries 7.8 A, and two meters would show 5.94 kW against a true 4.27 kW. On a known balanced load, one meter in one phase times three is enough. On an unbalanced three-wire load, the two-meter sum is still right but the tan φ formula is not.
A balanced three-wire motor load runs at a power factor of 0.4 lagging. What do the two wattmeters (R-Y-B sequence) show?
pf 0.4 means φ = 66.4°, which is past 60°. W2 = VL·IL·cos(30° + 66.4°) = VL·IL·cos(96.4°) < 0. The total is still W1 + W2 (W1 minus the size of W2).
Worked examples
Readings on a balanced star load · 400 V, 50 Hz, R-Y-B; star load Z = 20 + j15 Ω per phase; current coils in R and B, pressure coils to Y
- |Z| = 25 Ω, φ = 36.87° lagging; Vph = 400/√3 = 230.94 V, IL = 9.238 A
- W1 = 400 × 9.238 × cos(30° − 36.87°) = 3668.5 W
- W2 = 400 × 9.238 × cos(30° + 36.87°) = 1451.5 W
- Check: 3·I²·R = 3 × 9.238² × 20 = 5120 W = W1 + W2
- tan φ = √3(3668.5 − 1451.5)/5120 = 0.75, so φ = 36.87°
Answer: W1 ≈ 3.67 kW, W2 ≈ 1.45 kW, total 5.12 kW at pf 0.8 lagging
From two readings to the power factor · Balanced load on 400 V. (a) W1 = 5.2 kW, W2 = 1.8 kW. (b) W1 = 3.0 kW, W2 = −1.0 kW (read after reversing its current coil)
- (a) P = 7.0 kW, tan φ = √3 × 3.4/7.0 = 0.8413, φ = 40.07°, pf = 0.765
- (a) Q = √3 × 3.4 = 5.89 kVAR; IL = P/(√3·VL·pf) = 13.2 A
- (b) P = 3.0 − 1.0 = 2.0 kW, tan φ = √3 × 4.0/2.0 = 3.464, φ = 73.9°
- (b) pf = 0.277, Q = √3 × 4.0 = 6.93 kVAR: a lightly loaded motor is mostly reactive
Answer: (a) pf 0.765, P = 7.0 kW, Q = 5.89 kVAR; (b) pf 0.277, P = 2.0 kW
Blondel on an unbalanced three-wire load · 400 V, R-Y-B, star load without neutral: ZR = 40 Ω, ZY = j40 Ω, ZB = 40 − j20 Ω
- Neutral shift (Millman): VN′N = Σ(V/Z)/Σ(1/Z) = 142 V ∠140.1°
- Line currents |IR|, |IY|, |IB| = 8.8, 7.28, 2.44 A
- True power Σ|I|²R = 3095.2 + 0 + 238.2 = 3333.3 W
- W1 = Re(V_BY·conj IB) = 845.3 W, W2 = Re(V_RY·conj IR) = 2488 W; sum 3333.3 W
- The tan φ formula would give pf 0.761, which means nothing here: the load has no single φ
Answer: W1 + W2 = 3333.3 W = the true total power, even though the load is unbalanced
Coupled circuits: mutual inductance, dot rule, coefficient of coupling, transformers
Which syllabus covers thisAU EC3251 Unit V; EE3251 Unit IVVTU not in syllabusSPPU 203147 Unit 1MU ECC304 Module 1; EEC302 Module 1KTU PCEET302 Module 4
What are mutual inductance and the coefficient of coupling?
When two coils share magnetic flux, a changing current in one induces a voltage in the other: v2 = M·di1/dt. M is the mutual inductance, and it can never exceed √(L1·L2). The ratio k = M/√(L1·L2) is the coefficient of coupling, from 0 (no shared flux) to 1 (all flux shared). Coils of 20 mH and 45 mH with k = 0.8 have M = 0.8 × 30 = 24 mH. If k were above 1, some pair of currents would give negative stored energy, which no passive coil can have.
What does the dot rule say, and how does it change series inductance?
A dot marks the coil ends whose induced voltages rise together. Current entering a dotted end induces a voltage that is positive at the other coil's dotted end. In series, current entering both dots (aiding) gives Leq = L1 + L2 + 2M. Entering one dot and leaving by the other (opposing) gives L1 + L2 − 2M. For 20 mH, 45 mH and M = 24 mH that is 113 mH or 17 mH. Measure both and M = (L_aid − L_opp)/4.
How can coupled coils be replaced by a plain T of inductors?
If the two coils share a common bottom wire, the pair behaves exactly like a T: series arms L1 − M and L2 − M with a shunt M (dots aiding), or L1 + M, L2 + M and −M (opposing). With 20 mH, 45 mH and k = 0.8 an arm comes out as −4 mH. A negative inductance is fine in a model, though you cannot buy one. Both circuits in the figure carry the same currents: at 1 kHz with RL = 100 Ω, |I1| = 154.7 mA and |I2| = 77.8 mA.
What is reflected impedance, and how does a linear transformer differ from an ideal one?
Solve the secondary loop for I2 and put it back into the primary. The source then sees its own coil plus (ωM)²/(Z2 + ZL), the reflected impedance. Loading the secondary changes the primary current through this term. A linear transformer is just two coupled coils with finite L and k < 1. The ideal transformer is the limit k = 1 with L1 and L2 very large. Then V2/V1 = √(L2/L1) = n and the source sees ZL/n². With 20 H and 80 H at k = 1 the ratio is already 2 to four decimal places.
Coils of 20 mH and 45 mH measure 101 mH when connected series aiding. What will they measure connected series opposing?
Aiding: 20 + 45 + 2M = 101, so 2M = 36 mH and M = 18 mH. Opposing: 65 − 36 = 29 mH. The difference between the two readings is 4M.
Worked examples
Finding M and k from two series measurements · Series aiding reads 140 mH, series opposing reads 60 mH, coil 1 alone reads 40 mH
- L_aid − L_opp = 4M, so M = (140 − 60)/4 = 20 mH
- L_aid = L1 + L2 + 2M, so L2 = 140 − 40 − 40 = 60 mH
- k = M/√(L1·L2) = 20/√(40 × 60) = 0.408
Answer: M = 20 mH, L2 = 60 mH, k ≈ 0.41 (loose coupling)
A linear transformer with a resistive load · 10 V rms, 1 kHz source with Rs = 10 Ω; L1 = 20 mH, L2 = 80 mH, k = 0.9, dots aiding; RL = 200 Ω
- M = 0.9 × √(20 × 80) = 36 mH; ωM = 226.19 Ω, ωL1 = 125.66 Ω, ωL2 = 502.7 Ω
- Reflected impedance (ωM)²/(RL + jωL2) = 34.96 − j87.88 Ω
- I1 = 10/(10 + j125.66 + reflected) = 170.3 mA ∠−40°
- I2 = jωM·I1/(RL + jωL2) = 71.2 mA, so V2 = I2·RL = 14.24 V
- Power check: 10 × Re(I1) = 1.3034 W = Rs loss 0.2899 W + load 1.0135 W
Answer: |I1| ≈ 170 mA, |I2| ≈ 71.2 mA, V2 ≈ 14.24 V (an ideal 1:2 transformer would give 16.7 V)
T-equivalent with opposing dots · L1 = 20 mH, L2 = 45 mH, k = 0.8, secondary dot at the far end; find the T and the input inductance with the secondary shorted
- M = 24 mH; opposing dots: arms L1 + M = 44 mH and L2 + M = 69 mH, shunt −M = −24 mH
- Shorted secondary, from the T: 44 + (−24 × 69)/(−24 + 69) = 7.2 mH
- Directly: L1 − M²/L2 = 20 − 24²/45 = 7.2 mH (the dot position does not matter here)
Answer: Arms 44 mH and 69 mH, shunt −24 mH; input inductance 7.2 mH either way
Series resonance: Q factor, bandwidth, selectivity and voltage magnification
Which syllabus covers thisAU EC3251 Unit IV; EE3251 Unit IVVTU BEC304 Module 5; BEE302 Module 3SPPU not in syllabusMU not in syllabusKTU PCEET302 Module 1
What happens at series resonance?
In a series R-L-C circuit, XL = ωL rises with frequency and XC = 1/ωC falls. At f0 = 1/(2π√(LC)) they are equal and cancel, so the impedance is just R. It is at its smallest there, the current at its largest and in phase with the voltage (pf 1). With 10 Ω, 10 mH and 1 µF, f0 = 1591.5 Hz and a 10 V source drives 1 A. Below f0 the circuit is capacitive, above it inductive.
What is the Q factor, and why is it called voltage magnification?
At resonance the full source voltage sits across R, but L and C each carry I0·XL = Q times the source voltage, in antiphase. In the figure Q = ω0L/R = 10, so a 10 V source puts 100 V across the capacitor. Rate the capacitor for the magnified voltage, not the supply. Q also measures energy stored over energy lost per radian of a cycle, so a low-loss coil gives a high Q.
What are the half-power frequencies and the bandwidth?
The half-power frequencies f1 and f2 are where the current falls to I0/√2, so the power in R halves. There the net reactance equals ±R. Solving gives f1,2 = ∓R/(4πL) + √((R/4πL)² + f0²). The bandwidth f2 − f1 = R/(2πL) = f0/Q, and f0 = √(f1·f2) is their geometric mean. For the default circuit f1 = 1514 Hz, f2 = 1673.1 Hz and BW = 159.2 Hz. A narrow band means high selectivity.
Is the capacitor voltage largest at f0?
Not quite. VC = I/(ωC) and the 1/ω factor favours lower frequencies, so |VC| peaks at f0√(1 − 1/(2Q²)) and |VL| just above f0. At Q = 10 that is 1587.6 Hz against 1591.5 Hz, about 100.13 V instead of 100 V. At Q = 1 (R = 100 Ω) the peak falls to 0.707·f0. Tuning C at a fixed frequency, the current peaks at resonance, but VC peaks where XC = (R² + XL²)/XL.
In the default circuit (10 Ω, 10 mH, 1 µF) you double R to 20 Ω. What happens?
f0 = 1/(2π√(LC)) has no R in it. Q = ω0L/R falls from 10 to 5 and BW = R/(2πL) rises from 159.2 Hz to 318.3 Hz. The peak current halves too.
