A current divider is a parallel resistor network that splits an input current into fractional branch currents inversely proportional to their resistance. If you need to route 5mA to a sensor and 15mA to a dummy load from a single 20mA source, you don't need complex active switching; you just need two properly sized resistors in parallel. The direct answer to sizing them is calculating the inverse resistance ratio, but the real engineering challenge lies in driving the network with a stiff current source and managing failure modes.
The Topology: Nodes, Math, and the 'Stiff Source' Rule
To design reliably, we must define the physical nodes. A basic two-branch current divider consists of three primary nodes:
- Node A (Input): The top junction where the total input current ($I_{in}$) enters the network.
- Node B (Branch Split): The bottom junction where the current recombines before returning to ground. (Note: In a pure parallel loop, Node A and Node B are technically the two terminals of the parallel pair. For clarity in schematic tracing, we treat the top rail as Node A and the bottom ground rail as Node C, with the branches spanning between them).
- Node C (Ground Return): The common reference point (0V) where the branch currents sum back to $I_{in}$.
The governing equation for a two-resistor network ($R_1$ and $R_2$) is:
$I_1 = I_{total} \times \frac{R_2}{R_1 + R_2}$
Design Walkthrough: Splitting 20mA with E24 Resistors
Let's design a divider to split a 20mA input into a 5mA branch ($I_1$) and a 15mA branch ($I_2$).
1. Determine the Resistance Ratio
Current splits inversely to resistance. Since we want a 1:3 current ratio (5mA vs 15mA), our resistance ratio must be 3:1. Therefore, $R_1$ must be three times larger than $R_2$.
2. Pick Real Component Values
We need standard E24 series values. Let's set $R_2 = 100\Omega$. Consequently, $R_1 = 300\Omega$. Both are standard 5% through-hole values (e.g., Yageo CFR-25JB series).
3. Calculate Node Voltage and Power
The equivalent parallel resistance ($R_{eq}$) is:
$R_{eq} = \frac{300 \times 100}{300 + 100} = 75\Omega$
The voltage drop across the entire network (Node A to Node C) will be:
$V = I_{total} \times R_{eq} = 0.020A \times 75\Omega = 1.5V$
Now, verify power dissipation to select the right physical package:
- $P_1 = I_1^2 \times R_1 = (0.005)^2 \times 300 = 7.5mW$
- $P_2 = I_2^2 \times R_2 = (0.015)^2 \times 100 = 22.5mW$
Both values are well under the 250mW rating of a standard 1/4W axial resistor. No derating or heat sinking is required at a 25°C ambient.
Behavior Matrix: What Changes When Elements Shift
Assuming an ideal 20mA constant current source feeding the network, here is how the circuit reacts to component variations. This table is critical for tolerance analysis.
| Parameter Change | Effect on $I_1$ (300Ω branch) | Effect on $I_2$ (100Ω branch) | Effect on Node A Voltage |
|---|---|---|---|
| $R_1$ increases (e.g., to 330Ω) | Decreases (current shifts away) | Increases (absorbs shifted current) | Increases (higher $R_{eq}$) |
| $R_2$ decreases (e.g., to 91Ω) | Decreases | Increases | Decreases (lower $R_{eq}$) |
| $I_{in}$ scales to 40mA | Doubles to 10mA | Doubles to 30mA | Doubles to 3.0V |
| Temperature rises (PTC effect) | Slight decrease (copper/carbon drift) | Slight increase | Increases slightly |
Failure Modes: What Breaks at the Extremes?
Unlike series voltage dividers, where an open component simply floats the output node to zero current, parallel current dividers exhibit aggressive failure modes because the total input current must go somewhere.
Scenario A: $R_1$ Fails Open
If the 300Ω resistor burns out or a breadboard jumper pops loose, $I_1$ drops to 0mA. The entire 20mA from the source is now forced through $R_2$ (100Ω).
