Signal flow graphs are a powerful tool used in electrical engineering to analyze complex systems, particularly those involving feedback, multiple inputs and outputs, and interconnected subsystems. They visually represent the relationships between different variables and how signals propagate through a system, offering a more compact and intuitive form than traditional block diagrams. Understanding practical examples helps students and engineers grasp the concepts more effectively and apply them to real-world design and analysis tasks. This article expands on the fundamentals, presents detailed examples from control systems, analog circuits, filters, power systems, and digital signal processing, and introduces essential tools like Mason's gain formula for direct transfer function derivation.

What Are Signal Flow Graphs?

A signal flow graph (SFG) is a directed graph that models the flow of signals through a system. Nodes represent system variables (signals), and directed edges (branches) indicate the functional relationship between those variables. The weight of each edge, called the transmittance or gain, represents the multiplier or transfer function from the source node to the destination node. SFGs are particularly powerful for linear time-invariant (LTI) systems because they allow direct application of Mason's gain formula to compute the overall transfer function without algebraic manipulation.

Compared to block diagrams, SFGs are more concise: they eliminate the need for summing junctions and takeoff points, representing them simply as nodes. They also provide a clear visual representation of feedback loops and forward paths, making them ideal for analyzing systems with multiple loops or coupled components.

Fundamental Concepts and Terminology

Nodes and Branches

Every variable in the system is represented by a node. Nodes are of three types:

  • Source nodes (input nodes) have only outgoing branches.
  • Sink nodes (output nodes) have only incoming branches.
  • Mixed nodes have both incoming and outgoing branches.

Each directed branch has a transmittance (gain) that multiplies the signal from the source node to the destination node.

Paths and Loops

  • Forward path: A path from a source node to a sink node that travels along branches in the direction of signal flow and does not pass through any node more than once.
  • Loop: A closed path that starts and ends at the same node, without crossing any node twice.
  • Non-touching loops: Loops that share no nodes.
  • Forward path gain: Product of all branch transmittances along a forward path.
  • Loop gain: Product of branch transmittances around a loop.

Mason's Gain Formula: A Practical Tool

Mason's gain formula provides a systematic way to compute the overall transfer function T from a source to a sink node directly from the signal flow graph:

T = (1/Δ) * Σk (Pk Δk)

Where:

  • Δ = 1 - (sum of all individual loop gains) + (sum of gain products of all possible combinations of two non-touching loops) - (sum of gain products of three non-touching loops) + ...
  • Pk = gain of the k-th forward path.
  • Δk = the value of Δ for the part of the graph not touching the k-th forward path.

This formula eliminates the need to solve simultaneous equations and is especially valuable when the SFG contains many interlocking loops.

Example 1: Feedback Control System

Consider a standard negative feedback control system with a plant transfer function G(s), feedback transfer function H(s), input R(s), and output C(s). The block diagram is familiar: the error signal is the difference between input and feedback, and the plant output is fed back through H(s).

In signal flow graph form, we have:

  • Node R(s) as source.
  • Node E(s) representing the error.
  • Node C(s) as sink (output).
  • Branch from R(s) to E(s) with transmittance 1.
  • Branch from E(s) to C(s) with transmittance G(s).
  • Branch from C(s) back to E(s) through H(s) but note that the feedback is subtracted, so the branch transmittance from C(s) to E(s) is -H(s) (assuming negative feedback).

To apply Mason's formula:

  • Forward path: R → E → C, gain P1 = 1 × G(s) = G(s).
  • Loops: One loop from E → C → E with gain L1 = G(s) × (-H(s)) = -G(s)H(s).
  • Δ = 1 - (L1) = 1 + G(s)H(s).
  • Δ1 = 1 (since the path touches the loop).
  • Transfer function T = G(s) / (1 + G(s)H(s)).

This matches the well-known closed-loop transfer function. The SFG approach makes it easy to see the effect of the feedback loop and to extend to multiple feedback loops, such as in cascade control or state-feedback systems.

Example 2: Operational Amplifier with Negative Feedback

An inverting operational amplifier configuration is a classic application. Let the op-amp have open-loop gain A (very large), input resistor Ri, feedback resistor Rf, input voltage vin, output voltage vout. The inverting input is at virtual ground, but to analyze with SFG, consider the actual op-amp model with finite gain.

We can define nodes:

  • vin (source)
  • vd (differential input voltage: v+ - v-; non-inverting input grounded so vd = -v-)
  • vout (sink)
  • v- (inverting input node)

Branches:

  • From vin to v- through a voltage divider formed by Ri and Rf: transmittance = Rf / (Ri + Rf) (considering the current and the virtual short assumption is not yet applied; this is the forward path from input to the inverting node).
  • From vout to v- through feedback: transmittance = Ri / (Ri + Rf).
  • From v- to vd: since vd = -v-, transmittance = -1.
  • From vd to vout: transmittance = A (open-loop gain).