Worked examples
Resonance, Q and the capacitor voltage · Series R = 15 Ω, L = 25 mH, C = 160 nF, source 12 V rms
- LC = 4 × 10⁻⁹ s², ω0 = 1/√(LC) = 15811.4 rad/s, f0 = 2516.5 Hz
- At f0: Z = R = 15 Ω, I0 = 12/15 = 0.8 A
- XL = XC = ω0L = 395.3 Ω, Q = 395.3/15 = 26.35
- VC = VL = I0·XC = 316 V, which is Q × 12 V
Answer: f0 ≈ 2516 Hz, Q ≈ 26.4, VC ≈ 316 V from a 12 V source
Half-power frequencies · Series R = 47 Ω, L = 10 mH, C = 47 nF
- ω0 = 1/√(LC) = 46127 rad/s (f0 = 7341.3 Hz); α = R/(2L) = 2350 s⁻¹
- ω1 = −α + √(α² + ω0²) = 43836 rad/s, f1 = 6976.8 Hz
- ω2 = α + √(α² + ω0²) = 48536 rad/s, f2 = 7724.8 Hz
- BW = f2 − f1 = 748 Hz = R/(2πL); Q = f0/BW = 9.81; √(f1·f2) = 7341.3 Hz = f0
Answer: f1 ≈ 6977 Hz, f2 ≈ 7725 Hz, BW ≈ 748 Hz, Q ≈ 9.8
Tuning a mains circuit with C (and why it is dangerous) · Illustrative: a 0.5 H coil with 20 Ω resistance in series with a capacitor C on 230 V, 50 Hz
- XL = 2π × 50 × 0.5 = 157.08 Ω; resonance needs XC = XL, so C = 1/(ω·XL) = 20.26 µF
- At resonance I = 230/20 = 11.5 A and VC = I·XC = 1806 V, Q = 7.85
- Half-power C: XC = XL ± R gives C = 17.98 µF and 23.22 µF
- VC is largest at XC = (R² + XL²)/XL, C = 19.94 µF: 1821 V
Answer: C ≈ 20.3 µF; the capacitor would see about 1.8 kV from a 230 V supply
Parallel resonance: the practical tank, dynamic impedance and single tuned circuits
Which syllabus covers thisAU EC3251 Unit IV; EE3251 Unit IVVTU BEC304 Module 5; BEE302 Module 3SPPU 203147 Unit 6MU not in syllabusKTU PCEET302 Module 1
What is anti-resonance in a practical tank circuit?
A real coil has resistance, so a practical tank is (R + L) in parallel with C. At one frequency the inductive and capacitive branch currents cancel in their reactive parts, and the tank looks like a pure resistance: anti-resonance. Setting the imaginary part of the admittance to zero gives the formula below. The coil resistance pulls f_ar a little under f0. With 10 Ω, 1 mH and 100 nF, f0 = 15915.5 Hz and f_ar = 15835.7 Hz.
What is dynamic impedance, and what is current magnification?
At f_ar the tank's impedance is purely resistive and works out exactly to Zdyn = L/(CR), here 1000 Ω from a coil of only 10 Ω. That is about Q²·R with Q = ω0L/R = 10. So the line current is small: 1 mA from a 1 V source. The current going round the loop between L and C is much bigger, 9.95 times the line current here (ω_ar·L/R). That is current magnification, the parallel twin of the series circuit's voltage magnification.
Where exactly is |Z| largest, and what if the capacitor branch has resistance too?
|Z| does not peak at f_ar. With 10 Ω, 1 mH and 100 nF it peaks at 15915.1 Hz, between f_ar and f0, at 1005 Ω. With resistance RC in the capacitor branch too, f_ar = f0·√((L − C·RL²)/(L − C·RC²)). If RL = RC = √(L/C) = 100 Ω the tank is resistive at every frequency. If the brackets differ in sign (say RL > 100 Ω with RC = 0), there is no anti-resonance at all.
What is a single tuned circuit?
Put a tank in place of an amplifier's collector resistor. The transistor acts as a current source, so the output voltage is I·Z, large only near f_ar. The stage amplifies one band and rejects the rest, with bandwidth about f0/Q. In an FM receiver (88 to 108 MHz) a 0.1 µH coil tunes across the band with C from 32.7 pF down to 21.7 pF. The transistor's own output resistance sits across the tank and lowers the loaded Q, which widens the band.
A tank with Q = 10 draws 2 mA from its source at anti-resonance. Roughly how much current flows through its capacitor?
At f_ar the capacitor current is about Q times the line current (exactly ω_ar·L/R times). So about 20 mA circulates between L and C while only 2 mA comes from the source.
Worked examples
Anti-resonant frequency, Zdyn and the circulating current · Coil R = 25 Ω, L = 2 mH in parallel with C = 50 nF, fed from 5 V rms at f_ar
- 1/LC = 10¹⁰ s⁻², R²/L² = 1.5625 × 10⁸ s⁻², so ω_ar = 99216 rad/s and f_ar = 15790.7 Hz (f0 = 15915.5 Hz)
- Zdyn = L/(CR) = 2 × 10⁻³/(50 × 10⁻⁹ × 25) = 1600 Ω
- Line current = 5/1600 = 3.125 mA
- Capacitor current = V·ω·C = 24.8 mA; coil current = 25 mA
- Magnification 7.94 ≈ Q0 = ω0L/R = 8
Answer: f_ar ≈ 15791 Hz, Zdyn = 1600 Ω, 3.125 mA from the source, about 24.8 mA circulating
Resistance in both branches · L = 1 mH with RL = 20 Ω, in parallel with C = 100 nF with RC = 10 Ω
- f0 = 1/(2π√(LC)) = 15915.5 Hz
- L − C·RL² = 10⁻³ − 10⁻⁷ × 400 = 9.6 × 10⁻⁴; L − C·RC² = 9.9 × 10⁻⁴
- f_ar = f0 × √(9.6/9.9) = f0 × 0.9847 = 15672.5 Hz
- Check: Z at that frequency = 340 Ω, purely resistive (Im Z = 0)
- With RC = 0 it would be 15593.9 Hz: resistance in the C branch pushes f_ar back up
Answer: f_ar ≈ 15672 Hz, where the tank is a 340 Ω resistance
A single tuned circuit for the FM band · Illustrative: L = 0.1 µH, coil Q = 80 at 100 MHz; the tuned stage's transistor has 10 kΩ output resistance
- C = 1/(ω²L): 88 MHz needs 32.71 pF, 108 MHz needs 21.72 pF; 100 MHz needs 25.33 pF
- At 100 MHz, ωL = 62.83 Ω, so the coil's resistance is ωL/Q = 0.785 Ω
- Zdyn = L/(CR) = 5027 Ω (≈ Q·ωL), unloaded BW ≈ f0/Q = 1.25 MHz (the exact half-power band is also 1.25 MHz)
- Loaded by 10 kΩ: Rp = 5027 parallel 10000 = 3345 Ω, loaded Q = Rp/ωL = 53.2
- Loaded BW ≈ 100 MHz/53.2 = 1.88 MHz
Answer: C swings from 32.7 to 21.7 pF across the band; loading drops Q from 80 to 53 and widens the band to 1.88 MHz
First-order transients: source-free and driven RC and RL circuits
Which syllabus covers thisAU EC3251 Unit IV; EE3251 Unit IIIVTU BEC304 Module 3; BEE302 Module 3SPPU 203147 Unit 3MU ECC304 Module 3; EEC302 Module 2KTU PCEET302 Module 3
What is the time constant, and why does every first-order answer look alike?
A circuit with one capacitor or one inductor, plus resistors and DC sources, moves from its starting value to its final value along one exponential. The time constant is τ = RC or L/R, with R the Thevenin resistance the element sees. With 1 kΩ and 100 µF, τ = 100 ms. After one τ the change is 63.2 % done (12 V charging reaches 7.59 V); 99 % takes 4.61τ; after 5τ it is 99.3 % done, which engineers call settled.
What is a source-free (natural) response?
Remove the source and let the stored energy drain through the resistance. A capacitor charged to 12 V and discharging through 1 kΩ follows v = 12e^(−t/τ): half the voltage is gone after τ·ln 2 = 69.3 ms. Energy goes as v², so half the 7.2 mJ is gone after only 34.7 ms. An inductor does the same with its current: shorted through R, iL = I0·e^(−Rt/L), and the coil voltage reverses so the current keeps flowing the same way.
Where does the energy go when a capacitor charges through a resistor?
Charge 100 µF to 12 V. The source moves charge CV through a constant 12 V, so it delivers CV² = 14.4 mJ. The capacitor ends with ½CV² = 7.2 mJ. The other 7.2 mJ is heat in the resistor, whatever R is: a smaller R gives a bigger current for a shorter time and the same loss. Charging from a fixed voltage is never better than 50 % efficient, one reason switched supplies charge capacitors through inductors instead.
What does an RC circuit do with a square wave?
Each edge starts a new exponential from wherever the last one stopped. After a few cycles the capacitor voltage settles into a band. With 12 V and a period T = 2τ the band is 3.23 to 8.77 V, centred on the 6 V average. Make T = τ/5 and the ripple shrinks to 0.6 V: the RC averages. Make T = 20τ and vC follows the square wave almost fully. The Analog Electronics book uses these limits as the RC integrator and differentiator.
A 100 µF capacitor charges to 12 V through 1 kΩ and wastes 7.2 mJ in the resistor. You swap in a 10 kΩ resistor. How much energy is wasted now?
The loss is ½CV² = 7.2 mJ for any R. With 10 kΩ the current is ten times smaller but flows ten times longer (τ = 1 s), and i²R integrated over time comes out the same.
Worked examples
Charging an RC to a threshold · 12 V step, R = 4.7 kΩ, C = 22 µF, capacitor empty at t = 0. Find τ, vC at 50 ms, and when vC reaches 9 V.
- τ = RC = 4.7 kΩ × 22 µF = 103.4 ms; initial current 12 V/4.7 kΩ = 2.553 mA
- vC(t) = 12(1 − e^(−t/τ)); at 50 ms: 12(1 − e^(−0.4836)) = 4.601 V
- 9 V needs e^(−t/τ) = 3/12, so t = τ·ln 4 = 143.3 ms
Answer: τ = 103.4 ms, vC(50 ms) = 4.6 V, 9 V at 143 ms
An inductor energised through a divider (use Thevenin first) · 24 V source, R1 = 60 Ω in series, R2 = 40 Ω to ground; a 0.5 H coil is switched across R2 at t = 0.