- New $I_2$: 20mA (a 33% overload compared to the design 15mA).
- New Node Voltage: $20mA \times 100\Omega = 2.0V$.
- Risk: If $R_2$ was sized tightly for power, this 33% current spike yields a 77% power spike ($I^2R$), potentially causing a cascading thermal failure.
Scenario B: $R_1$ Fails Short
If a solder bridge shorts across $R_1$, the resistance of that branch drops to near zero.
- New $I_1$: ~20mA (all current takes the path of least resistance).
- New $I_2$: 0mA (the 100Ω resistor is bypassed entirely).
- New Node Voltage: ~0V.
- Risk: The load on Branch 2 loses power completely. Furthermore, if your current source is actually a voltage source with a series limiting resistor (a pseudo-current source), a short here will pull the rail voltage down to zero, potentially browning out other microcontrollers on the same bus.
Decision Path: Passive Divider vs. Active Mirror
When should you use a simple passive resistor network versus an active semiconductor solution? Use this decision tree to terminate your design choices.
| Design Requirement | Passive Resistor Divider | Active Current Mirror (e.g., LM334, Matched BJTs) | Default Pick |
|---|---|---|---|
| Cost & Board Space | <$0.05, minimal footprint | >$0.50, requires 2-3 components | Passive |
| Precision / Matching | Limited to 1% resistor tolerance | Excellent (laser-trimmed silicon matching) | Active |
| Load Voltage Independence | Fails if load voltage varies widely | Maintains ratio regardless of load voltage | Active |
| High Frequency / AC Signals | Poor (parasitic capacitance ruins phase) | Good (transistor bandwidth limits apply) | Active |
Step-by-Step Breadboard Verification
Do not trust the math until you verify it on the bench. Here is the exact procedure to test your 20mA divider using an LM317 configured as a constant current source.
- Build the Stiff Source: Wire an LM317 in constant-current mode. Place a set resistor ($R_{set}$) between the OUT and ADJ pins. To get 20mA, calculate $R_{set} = 1.25V / 0.020A = 62.5\Omega$. Use a 62Ω 1% metal film resistor (e.g., Vishay MRS25 series) for the closest standard value, yielding ~20.1mA.
- Connect the Divider: Connect the LM317 output to Node A. Place your 300Ω ($R_1$) and 100Ω ($R_2$) resistors in parallel between Node A and the ground rail.
- Verify Node Voltage First: Before breaking the circuit to measure current, use your DMM in voltage mode across the parallel pair. You should read 1.5V (±50mV). If you read 5V or 0V, your LM317 is wired incorrectly or oscillating (add a 1µF ceramic cap across the input if it oscillates).
- Measure Branch 1 Current: Turn off the power. Pull the top lead of $R_1$. Insert your DMM in series (mA mode) between the LM317 output and $R_1$. Power on. Expect 5.0mA.
- Measure Branch 2 Current: Repeat the series insertion for $R_2$. Expect 15.1mA (accounting for the 62Ω set resistor variance).
Final Verdict: When to Commit to Passive Division
Current dividers are elegant, but they are unforgiving of sloppy source design. The topology works flawlessly provided you respect the boundary conditions: the source must be stiff, and the loads connected to the branches must not alter the equivalent resistance of the branch itself. If your branch load is a microcontroller drawing variable current, a passive current divider will fail catastrophically as the MCU switches between sleep and active modes, shifting the node voltage and starving the other branch.
For fixed, static loads—like biasing a photodiode or splitting a reference current for a DAC—a passive network of 1% metal film resistors is the most cost-effective, lowest-noise solution on the bench. Calculate your inverse ratios, verify your $I^2R$ thermal limits, and always test the open-failure mode to ensure your remaining branch can survive the full source current.
For deeper reading on DC network theorems, refer to the All About Circuits textbook chapter on divider circuits, or review the Electronics Tutorials guide on parallel current division for additional mathematical proofs.