We have two forward paths from vin to vout:

  • Path 1: vin → v- → vd → vout, gain = [Rf/(Ri+Rf)] × (-1) × A.
  • Path 2: vin → v- → vout (if we consider a direct branch? Actually, there is no direct branch; the only way is through vd. So only one forward path exists.

Loops:

  • Loop: v- → vd → vout → v- with gain = (-1) × A × [Ri/(Ri+Rf)] = -A Ri/(Ri+Rf).

Using Mason's formula: Δ = 1 - ( -A Ri/(Ri+Rf) ) = 1 + A Ri/(Ri+Rf).

Δ1 = 1.

Transfer function T = P1/Δ = [-A Rf/(Ri+Rf)] / [1 + A Ri/(Ri+Rf)] = -A Rf / (Ri+Rf + A Ri).

For very large A, the term A Ri dominates the denominator, leading to T ≈ -Rf/Ri, the ideal inverting gain. This SFG approach rigorously demonstrates how finite gain affects the closed-loop gain and reveals the loop gain A Ri/(Ri+Rf), which drives stability analysis.

Example 3: Filter Design — Active Low-Pass Filter

Consider a second-order Sallen-Key low-pass filter. The output of an op-amp is fed back through a capacitor- resistor network to create a complex conjugate pole pair. The SFG can model the two integrator stages and the feedback pathes.

Define nodes: Vin, Vx (node between R1 and C1), Vy (node between R2 and C2), and Vout. The op-amp is configured as a unity-gain buffer, so Vout = Vy. Branches represent the transfer functions of the RC stages. The complete SFG allows the derivation of the transfer function Vout/Vin.

By applying Mason's formula, the denominator yields the characteristic polynomial: 1 + s(R1C1 + R2C2 + (1-K)R1C2) + s² R1R2C1C2, where K is the gain of the non-inverting stage (here K=1). Engineers use this to choose component values for a desired cutoff frequency and Q factor.

Example 4: Power System Load Flow Analysis

In power systems, signal flow graphs can model the linearized relationship between bus voltages and injected currents. For a power system with n buses, the nodal admittance matrix Y relates current injections to bus voltages: I = Y V. The SFG representation treats each bus voltage as a node and each admittance element Yij as a branch from node i to node j with transmittance 1/Yij (or more accurately, using the impedance).

A simplified example: two-bus system with generation at bus 1 and load at bus 2, connected by a transmission line with series impedance Z and shunt admittance. The SFG can help compute the voltage at bus 2 given the voltage at bus 1 and the load current. By identifying loops formed by shunt elements and forward paths, engineers can derive sensitivity factors and identify critical contingencies where voltage collapse might occur. Smart grid applications often use such models for real-time stability assessment.

Advanced Application: Digital Signal Processing

Signal flow graphs are widely used in digital signal processing (DSP) to represent discrete-time systems, particularly FIR and IIR filters. In a digital SFG, nodes represent delayed signal values (e.g., x[n], x[n-1], ...) and branches represent multiplication by coefficients and the unit delay operator z-1.

Consider a second-order IIR filter described by the difference equation: y[n] = b0x[n] + b1x[n-1] + b2x[n-2] - a1y[n-1] - a2y[n-2]. The SFG, often drawn in a "direct form II" structure, contains two delay elements, five multiplier branches, and two summing nodes. Mason's gain formula yields the transfer function H(z) = (b0 + b1z-1 + b2z-2) / (1 + a1z-1 + a2z-2).

The SFG representation helps in analyzing quantization effects, overflow conditions, and stability in fixed-point implementations. It also facilitates the conversion to parallel or cascade forms for better numerical performance. For a comprehensive tutorial, see The Scientist and Engineer's Guide to Digital Signal Processing.

Conclusion

Practical examples like feedback control systems, operational amplifiers, active filters, power system load flow, and digital filters demonstrate the versatility of signal flow graphs in electrical engineering. They provide a clear visual method to analyze and design complex systems, making them an essential tool for engineers and students alike. Mastering SFG construction and Mason's gain formula equips you with a systematic approach to transfer function derivation, stability analysis, and sensitivity studies across many domains. For further reading, explore Wikipedia's detailed article on signal-flow graphs or a dedicated textbook on control systems. The ability to transform a messy circuit or block diagram into a clean graph of interconnected nodes is a skill that pays dividends throughout a career in electrical engineering.