- Thevenin seen by the coil: Vth = 24 × 40/100 = 9.6 V, Rth = 60 || 40 = 24 Ω
- τ = L/Rth = 0.5/24 = 20.83 ms; final current Vth/Rth = 0.4 A
- iL(10 ms) = 0.4(1 − e^(−10/20.83)) = 152.5 mA
- Energy finally stored: ½LI² = ½ × 0.5 × 0.4² = 40 mJ
Answer: τ = 20.8 ms, iL(10 ms) = 152 mA rising to 400 mA, 40 mJ stored
RC on a square wave: the steady band · 0/5 V square wave, period 1 ms, R = 1 kΩ, C = 1 µF, starting empty
- τ = 1 ms, half period h = 0.5 ms, a = e^(−h/τ) = 0.6065
- Steady band: Vlow = V·a/(1 + a) = 1.888 V, Vhigh = V/(1 + a) = 3.112 V
- Ripple = 1.225 V peak to peak; peak current at each on-edge (5 − 1.888)/1 kΩ = 3.11 mA
- The low point after cycle k is Vlow(1 − a^(2k)), within 1 % of Vlow once a^(2k) ≤ 0.01: k = 5 cycles
Answer: vC swings 1.89 to 3.11 V after about 5 cycles
Switching and initial conditions: L and C at t = 0−, 0+ and steady state
Which syllabus covers thisAU EC3251 Unit IVVTU BEC304 Module 3; BEE302 Module 3SPPU 203147 Unit 3MU ECC304 Module 3; EEC302 Module 2KTU PCEET302 Module 3
Which quantities cannot jump when a switch operates?
The inductor current and the capacitor voltage. Each holds energy (½Li², ½Cv²), and changing that energy in zero time would need infinite power. A 100 mH coil carrying 2 A holds 200 mJ; removing it in 1 µs would take 200 kW. So iL(0+) = iL(0−) and vC(0+) = vC(0−). Everything else may jump: the inductor voltage, the capacitor current and every resistor's current and voltage.
How do you find the values at 0−, 0+ and ∞?
At 0− the circuit has sat for a long time with DC, so the inductor is a short and the capacitor an open circuit: solve a resistive circuit. At 0+ replace each inductor by a current source iL(0−) and each capacitor by a voltage source vC(0−), set the switches to their new state and solve again. At ∞ it is DC once more. In the first preset, 24 V through 12 Ω gives iL = 2 A; at 0+ that 2 A is forced through R2 = 100 Ω, so vL(0+) = −200 V.
How do you get the slopes di/dt and dv/dt at 0+?
From the element laws: diL/dt = vL/L and dvC/dt = iC/C, using the 0+ values. In the RLC preset (12 V, R = 20 Ω, L = 100 mH, C = 10 µF), iL and vC both start at 0, so vL(0+) = 12 V and diL/dt = 120 A/s, while dvC/dt(0+) = 0. For a second derivative, differentiate the circuit equation once: here C·d²vC/dt² = diL/dt at 0+, which gives 12 × 10⁶ V/s². These slopes fix the constants of the second-order solution.
Why does opening a switch in series with a coil cause a spark?
The coil insists on keeping its current. With 2 A flowing and only a 1 kΩ path left, it drives 2000 V across that path, and the open switch sees the source plus this. With no path at all the voltage climbs until the air gap at the contacts breaks down and arcs. The fix is a path: a freewheeling diode across the coil clamps the reversal to about a volt (illustrative), and an RC snubber or a resistor limits it to iL·R, with τ = L/R (0.1 ms for 1 kΩ).
In the RC changeover circuit, vC(0−) = 8.25 V. At t = 0 the switch moves to R2 = 1 kΩ. What is vC just after?
A capacitor voltage cannot jump: vC(0+) = 8.25 V. What jumps is the current: iC goes from 0 to −8.25 V × (1/1 kΩ + 1/2.2 kΩ) = −12 mA.
Worked examples
The inductive kick when a switch opens · 48 V source, R1 = 16 Ω, then a switch, then a 0.4 H coil with R2 = 220 Ω across it. The switch has been closed a long time and opens at t = 0.
- 0−: the coil is a short, so iL = 48/16 = 3 A and R2 carries nothing
- 0+: iL = 3 A must flow through R2: vL(0+) = −3 × 220 = −660 V
- The open switch sees 48 − (−660) = 708 V; diL/dt = vL/L = −1650 A/s
- τ = L/R2 = 1.818 ms; iL(1 ms) = 3e^(−1/1.818) = 1.731 A; iL(∞) = 0
Answer: vL jumps from 0 to −660 V; the switch sees 708 V; the current decays with τ = 1.82 ms
A changeover switch on a capacitor · V1 = 20 V, R1 = 2 kΩ to contact A, R3 = 8 kΩ across C = 47 µF; at t = 0 the switch moves to contact B, R2 = 4 kΩ to ground.
- 0−: C open, so vC = 20 × 8/(2 + 8) = 16 V and iC = 0
- 0+: vC = 16 V (no jump); R2 carries 16/4 kΩ = 4 mA and R3 2 mA
- iC(0+) = −(4 + 2) = −6 mA; dvC/dt = iC/C = −127.7 V/s
- ∞: vC = 0; τ = C(R2 || R3) = 47 µF × 2.667 kΩ = 125.3 ms
Answer: vC(0+) = 16 V, iC(0+) = −6 mA, dvC/dt(0+) = −128 V/s, vC(∞) = 0
Series RLC switched onto a precharged capacitor · 10 V, R = 50 Ω, L = 0.2 H, C = 20 µF charged to 4 V (+ at the top); the switch closes at t = 0, no current before.
- 0+: i = 0 (coil), vC = 4 V (capacitor), so vL = 10 − 50 × 0 − 4 = 6 V
- di/dt(0+) = vL/L = 6/0.2 = 30 A/s; dvC/dt(0+) = i/C = 0
- Differentiate L di/dt = V − Ri − vC: L·d²i/dt² = −R·di/dt − dvC/dt = −50 × 30 − 0
- d²i/dt²(0+) = −7500 A/s²; final values: i(∞) = 0, vC(∞) = 10 V
Answer: i(0+) = 0, di/dt = 30 A/s, d²i/dt² = −7500 A/s², vC goes 4 → 10 V
Second-order RLC transients: overdamped, critically damped and underdamped
Which syllabus covers thisAU EC3251 Unit IV; EE3251 Unit IIIVTU BEC304 Module 3SPPU 203147 Unit 3MU ECC304 Module 3; EEC302 Module 2KTU PCEET302 Module 3
Where do α and ω0 come from?
Write KVL round a series RLC loop and differentiate once: L·i″ + R·i′ + i/C = 0. Try i = e^(st) and you get s² + (R/L)s + 1/(LC) = 0, written s² + 2αs + ω0² with α = R/2L (how fast energy is lost) and ω0 = 1/√(LC) (how fast it swaps between L and C). With 10 mH and 10 µF, ω0 = 3162 rad/s (503 Hz). The two roots decide the shape of every voltage and current.
What are the overdamped, critically damped and underdamped cases?
If α > ω0 the roots are real and different: two decaying exponentials, no overshoot (overdamped). If α = ω0 they meet at −ω0: the fastest response with no overshoot (critical). If α is below ω0 they become a complex pair −α ± jωd, ωd = √(ω0² − α²): a ringing response (underdamped). For the series circuit the boundary is R = 2√(L/C) = 63.2 Ω with 10 mH and 10 µF. As R falls the roots slide along the real axis, meet, then ride round the circle |s| = ω0.
How much does an underdamped circuit overshoot?
For a series RLC with a step, the capacitor voltage peaks at t = π/ωd and overshoots by e^(−απ/ωd). With R = 20 Ω, 10 mH and 10 µF: α = 1000 /s, ωd = 3000 rad/s, ζ = α/ω0 = 0.316, so a 10 V step peaks at 13.51 V (35.1 % overshoot) after 1.047 ms, then rings at 477 Hz while the envelope shrinks as e^(−αt). The Control Systems book treats the same maths for any second-order system.
Why does more R damp a series circuit but not a parallel one?
In series every bit of current passes through R, so a bigger R burns the swinging energy faster: α = R/2L. In parallel R sits across L and C; a bigger R takes less current and wastes less, so the tank rings longer: α = 1/(2RC). They are duals. With 10 mH and 10 µF the series circuit is critically damped at 63.2 Ω, the parallel one at 15.8 Ω, and the product of the two is L/C = 1000 Ω².
A series RLC (10 mH, 10 µF, R = 20 Ω) overshoots a 10 V step by 35 %. You change R to 100 Ω. What happens?
Critical damping is R = 2√(L/C) = 63.2 Ω. At 100 Ω, α = 5,000 /s exceeds ω0 = 3162 rad/s: two real roots, a slow rise to 10 V with no overshoot.
Worked examples
An overdamped series RLC · 24 V step, R = 1 kΩ, L = 0.1 H, C = 1 µF, at rest before t = 0
- α = R/2L = 5000 /s, ω0 = 1/√(LC) = 3162.3 rad/s; α > ω0 so overdamped (Rcrit = 2√(L/C) = 632.5 Ω)
- Roots: s = −5000 ± √(5000² − 3162.3²) = −1127 and −8873 /s
- i(t) = V/(L(s1 − s2))·(e^(s1·t) − e^(s2·t)), peak at t = ln(s2/s1)/(s1 − s2) = 0.2664 ms: i = 20.03 mA
- vC(1 ms) = 24[1 + (s2·e^(s1·t) − s1·e^(s2·t))/(s1 − s2)] = 15.093 V
Answer: Roots −1127 and −8873 /s; peak current 20 mA at 0.266 ms; vC(1 ms) = 15.09 V
Overshoot of an underdamped series RLC · 12 V step, R = 40 Ω, L = 25 mH, C = 4.7 µF
- α = 40/(2 × 0.025) = 800 /s, ω0 = 2917.3 rad/s: underdamped (Rcrit = 145.9 Ω)
- ωd = √(ω0² − α²) = 2805.5 rad/s, ringing at 447 Hz
- Overshoot = e^(−απ/ωd) = 40.8 %, at t = π/ωd = 1.12 ms
- Peak vC = 12 × (1 + 0.4083) = 16.9 V
Answer: vC peaks at 16.9 V (40.8 % overshoot) after 1.12 ms; R ≥ 146 Ω would remove it
A parallel tank hit by a current step · 50 mA step into R = 200 Ω, L = 20 mH and C = 5 µF in parallel, all at rest
- α = 1/(2RC) = 1/(2 × 200 × 5 µF) = 500 /s, ω0 = 3162.3 rad/s: underdamped
- ωd = 3122.5 rad/s; the inductor current overshoots by e^(−απ/ωd) = 60.5 %
- Peak iL = 50 × (1 + 0.6047) = 80.2 mA
- Critical damping in parallel needs R = ½√(L/C) = 31.6 Ω, a SMALLER resistor
Answer: iL peaks at 80.2 mA; R = 31.6 Ω would make it critically damped
Transients with sinusoidal sources: switching angle and the DC offset
Which syllabus covers thisAU EE3251 Unit IIIVTU BEC304 Module 3SPPU not in syllabusMU not in syllabusKTU PCEET302 Module 3
What happens when an RL circuit is switched onto a sine wave?
The current is the steady-state sine you would get from phasors plus a natural term Ae^(−Rt/L). The natural term exists only to make i(0+) = 0, because the coil current was zero just before. On 230 V, 50 Hz (Vm = 325.3 V) with R = 5 Ω and L = 50 mH, |Z| = 16.48 Ω, the current lags by φ = 72.3° and its steady peak is 19.73 A. The offset decays with τ = L/R = 10 ms.
Which switching angle gives no transient, and which the worst?
The offset at 0+ is −Im·sin(θ − φ). Close the switch at θ = φ and the steady sine already starts at zero: no transient at all. Close it at θ = φ ± 90° and the offset is a full Im. In the 5 Ω, 50 mH circuit, switching at the voltage zero (θ = 0) gives an offset of 18.8 A and a first peak of 27.5 A, 1.39 times the steady peak. At the largest-offset angle, 162.3°, the peak is 27.1 A: slightly less, because it comes later and the offset has decayed more. For any R and L the largest first peak is at the voltage zero. Longer τ pushes it towards 2×.
What changes when a capacitor is switched onto AC?
The capacitor voltage follows the same law with φ = atan(ωRC), so it too can carry a decaying offset. The current, though, is not held: at 0+ the empty capacitor is a short and i jumps to v(0)/R. Switching 100 µF through 100 Ω at the voltage peak gives 3.25 A at once against a steady peak of 3.1 A; with only 1 Ω in the path it would be 325 A. That inrush is why capacitor banks are switched near voltage zero or through series reactors.
Where does the DC offset matter in practice?
In short circuits. A fault path is mostly inductive, and a fault can strike at any point of the wave. Take X/R = 10 (illustrative): τ = 31.8 ms, so the offset keeps 53 % of itself after each 20 ms cycle. At the worst angle the first peak reaches 1.73 times the symmetrical peak. Breakers and busbars must survive that asymmetrical first peak, not just the steady fault current. The Power Systems book covers fault levels.
An RL load has φ = 72° (current lags by 72°). The switch happens to close at θ = 72° on the voltage wave. What does the current do?
The offset at 0+ is −Im·sin(θ − φ) = −Im·sin 0 = 0. The steady current at that instant is already zero, so the coil's zero current needs no correction term.
Worked examples
An RL load switched at the voltage zero · 230 V, 50 Hz (Vm = 325.3 V), R = 8 Ω, L = 60 mH, switch closes at θ = 0
- X = ωL = 18.85 Ω, |Z| = 20.48 Ω, φ = atan(X/R) = 67°
- Im = 325.3/20.48 = 15.88 A; τ = L/R = 7.5 ms
- Offset at 0+ = −Im·sin(0 − 67°) = 14.62 A
- At t = 10 ms: i = 15.88·sin(180° − 67°) + 14.62·e^(−10/7.5) = 18.48 A
- First peak (search over the first cycle): 20.58 A at 8.31 ms
Answer: Steady 15.9 A peak, offset 14.6 A, first peak 20.6 A (1.3×)
The worst switching angle for the same load · The 8 Ω, 60 mH load above
- Worst θ = φ ± 90° = 157° (or 337°): the offset is a full ±Im = −15.88 A
- First peak: −20.18 A at 9.62 ms, 1.27 times the steady peak
- A scan of θ finds the largest first peak at θ = 0° (the voltage zero), 20.58 A: that peak arrives sooner, so less offset has decayed
- Setting d(peak)/dθ = 0 alongside di/dt = 0 gives tan(θ − φ) = −tan φ, so θ = 0 for any R and L
- The offset falls to 5 % after τ·ln 20 = 22.5 ms, about 1.1 cycles
- No-transient angle: θ = φ = 67°
Answer: Close at 67° for no offset; the full offset comes at 157°, the largest first peak (20.6 A) at the voltage zero
A capacitor switched at the voltage peak · 230 V, 50 Hz, R = 10 Ω in series with C = 200 µF, capacitor empty, switch closes at θ = 90°
- Xc = 1/(ωC) = 15.92 Ω, |Z| = 18.8 Ω; vC lags the source by φ = atan(ωRC) = 32.1°
- Steady vC peak = Vm·Xc/|Z| = 275.4 V; steady current peak = Vm/|Z| = 17.3 A
- At 0+ the empty capacitor is a short: i(0+) = 325.3/10 = 32.53 A
- vC offset at 0+ = −275.4·sin(90° − 32.1°) = −233.2 V, decaying with τ = RC = 2 ms
- First vC peak: −276.1 V at 11.77 ms
Answer: Inrush 32.5 A against a steady 17.3 A peak; vC first peak −276 V vs 275 V steady
The Laplace transform in circuits: s-domain models with initial conditions
Which syllabus covers thisAU EE3251 Unit IIIVTU BEC304 Module 4; BEE302 Module 4SPPU 203147 Unit 4MU EEC302 Module 4KTU PCEET302 Module 3
Why solve a circuit in the s-domain at all?
Because derivatives become multiplication. Switch a 10 V step into a loop of 100 Ω, 100 mH and 10 µF and Kirchhoff gives a second-order differential equation. Transform it and L di/dt becomes sL·I(s) − L·i(0−), the capacitor's integral becomes I(s)/(sC) + v(0−)/s, and the loop equation is plain algebra: solve for I(s), then turn it back into i(t). The initial conditions ride along inside the algebra, so you never fit constants afterwards. The Signals and Systems book covers the transform itself and its region of convergence; this chapter is about circuits.
What do R, L and C look like in the s-domain?
R stays R. An inductor becomes an impedance sL plus a source for its initial current: either a voltage source L·i(0−) in series or a current source i(0−)/s in parallel. A capacitor becomes 1/(sC) plus either a voltage source v(0−)/s in series or a current source C·v(0−) in parallel. The two forms are source transformations of each other, so pick the one that suits mesh (series) or nodal (parallel) analysis. With iL(0−) = 50 mA and L = 100 mH the series source is 5 mV·s; with vC(0−) = 4 V and C = 10 µF the parallel one is 40 µA·s.
How do you get i(t) back from I(s)?
Split I(s) into partial fractions and invert each term. A simple pole p gives k·e^(pt), with the residue k = (s − p)I(s) evaluated at s = p (the cover-up rule). A complex pair σ ± jω gives one damped cosine 2|k|e^(σt)cos(ωt + ∠k), so you only compute one residue. A repeated pole gives a t·e^(pt) term. For the 100 Ω loop the poles are −500 ± j866 /s, the residue is 0.0321 A at −38.9°, so i(t) rings at 138 Hz while it decays with a 2 ms time constant.
Do mesh and nodal analysis still work in the s-domain?
Yes, unchanged. Every element is an impedance Z(s) (R, sL, 1/sC) or an admittance, the initial conditions are extra sources, and KVL and KCL are written exactly as for phasors with s in place of jω. A mesh with three meshes still gives a 3 × 3 matrix, only now each entry is a polynomial in s. With all initial conditions zero, the ratio of output to input transforms is the network function of chapter 23, and setting s = jω gives the steady-state AC answer of chapter 10.
A series RLC loop has iL(0−) = 50 mA when a step is applied. You draw the s-domain circuit but forget the L·i(0−) source. What goes wrong?
Without the source, I(s) = 60/(s² + 1000s + 10⁶) and i(0+) = 0; the inductor's current cannot jump, so it must start at 50 mA. At 0.5 ms the wrong answer is 22.6 mA against the correct 48.6 mA.
Worked examples
Step into a charged series RLC loop · V = 10 V step at t = 0, R = 100 Ω, L = 100 mH, C = 10 µF, iL(0−) = 50 mA, vC(0−) = 4 V (illustrative)
- KVL in s: 10/s = 100·I + 0.1s·I − 0.1 × 0.05 + I/(10 µF·s) + 4/s
- I(s) = (0.005s + 6)/(0.1s² + 100s + 10⁵) = (0.05s + 60)/(s² + 1000s + 10⁶)
- Poles: s = −500 ± j866 /s (underdamped: R < 2√(L/C) = 200 Ω)
- Residue at the upper pole: (0.05p + 60)/(p − p*) = 0.025 − j0.0202 = 0.0321∠−38.9°
- i(t) = 2 × 0.0321 e^(−500t) cos(866t − 38.9°) A; check: i(0) = 50 mA
- At t = 1 ms: i = 38.32 mA (an RK4 solution of the loop equation agrees)
Answer: i(t) = 0.0643 e^(−500t) cos(866t − 38.9°) A, starting at 50 mA and ending at 0
An inductor already carrying more than its final current · 24 V step into R = 50 Ω in series with L = 0.5 H; iL(0−) = 1 A, left over from an earlier circuit state
- Series model: (24/s + 0.5 × 1) = (50 + 0.5s)·I(s), so I(s) = (s + 48)/(s(s + 100))
- Parallel model check: the L branch carries V_L/(sL) + 1/s, and KVL gives the same I(s)
- Cover-up: at s = 0, k = 48/100 = 0.48 A; at s = −100, k = (−100 + 48)/(−100) = 0.52 A
- i(t) = 0.48 + 0.52e^(−100t) A: starts at 1 A and falls to V/R = 0.48 A
- It reaches 0.5 A when 0.52e^(−100t) = 0.02: t = ln(26)/100 = 32.6 ms
Answer: i(t) = 0.48 + 0.52e^(−100t) A; 0.5 A after 32.6 ms
Nodal analysis with the capacitor's current-source model · C = 47 µF charged to 9 V is connected at t = 0 through R = 1 kΩ to a 5 V source
- Parallel model: 1/(sC) with a current source C·v(0−) = 47 µF × 9 V = 423 µA·s into the top node
- KCL at the capacitor node: (Vc − 5/s)/1000 + sC·Vc = C × 9
- Vc(s) = 5/s + 4/(s + 1/τ), τ = RC = 47 ms
- vC(t) = 5 + 4e^(−t/0.047) V; i(t) = (5 − vC)/R = −4e^(−t/τ) mA (the capacitor discharges into the source)
- At t = τ: vC = 5 + 4/e = 6.47 V
Answer: vC falls from 9 V to 5 V with τ = 47 ms; 6.47 V at t = 47 ms
Standard signals, waveform synthesis and the initial and final value theorems
Which syllabus covers thisAU EE3251 Unit IIIVTU BEC304 Module 4; BEE302 Module 4SPPU 203147 Unit 4MU EEC302 Module 4KTU not in syllabus
Which standard signals do circuit tests use, and what are their transforms?
Four, each the integral of the one before. The unit impulse δ(t), a kick of zero width and unit area, transforms to 1. The unit step u(t), a switch closing, transforms to 1/s. The unit ramp r(t) = t·u(t), a steadily rising voltage, transforms to 1/s². The parabola t²/2·u(t) transforms to 1/s³. So a 12 V step is 12/s and a 3 V/s ramp is 3/s². Because each is the integral of the previous one, the step response of a circuit is the integral of its impulse response.
How do you build an odd-shaped waveform from steps and ramps?
Switch pieces on at the right moments. A 5 V pulse lasting 2 s is 5u(t) − 5u(t − 2): one step on, an equal step off. The time-shift theorem says a piece delayed by a seconds picks up e^(−as) in its transform, so this pulse is 5(1 − e^(−2s))/s, which is 4.3233 at s = 1. Corners are slope changes: a trapezoid rising at 4 V/s for 1 s, flat until 3 s and back to 0 at 4 s is 4r(t) − 4r(t − 1) − 4r(t − 3) + 4r(t − 4).
How is a repeating waveform transformed?
Transform one period, F1(s), and divide by 1 − e^(−Ts), because the waveform is that period repeated every T seconds and the shifts form a geometric series. A 5 V square wave, high for 1 s in every 2 s, has F1(s) = 5(1 − e^(−s))/s, so F(s) = 5/(s(1 + e^(−s))). At s = 1 that is 3.6553, matching a numerical integral over 20 periods. Near s = 0, sF(s) tends to 2.5 V: the average value, not a final value, because the waveform never settles.
What do the initial and final value theorems give, and when do they fail?
f(0+) = lim sF(s) as s → ∞, and f(∞) = lim sF(s) as s → 0, without inverting anything. For F(s) = (2s² + 8s + 6)/(s(s² + 2s + 5)) they give f(0+) = 2 and f(∞) = 6/5 = 1.2, and the full inverse agrees. The final value theorem only holds when every pole of sF(s) lies in the left half plane. Flip the poles to 1 ± j2 and the formula still says 1.2, but f(t) swings ever wider, like e^t: at t = 5 s it is already −551.
What does convolution have to do with circuits?
A linear circuit's response to any input is its impulse response h(t) convolved with the input: y(t) = ∫ h(u)x(t − u) du. In the s-domain that integral becomes a product, Y(s) = H(s)X(s), which is why transforms are easier. For an RC low-pass with τ = 0.5 s, h(t) = 2e^(−2t), and a 5 V pulse lasting 2 s gives y = 5(1 − e^(−4)) = 4.908 V at t = 2 s and 0.664 V one second later, both from the convolution integral and from Y(s).
F(s) = 10/(s(s − 2)). The final value theorem gives lim sF(s) = −5 as s → 0. Is f(∞) = −5?
sF(s) = 10/(s − 2) has a pole in the right half plane, so the theorem does not apply. Partial fractions give f(t) = −5 + 5e^(2t): it starts at 0 and runs away.
Worked examples
Transform of a trapezoidal pulse · f(t) rises from 0 to 4 V in 1 s, stays at 4 V until t = 3 s, falls to 0 at t = 4 s
- Slopes: +4 V/s from 0, 0 from 1 s, −4 V/s from 3 s, 0 from 4 s
- So f(t) = 4r(t) − 4r(t − 1) − 4r(t − 3) + 4r(t − 4)
- Each ramp k·r(t − a) transforms to k·e^(−as)/s²
- F(s) = 4(1 − e^(−s) − e^(−3s) + e^(−4s))/s²
- Check at s = 0.5: formula 4.8908, numerical integral of f(t)e^(−0.5t) 4.8908
- As s → 0, F(s) → the area under f(t) = 4 × (4 + 2)/2 = 12 V·s
Answer: F(s) = 4(1 − e^(−s) − e^(−3s) + e^(−4s))/s²; F(0) = 12 V·s
A square wave as a periodic function · v(t) = 5 V for 0 ≤ t < 1 s and 0 for 1 ≤ t < 2 s, repeating every 2 s
- One period: 5u(t) − 5u(t − 1), so F1(s) = 5(1 − e^(−s))/s
- Periodic rule: F(s) = F1(s)/(1 − e^(−2s))
- 1 − e^(−2s) = (1 − e^(−s))(1 + e^(−s)), so F(s) = 5/(s(1 + e^(−s)))
- At s = 1: F = 5/(1 + e^(−1)) = 3.6553
- sF(s) → 2.5 as s → 0: the average; the final value theorem does not apply because v(t) keeps switching
Answer: F(s) = 5/(s(1 + e^(−s))); F(1) = 3.655
Initial and final values without inverting · F(s) = (2s² + 8s + 6)/(s(s² + 2s + 5)), then the same numerator over s(s² − 2s + 5)
- IVT: sF(s) = (2s² + 8s + 6)/(s² + 2s + 5) → 2 as s → ∞, so f(0+) = 2
- Poles of sF(s): −1 ± j2, both in the left half plane, so the FVT applies
- FVT: sF(s) at s = 0 is 6/5 = 1.2, so f(∞) = 1.2
- Check by partial fractions: f(t) = 1.2 + 2Re(k·e^((−1 + j2)t)) with k = 0.4 − j1.2; f(0) = 2, f(30 s) = 1.2
- Second F(s): the same limit gives 1.2, but its poles 1 ± j2 are in the right half plane; f(5 s) = −551
Answer: f(0+) = 2 and f(∞) = 1.2 for the first; the second has no final value
Network functions: driving point and transfer functions, poles, zeros and stability
Which syllabus covers thisAU not in syllabusVTU not in syllabusSPPU 203147 Unit 6MU ECC304 Module 4; EEC302 Module 6KTU not in syllabus
What is the difference between a driving-point function and a transfer function?
Both are ratios of transforms with every initial condition zero. A driving-point function looks into one port: Z(s) = V(s)/I(s) or Y(s) = 1/Z(s), measured where you drive. A transfer function links two different ports: a voltage ratio V2/V1, a current ratio, a transfer impedance V2/I1 or transfer admittance I2/V1. A two-section RC ladder with 1 kΩ and 1 µF in each section has V2/V1 = 1/(10⁻⁶s² + 0.003s + 1), with poles at −382 and −2618 rad/s.
What do the poles and zeros tell you about the circuit?
Factor the function as K(s − z1)(s − z2).../((s − p1)(s − p2)...). Zeros are the complex frequencies where the function is zero; poles are the network's natural modes, the shapes its impulse response is built from. A pole at −a gives e^(−at); a pair σ ± jω gives e^(σt)cos(ωt). Further left means faster decay, further from the real axis means faster ringing. R1 = R2 = 100 Ω with L = 10 mH and C = 1 µF puts poles at −5000 ± j8660 /s: a 0.2 ms decay with 1378 Hz ringing.
What must a passive driving-point function look like?
If Z(s) comes from real R, L, C and transformers, then: all polynomial coefficients are real and positive; no power of s is missing between the highest and lowest unless the polynomial is purely even or purely odd; the degrees of numerator and denominator differ by at most 1, and so do their lowest powers; poles and zeros lie in the left half plane, and any on the jω axis are simple. Transfer functions are freer: their zeros may sit in the right half plane, as in a lattice network.
How do you find a residue graphically from the pole-zero plot?
Draw vectors to the chosen pole from every zero and from every other pole. The residue is K times the product of the zero vectors divided by the product of the pole vectors, lengths multiplying and angles adding. For H(s) = 12(s + 3)/((s + 1)(s + 2)(s + 6)) at s = −1, the zero vector is 2, the pole vectors are 1 and 5, so k = 12 × 2/5 = 4.8. The other residues are −3 and −1.8, so h(t) = 4.8e^(−t) − 3e^(−2t) − 1.8e^(−6t).
When is a network stable, and how do you test it without finding roots?
A network is stable when every pole is in the open left half plane, so the denominator must be a Hurwitz polynomial. The Routh test tells you without factoring: the number of sign changes in its first column equals the number of right-half-plane roots. Passive RLC networks never fail it (lossless LC poles at worst sit on the jω axis). Add a dependent source and they may not: a parallel tank with R = 1 kΩ whose transconductance gm feeds current back in has poles in the right half plane once gm exceeds 1/R = 1 mS, and at exactly 1 mS it oscillates at 1592 Hz.
Someone claims Z(s) = (s² − 2s + 5)/(s² + 3s + 2) is the input impedance of a passive RLC network. Is it possible?
A passive driving-point function has positive coefficients and no right-half-plane poles or zeros. s² − 2s + 5 has roots 1 ± j2, so no passive network can have this impedance.
Worked examples
Driving-point impedance of an R-L-C one-port · R1 = 100 Ω in series with [L = 10 mH in parallel with (R2 = 100 Ω + C = 1 µF)]
- Z = R1 + sL(R2 + 1/sC)/(sL + R2 + 1/sC); multiply through by sC
- Z = ((R1 + R2)LC s² + (R1R2C + L)s + R1)/(LC s² + R2C s + 1)
- Numbers: Z = (2×10⁻⁶ s² + 0.02s + 100)/(10⁻⁸ s² + 10⁻⁴ s + 1) = 200(s² + 10⁴s + 5×10⁷)/(s² + 10⁴s + 10⁸)
- Zeros: −5000 ± j5000 /s; poles: −5000 ± j8660 /s
- Checks: Z(0) = 100 Ω (L shorts the branch), Z(∞) = 200 Ω (C shorts, L opens): R1 then R1 + R2
Answer: Z(s) = 200(s² + 10⁴s + 5×10⁷)/(s² + 10⁴s + 10⁸): every pole and zero in the left half plane
Residues from the pole-zero plot · H(s) = 12(s + 3)/((s + 1)(s + 2)(s + 6))
- At p = −1: zero vector (−1) − (−3) = 2; pole vectors (−1) − (−2) = 1 and (−1) − (−6) = 5; k = 12 × 2/(1 × 5) = 4.8
- At p = −2: zero vector 1; pole vectors −1 and 4; k = 12 × 1/(−1 × 4) = −3
- At p = −6: zero vector −3; pole vectors −5 and −4; k = 12 × (−3)/20 = −1.8
- Sum 4.8 − 3 − 1.8 = 0, as it must be when the denominator degree exceeds the numerator's by 2 (h(0+) = 0)
- h(t) = 4.8e^(−t) − 3e^(−2t) − 1.8e^(−6t); h(1 s) = 1.3554
Answer: Residues 4.8, −3 and −1.8
Routh test on a third-order loop · Characteristic polynomial s³ + 5s² + 8s + K, K > 0 a gain
- Routh rows: [1, 8], [5, K], [(40 − K)/5, 0], [K]
- First column 1, 5, (40 − K)/5, K: all positive when 0 < K < 40
- At K = 40 the third row is zero; the auxiliary polynomial 5s² + 40 = 0 gives s = ± j√8
- So at K = 40 the loop oscillates at √8 = 2.828 rad/s; for K = 41 the first column changes sign twice: two right-half-plane poles
Answer: Stable for 0 < K < 40; oscillates at 2.83 rad/s at K = 40
Network topology: graphs, trees, incidence, tie-set and cut-set matrices
Which syllabus covers thisAU EC3251 Unit VVTU not in syllabusSPPU 203147 Unit 2MU ECC304 Module 2; EEC302 Module 3KTU not in syllabus
What are a graph, a tree, twigs and links?
Replace every element of a network by a line and keep the junctions: that is its graph, with n nodes and b branches, each given a reference direction. A tree is a set of branches that touches every node and closes no loop; it always has n − 1 branches, called twigs. The other b − n + 1 branches are links, the co-tree. Every link closes exactly one loop with the tree. A Wheatstone bridge has 4 nodes and 6 branches: 3 twigs, 3 links and 16 different trees.
What is the incidence matrix, and how does it count the trees?
The incidence matrix has a row per node and a column per branch: +1 where a branch leaves the node, −1 where it enters, 0 elsewhere. Drop the reference node's row to get the reduced matrix A, of size (n − 1) × b. KCL for the whole network is then one line: A·i = 0. Kirchhoff's matrix-tree theorem counts the trees as det(A·Aᵀ): 8 for a square with one diagonal, 16 for the bridge and 45 for a square with a centre node joined to all four corners.
What is a fundamental tie-set matrix?
Pick a tree. Each link with its unique tree path forms a fundamental loop, oriented the same way as the link. The tie-set matrix B has one row per loop and one column per branch: +1 if the branch runs with the loop, −1 against, 0 if absent. KVL is B·v = 0, and every branch current is a combination of the link currents: i = Bᵀ·IL. With the twigs listed first, B = [Bt | I]: the link columns form an identity matrix. A bridge with tree {1, 2, 3} has three loops, one each for links 4, 5 and 6.
What is a fundamental cut-set matrix?
Remove one twig and the tree falls into two parts; the twig plus every link that joins the two parts is a fundamental cut-set, oriented with the twig. The cut-set matrix Q has one row per twig: +1 for a branch crossing the cut in the twig's direction, −1 against. KCL is Q·i = 0, and every branch voltage follows from the twig voltages: v = Qᵀ·vt. With twigs first, Q = [I | Ql], and Ql = −Btᵀ ties it to the tie-set matrix. Each row of A is also a cut-set, around one node.
How do these matrices give loop and node equations?
For branches made of a source E in series with a resistance R (v = E + R·i), loop analysis is B·R·Bᵀ·IL = −B·E: b − n + 1 equations in the link currents. Node analysis is A·G·Aᵀ·e = A·G·E: n − 1 equations in the node voltages. Choose whichever is smaller. On the bridge with a 12 V, 10 Ω source and arms of 100, 200, 300 and 150 Ω with 50 Ω across, both give a source current of 66.29 mA. For planar graphs, duality swaps loops with nodes and series with parallel.
A network graph has 6 nodes and 10 branches. How many independent KVL (loop) equations does it have?
The number of links is b − n + 1 = 10 − 6 + 1 = 5, and each link closes one independent loop. KCL gives the other n − 1 = 5 equations, and 5 + 5 = 10 matches the 10 branch currents.
Worked examples
Counting the trees of a square with a diagonal · Nodes 0, n1, n2, n3; branches 1: 0→n1, 2: n1→n2, 3: n2→n3, 4: n3→0, 5: n1→n3
- A tree needs n − 1 = 3 of the 5 branches: C(5, 3) = 10 candidate sets
- Two of them are triangles and close a loop: {1, 4, 5} and {2, 3, 5}
- So there are 10 − 2 = 8 trees
- Matrix-tree theorem: A rows n1 = [−1 1 0 0 1], n2 = [0 −1 1 0 0], n3 = [0 0 −1 1 −1]
- A·Aᵀ = [[3, −1, −1], [−1, 2, −1], [−1, −1, 3]]; det = 3(6 − 1) + 1(−3 − 1) − 1(1 + 2) = 8
Answer: 8 trees, by enumeration and by det(A·Aᵀ)
Loop analysis of a bridge with a tie-set matrix · Bridge graph: 1: n1→n2 (100 Ω), 2: n1→n3 (200 Ω), 3: n2→0 (300 Ω), 4: n3→0 (150 Ω), 5: n2→n3 (50 Ω), 6: 0→n1 (12 V source, + at n1, 10 Ω); tree {1, 2, 3}
- Loops of links 4, 5, 6: B rows [−1 1 −1 1 0 0], [1 −1 0 0 1 0], [1 0 1 0 0 1]
- B·R·Bᵀ = [[750, −300, −400], [−300, 350, 100], [−400, 100, 410]] Ω
- Right side −B·E = [0, 0, 12] V (the source sits in link 6)
- Link currents: I4 = 42.27 mA, I5 = 17.29 mA, I6 = 66.29 mA
- Twig currents from i = Bᵀ·IL: i1 = 41.31 mA, i2 = 24.98 mA, i3 = 24.02 mA; node analysis gives the same
Answer: Source current 66.29 mA; 17.29 mA flows in the 50 Ω branch from n2 to n3 (the bridge is unbalanced: 100 × 150 ≠ 200 × 300)
Cut-set matrix and branch voltages from twig voltages · The same bridge and tree {1, 2, 3}
- Remove twig 1 (n1→n2): the tree splits into {n1, n3} and {n2, 0}; branches 1, 4, 5, 6 cross
- Rows (twigs 1, 2, 3): [1 0 0 1 −1 −1], [0 1 0 −1 1 0], [0 0 1 1 0 −1]
- Check: the link columns equal −Btᵀ from the tie-set matrix, and Q·Bᵀ = 0
- Twig voltages from the solution: v1 = 4.131 V, v2 = 4.996 V, v3 = 7.206 V
- Links from v = Qᵀ·vt: v4 = v1 − v2 + v3 = 6.341 V, v5 = −v1 + v2 = 0.865 V, v6 = −v1 − v3 = −11.337 V
Answer: v4 = 6.34 V, v5 = 0.865 V, v6 = −11.34 V (so the source's terminal voltage is 11.34 V)
Two-port networks: Z, Y, h and ABCD parameters, reciprocity and symmetry
Which syllabus covers thisAU not in syllabusVTU BEC304 Module 5; BEE302 Module 5SPPU 203147 Unit 5MU ECC304 Module 5; EEC302 Module 5KTU PCEET302 Module 4
What is a two-port, and why is it described by four numbers?
A two-port is any network you reach only through two pairs of terminals: an input port and an output port, such as an amplifier, a filter or a transmission line. Four quantities live at the ports: V1, I1, V2 and I2, both currents flowing in at the top terminal. For a linear network, any two of them fix the other two, so four coefficients describe the whole box. A T of 20, 40 and 60 Ω has Z = [80, 60; 60, 100] Ω, and that matrix replaces the schematic.
How are the Z and Y parameters measured?
Z parameters come from open-circuit tests: drive a current into one port, leave the other open, and divide the port voltages by that current. Z11 = V1/I1 with I2 = 0 is the input resistance with the output open: 20 + 60 = 80 Ω for our T. Y parameters come from short-circuit tests: apply a voltage to one port, short the other, divide the currents by that voltage. Y11 = 1/(20 + 40 || 60) = 22.73 mS. Y is the inverse matrix of Z whenever both exist.
Where do the h and ABCD parameters come from?
Hybrid (h) parameters mix the two tests: h11 and h21 with the output shorted, h12 and h22 with the input open. They suit transistors, whose input is low-impedance and output high: h21 with the output shorted is the current gain (hfe in a datasheet). ABCD (transmission) parameters relate the input to the output, V1 = A·V2 − B·I2 and I1 = C·V2 − D·I2. Cascaded stages simply multiply their ABCD matrices, and a load ZL gives Zin = (A·ZL + B)/(C·ZL + D): 62 Ω for our T with 100 Ω.
What do reciprocity and symmetry mean for a two-port?
A network of R, L, C and coupled coils is reciprocal: the transfer from port 1 to port 2 equals the transfer back, so Z12 = Z21, Y12 = Y21, h12 = −h21 and AD − BC = 1. That leaves three independent numbers. A dependent source breaks it: a π with a 40 mS forward source has Y21 = 39.98 mS but Y12 = −0.02 mS. A symmetric network looks the same from either end: Z11 = Z22 and, if reciprocal, A = D. A T with 30, 30 and 60 Ω is both.
A T network has series arms of 20 Ω and 40 Ω and a 60 Ω shunt arm. What is Z12?
Z12 = V1/I2 with port 1 open. Current I2 flows through the 40 Ω arm and the 60 Ω shunt; with no current in the 20 Ω arm, V1 equals the shunt voltage, 60·I2. So Z12 = 60 Ω, the shared (shunt) arm.
Worked examples
Z and Y parameters of a resistive T · T network: series arms 20 Ω (port 1 side) and 40 Ω (port 2 side), shunt arm 60 Ω
- Port 2 open, I1 in: V1 = (20 + 60)·I1, V2 = 60·I1, so Z11 = 80 Ω, Z21 = 60 Ω
- Port 1 open, I2 in: Z22 = 40 + 60 = 100 Ω, Z12 = 60 Ω
- ΔZ = 80 × 100 − 60 × 60 = 4400 Ω²
- Y = Z⁻¹: Y11 = 100/4400 = 22.73 mS, Y12 = Y21 = −60/4400 = −13.64 mS, Y22 = 80/4400 = 18.18 mS
- Check Y11 by its own test: port 2 shorted, Rin = 20 + 40 || 60 = 44 Ω, 1/44 Ω = 22.73 mS
Answer: Z = [80, 60; 60, 100] Ω, Y = [22.73, −13.64; −13.64, 18.18] mS
h and ABCD parameters, and the loaded input impedance · The same T (20, 40, 60 Ω), loaded by RL = 100 Ω
- h11 = ΔZ/Z22 = 4400/100 = 44 Ω (input resistance with the output shorted: 20 + 40 || 60)
- h12 = Z12/Z22 = 0.6, h21 = −Z21/Z22 = −0.6, h22 = 1/Z22 = 10 mS
- A = Z11/Z21 = 1.333, B = ΔZ/Z21 = 73.33 Ω, C = 1/Z21 = 16.67 mS, D = Z22/Z21 = 1.667
- AD − BC = 2.2222 − 1.2222 = 1: reciprocal
- Zin = (A·100 + B)/(C·100 + D) = 206.67/3.3333 = 62 Ω; by reduction 20 + 60 || 140 = 62 Ω
Answer: h = [44 Ω, 0.6; −0.6, 10 mS], A = 1.333, B = 73.33 Ω, C = 16.67 mS, D = 1.667; Zin = 62 Ω
A non-reciprocal π: an amplifier-like two-port · π network: 2 kΩ across port 1, 10 kΩ across port 2, 50 kΩ between the ports, and a source gm·V1 = 40 mS × V1 drawn from the output node; load RL = 5 kΩ
- Y11 = 1/2k + 1/50k = 0.52 mS, Y12 = −1/50k = −0.02 mS
- Y21 = gm − 1/50k = 39.98 mS, Y22 = 1/10k + 1/50k = 0.12 mS
- Y12 ≠ Y21, so the network is not reciprocal (AD − BC = Y12/Y21 = −0.0005)
- Voltage gain: Av = −Y21/(Y22 + 1/RL) = −39.98/0.32 = −124.94
- Input impedance: Zin = 1/(Y11 − Y12·Y21/(Y22 + 1/RL)) = 331.3 Ω: the 50 kΩ feedback resistor looks far smaller from the input because the output swings 125 times as much the other way
Answer: Av ≈ −124.9, Zin ≈ 331 Ω; Y12 ≠ Y21 (non-reciprocal)
Interconnected two-ports: series, parallel, cascade, T and π equivalents, ladders
Which syllabus covers thisAU not in syllabusVTU not in syllabusSPPU not in syllabusMU ECC304 Modules 4, 5; EEC302 Module 5KTU PCEET302 Module 4
Why do cascaded two-ports multiply their ABCD matrices?
In a cascade the output voltage of stage 1 is the input voltage of stage 2, and the current leaving stage 1 (−I2) is the current entering stage 2. ABCD parameters are written exactly in those terms, so the overall matrix is the product T = T1·T2, in the order the signal meets them. Order matters: our T followed by our π gives A = 2.828, the π first gives A = 1.812. Both products still satisfy AD − BC = 1, because each stage is reciprocal.
When do Z parameters add in series and Y parameters add in parallel?
Only when every port still carries equal and opposite currents in its two terminals after the joining (the port condition). Join two three-terminal networks in series and the common wire of the top one shorts the lower one's top terminals: our T over our π gives Z11 = 700 Ω, not the sum 722.2 Ω. Brune's test catches it: join one port, open the others and drive; the gap the second joint would close shows 61.5 mV per volt. A 1:1 isolating transformer restores the sum. Parallel joins of networks sharing a common wire are safe, and Y adds.
How are the T and π equivalents of a two-port found?
A reciprocal two-port has three independent parameters, so three impedances can always reproduce it. The T comes from Z: series arms Z11 − Z12 and Z22 − Z12, shunt arm Z12. The π comes from Y: shunt arms Y11 + Y12 and Y22 + Y12, series arm −1/Y12. Converting one into the other is star-delta again. Our T of 100, 100 and 400 Ω has the π equivalent 900 Ω, 225 Ω, 900 Ω, and its Z matrix is identical.
What happens along a ladder of identical sections?
Every symmetric section has an image impedance Z0 = √(B/C) = √(Zsc·Zoc): load it with Z0 and its input also shows Z0, so a chain of them all sees Z0. The propagation constant γ comes from cosh γ = A, and each section multiplies the voltage by e^(−γ). For the T of 100, 100 and 400 Ω: Z0 = 300 Ω and cosh γ = 1.25, so γ = ln 2 and every section halves the voltage, a 6.02 dB attenuator. With a mismatched load, Zin still creeps towards Z0 as sections are added.
Three identical T sections (100, 100, 400 Ω, image impedance 300 Ω) are cascaded and terminated in 300 Ω. What fraction of the input voltage reaches the load?
cosh γ = A = 1.25 gives γ = ln 2, so each section halves the voltage when it is terminated in its image impedance, and a chain terminated in Z0 is matched at every joint: (1/2)³ = 1/8 (18.06 dB).
Worked examples
Cascading a T and a π · N1: T with series arms 100 Ω and 100 Ω, shunt 400 Ω. N2: π with shunt arms 400 Ω and 400 Ω, series 100 Ω. Load 300 Ω
- T1: A = 1 + 100/400 = 1.25, B = 100 + 100 + 100·100/400 = 225 Ω, C = 1/400 = 2.5 mS, D = 1.25
- T2: A = 1 + 100/400 = 1.25, B = 100 Ω, C = 2/400 + 100/400² = 5.625 mS, D = 1.25
- T = T1·T2: A = 1.25 × 1.25 + 225 × 0.005625 = 2.8281, B = 1.25 × 100 + 225 × 1.25 = 406.25 Ω
- C = 0.0025 × 1.25 + 1.25 × 0.005625 = 10.156 mS, D = 0.0025 × 100 + 1.25 × 1.25 = 1.8125; AD − BC = 1
- Zin = (A·300 + B)/(C·300 + D) = 1254.7/4.8594 = 258.2 Ω (the same as reducing from the load end)
Answer: A = 2.828, B = 406.2 Ω, C = 10.16 mS, D = 1.812; Zin = 258.2 Ω
A series connection that breaks the rule · The same T on top and π below, joined in series directly (the T's common wire to both top terminals of the π)
- Z of the T: [500, 400; 400, 500] Ω. Z of the π: Z11 = 400 × 500/900 = 222.2 Ω, Z21 = 400²/900 = 177.8 Ω
- Rule of thumb: Z = Z1 + Z2 = [722.2, 577.8; ...] Ω
- Actual circuit: the T's common wire ties both ends of the π's 100 Ω arm together, leaving the two 400 Ω shunts in parallel (200 Ω) under the T
- So Z11 = 100 + 400 + 200 = 700 Ω, Z12 = 400 + 200 = 600 Ω
- Brune's test: with only port 1 joined and 1 V applied, the open gap at port 2 shows (Z11π − Z21π)/(500 + Z11π) = 0.0615 V, not 0
Answer: Z11 = 700 Ω, not 722.2 Ω: the sum is invalid; an ideal 1:1 transformer at one port of the π would make it valid
A three-section attenuator ladder · Three identical T sections, each 100 Ω, 100 Ω series and 400 Ω shunt
- One section: Zoc = 500 Ω, Zsc = 100 + 100 || 400 = 180 Ω, Z0 = √(500 × 180) = 300 Ω
- cosh γ = A = 1.25, γ = acosh(1.25) = ln 2 = 0.6931 Np = 6.021 dB per section
- Three sections terminated in 300 Ω: V2/V1 = e^(−3γ) = 0.125, Zin = 300 Ω
- T³ gives Zin = B/D = 290.77 Ω with the output shorted and A/C = 309.52 Ω with it open: both already close to 300 Ω
Answer: Z0 = 300 Ω, 6.02 dB per section, output 1/8 of the input (18.06 dB) into 300 Ω
Passive filters: constant-k and m-derived low-pass and high-pass sections
Which syllabus covers thisAU not in syllabusVTU not in syllabusSPPU 203147 Unit 5MU not in syllabusKTU not in syllabus
How does a ladder of L and C pass some frequencies and block others?
Image-parameter theory treats the filter as a chain of identical symmetric sections, each terminated in its own image impedance. For a T section with total series arm Z1 and shunt arm Z2, cosh γ = 1 + Z1/(2Z2). With lossless arms Z1/Z2 is real; while it stays between −4 and 0, γ is purely imaginary: only a phase shift, no loss (the pass band). Outside that range γ gets a real part α: attenuation in nepers per section, 8.686 dB per neper.
What makes a section constant-k, and how is it designed?
If Z1·Z2 = k² does not depend on frequency, the section is constant-k; with L and C, k = R0 = √(L/C). The low-pass T has L/2 in each series arm and C in the shunt arm, and from R0 and fc: L = R0/(π fc), C = 1/(π R0 fc). For R0 = 500 Ω and fc = 5 kHz that is L = 31.8 mH and C = 127 nF. Past fc the image attenuation is 2·acosh(f/fc) nepers: slow at first, only 7.7 dB of exact loss at 1.3 fc between 500 Ω terminations.
How is the high-pass section different?
Swap the roles: series capacitors and a shunt inductor. Z1 = 1/(jωC) and Z2 = jωL still multiply to L/C = R0², so the section is constant-k, but now the pass band lies above fc. The T has 2C in each series arm and L in the shunt arm, with C = 1/(4π R0 fc) and L = R0/(4π fc). Its image attenuation below fc is 2·acosh(fc/f) nepers, a mirror image of the low-pass on a logarithmic frequency scale.
What does an m-derived section fix?
The constant-k section rolls off slowly and its image impedance R0√(1 − (f/fc)²) sags across the pass band. Zobel's m-derived section (m between 0 and 1) keeps the same Z0T but puts a series L-C in the shunt arm; it resonates at f∞ = fc/√(1 − m²), shorting the line. At 1.3 fc an m = 0.6 section gives 24.2 dB against 7.7 dB. Beyond f∞ its attenuation falls back, so practical filters combine a constant-k section, an m-derived section and m = 0.6 half-sections at the ends, which keep the terminating impedance flattest in the pass band.
An m-derived low-pass section has m = 0.8 and fc = 10 kHz. Where is its frequency of infinite attenuation?
f∞ = fc/√(1 − m²) = 10 kHz/√(1 − 0.64) = 10/0.6 = 16.67 kHz. It is always above fc, and moves closer to fc as m gets smaller (m = 0.6 puts it at 1.25 fc).
Worked examples
Designing a constant-k low-pass T · R0 = 300 Ω, fc = 20 kHz
- L = R0/(π fc) = 300/(π × 20 000) = 4.775 mH, so each series arm is L/2 = 2.387 mH
- C = 1/(π R0 fc) = 1/(π × 300 × 20 000) = 53.05 nF
- Check: √(L/C) = 300 Ω and 1/(π√(LC)) = 20 kHz
- At 10 kHz (x = 0.5): cosh γ = 1 − 2x² = 0.5, so α = 0 and the phase shift is β = 60°; Z0T = 300√(1 − 0.25) = 259.8 Ω
- At 40 kHz (x = 2): α = 2·acosh(2) = 2.634 Np = 22.9 dB
Answer: L/2 = 2.39 mH per arm, C = 53.1 nF; 22.9 dB image attenuation at twice the cut-off
Designing a constant-k high-pass T · R0 = 300 Ω, fc = 2 kHz
- C = 1/(4π R0 fc) = 1/(4π × 300 × 2000) = 132.6 nF, so each series arm is 2C = 265.3 nF
- L = R0/(4π fc) = 300/(4π × 2000) = 11.94 mH in the shunt arm
- Check: √(L/C) = 300 Ω and 1/(4π√(LC)) = 2000 Hz
- At 1 kHz (half the cut-off): α = 2·acosh(fc/f) = 2·acosh(2) = 2.634 Np = 22.9 dB
Answer: 2C = 265 nF per series arm, shunt L = 11.9 mH; 22.9 dB at 1 kHz
An m-derived section for a sharper edge · The low-pass design above (L = 4.775 mH, C = 53.05 nF, fc = 20 kHz) with m = 0.6
- Series arms: mL/2 = 0.6 × 4.775/2 = 1.432 mH each
- Shunt arm: mC = 31.83 nF in series with (1 − m²)L/(4m) = 0.64 × 4.775/2.4 = 1.273 mH
- f∞ = fc/√(1 − m²) = 20/0.8 = 25 kHz; check: 1/(2π√(1.273 mH × 31.83 nF)) = 25 kHz
- At 22 kHz (1.1 fc) the image attenuation is 14.9 dB, against 7.7 dB for the constant-k section
- Image impedance at 10 kHz: 259.8 Ω, the same as the constant-k T, so the two sections can be cascaded
Answer: Series 1.43 mH arms, shunt 1.27 mH + 31.8 nF; infinite attenuation at 25 kHz
Network synthesis: Hurwitz polynomials, positive real functions, Foster and Cauer forms
Which syllabus covers thisAU not in syllabusVTU not in syllabusSPPU not in syllabusMU ECC304 Modules 4, 6KTU not in syllabus
What is a Hurwitz polynomial, and how do you test one?
A polynomial is (strictly) Hurwitz when every root has a negative real part, so every natural response it describes dies away. All coefficients must then be present and positive, but that alone is not enough beyond second order. Two equivalent tests: the Routh array (no sign change in its first column), or the continued fraction of the even part over the odd part, which must give as many quotients as the degree, all positive. s³ + s² + 2s + 8 has positive coefficients but fails both: its roots are −2 and 0.5 ± j1.94.
What is a positive real function, and why does it matter?
A driving-point impedance Z(s) of any network of positive R, L, C and coupled coils is positive real (PR): real for real s, and Re Z ≥ 0 whenever Re s ≥ 0. A practical test: N(s) + D(s) must be Hurwitz and Re Z(jω) ≥ 0 at every frequency. If a function is not PR, no passive network can realise it. Z = (s − 1)/(s + 2) fails: at ω = 0 it is −0.5 Ω, a negative resistance, and it stays negative up to √2 rad/s.
What does the impedance of a lossless LC network look like?
An LC impedance is an odd rational function: on the jω axis it is a pure reactance jX(ω). Its poles and zeros are all on the jω axis, simple, and they alternate (interlace): a pole or zero at 0, then zero, pole, zero ... up to a pole or zero at infinity. X(ω) always rises with frequency between poles. Z(s) = (s² + 1)(s² + 16)/(s(s² + 4)) has a pole at 0, a zero at 1 rad/s, a pole at 2, a zero at 4 and a pole at infinity.
How do Foster and Cauer forms turn Z(s) into a circuit?
Foster I expands Z(s) in partial fractions: k∞·s is a series inductor, k0/s a series capacitor, each 2ki·s/(s² + ωi²) a parallel LC tank, all in series. Foster II does the same to Y(s), giving parallel branches. Cauer forms divide repeatedly (a continued fraction) into a ladder: Cauer I removes the pole at infinity each time (series L, shunt C), Cauer II the pole at zero (series C, shunt L). Each canonical form uses the minimum number of elements, equal to the degree: 4 for our example.
How do RC networks differ from LC networks?
An RC impedance has its poles and zeros on the negative real axis, simple and interlaced, with the critical frequency nearest the origin a pole, so Z falls as frequency rises. The same Foster expansion works with real poles: Z(s) = (s + 2)(s + 5)/(s(s + 3)) = 1 + (10/3)/s + (2/3)/(s + 3), which is a 1 Ω resistor, a 0.3 F capacitor and a parallel RC (0.222 Ω with 1.5 F) in series. An RL impedance is the mirror case: the critical frequency nearest the origin is a zero.
Which of these could be the impedance of a lossless LC network?
The critical frequencies must alternate. (s² + 1)/(s(s² + 4)): pole at 0, zero at 1, pole at 2, zero at infinity, alternating. The first has pole at 0, pole at 1: two poles in a row. The third is an even function, real on the jω axis, so it is not a reactance at all.
Worked examples
Two Hurwitz tests · P1(s) = s⁴ + 2s³ + 6s² + 4s + 3 and P2(s) = s³ + s² + 2s + 8
- P1, Routh array first column: 1, 2, (2·6 − 4)/2 = 4, (4·4 − 2·3)/4 = 2.5, 3: no sign change
- P1, continued fraction of (s⁴ + 6s² + 3)/(2s³ + 4s): quotients 0.5s, 0.5s, 1.6s, 0.8333s: four, all positive
- P2, Routh first column: 1, 1, (1·2 − 8)/1 = −6, 8: two sign changes, so two right-half-plane roots
- Check: P2 = (s + 2)(s² − s + 4), roots −2 and 0.5 ± j1.936
Answer: P1 is Hurwitz; P2 is not (two roots with real part +0.5) even though all its coefficients are positive
Foster I and Foster II of an LC impedance · Z(s) = (s² + 1)(s² + 16)/(s(s² + 4)), normalised units (Ω, H, F, rad/s)
- Foster I: Z = k∞·s + k0/s + 2k·s/(s² + 4), with k∞ = 1 and k0 = (1 × 16)/4 = 4
- 2k = [(s² + 1)(s² + 16)/s²] at s² = −4 = (−3)(12)/(−4) = 9
- Elements: series L = 1 H, series C = 1/k0 = 0.25 F, tank C = 1/9 = 0.1111 F with L = 9/4 = 2.25 H
- Foster II: Y = s(s² + 4)/((s² + 1)(s² + 16)), residues 2k1 = 3/15 = 0.2, 2k2 = 12/15 = 0.8
- Branches: L = 1/(2k), C = 2k/ω²: 5 H with 0.2 F, and 1.25 H with 0.05 F, in parallel
Answer: Foster I: 1 H + 0.25 F + (2.25 H parallel 0.111 F); Foster II: (5 H + 0.2 F) parallel (1.25 H + 0.05 F)
Cauer I and Cauer II of the same impedance · Z(s) = (s⁴ + 17s² + 16)/(s³ + 4s)
- Cauer I: Z = s + (13s² + 16)/(s³ + 4s): series L1 = 1 H
- (s³ + 4s)/(13s² + 16) = s/13 + (36/13)s/(13s² + 16): shunt C2 = 1/13 = 0.0769 F
- (13s² + 16)/((36/13)s) = (169/36)s + ...: series L3 = 4.694 H; remainder 16 gives shunt C4 = 9/52 = 0.1731 F
- Cauer II (ascending powers): (16 + 17s² + s⁴)/(4s + s³) = 4/s + ...: series C1 = 0.25 F, then shunt L2 = 13/4 = 3.25 H
- Continuing: series C3 = 9/169 = 0.0533 F and shunt L4 = 13/9 = 1.444 H
Answer: Cauer I: 1 H, 0.0769 F, 4.69 H, 0.173 F; Cauer II: 0.25 F, 3.25 H, 0.0533 F, 1.44 H (four elements in every form)
Keep going
- Analog Electronics interactive book
- Signals and Systems interactive book
- Electrical Machines interactive book
- Power Systems interactive book
- Control Systems interactive book
- GATE EE electric circuits questions
- GATE EE AC circuits and three-phase questions
- GATE ECE networks questions
- All Skill-Lync interactive books
Recommended reading
- Engineering Circuit Analysis, W. H. Hayt, J. E. Kemmerly and S. M. Durbin
- Fundamentals of Electric Circuits, C. K. Alexander and M. N. O. Sadiku
- Network Analysis, M. E. Van Valkenburg
- Network Analysis and Synthesis, F. F. Kuo
- Circuits and Networks: Analysis and Synthesis, A. Sudhakar and S. S. Palli
- Circuit Theory (Analysis and Synthesis), A. Chakrabarti
- Networks and Systems, D. Roy Choudhury
- Electric Circuits (Schaum's Outline), M. Nahvi and J. A. Edminister
How this book was made
Chapters follow the official syllabi listed above. Every interactive's formulas and every worked-example number were checked by script. Explanations, animations and examples are original to Skill-Lync. Topics informed by MIT OpenCourseWare 6.002 (Circuits and Electronics), CC BY-NC-SA. All explanations, animations and examples are original to Skill-Lync; MIT does not endorse this page.