A current splitter circuit (often called a current divider) takes a single input current and routes it into two or more parallel branches at a precise ratio. Unlike a voltage divider, which relies on a stiff voltage source and series resistors, a current splitter requires a constant current source and parallel impedances. If you hook a parallel resistor network to a standard 5V voltage source, you have not built a current splitter; you have just built two independent loads.

Getting the split ratio right on the bench comes down to choosing between a passive resistive topology and an active semiconductor mirror. Below is the exact decision framework, component math, and failure-mode analysis you need to build one that actually works.

The Core Decision: Passive Resistive vs. Active BJT Current Splitter

The most common mistake makers make is defaulting to an active transistor mirror when a passive resistor network would suffice, or vice versa. Your choice must be driven by the nature of your load, not just the source.

Load ConditionTopologyConcrete Part Pick
Fixed resistive load, <100mA, low costPassive Resistive SplitterYageo MFR-25 Series (1% Metal Film)
Dynamic load, high precision, <100mAActive BJT Current MirrorNexperia BCM847BS (Matched Dual PNP)
High current (>500mA), dynamic loadActive MOSFET MirrorALD1101 (Matched Dual NMOS Pair)

Why this topology over the alternative? A passive splitter is completely immune to high-frequency noise injection and thermal runaway, but it is strictly load-dependent. If the resistance of Branch 1 changes (e.g., an LED heats up and its forward voltage drops), the current split shifts. An active BJT mirror forces the current ratio based on transistor geometry and Vbe matching, ignoring load impedance variations entirely—provided you stay within the compliance voltage of the source.

Topology Deep-Dive: Node Labels and the Math That Actually Matters

Let us map the standard passive topology. We define four critical nodes:

  • Node A (Source Input): Where the constant current ($I_{total}$) enters the network.
  • Node B (Split Junction): The parallel branching point. Kirchhoff's Current Law dictates that $I_{total} = I_1 + I_2$ here.
  • Node C (Branch 1 Return): The ground or return path for the first load/resistor.
  • Node D (Branch 2 Return): The ground or return path for the second load/resistor.

The governing formula for a two-branch passive splitter is:

$$I_1 = I_{total} \times \frac{R_2}{R_1 + R_2}$$

Notice the inverse relationship: Branch 1 current is determined by the resistance of Branch 2. For a comprehensive breakdown of the underlying Kirchhoff's laws, the All About Circuits current divider chapter remains the definitive open-source reference.

Design Walkthrough: Building a 30mA / 20mA Passive Splitter

Let us design a real circuit. We have a 50mA constant current source (perhaps an LM334-based sink or an op-amp controlled MOSFET). We need to split this into Branch 1 (30mA) and Branch 2 (20mA) to bias two separate transistor stages.

First, we establish the resistance ratio. Since voltage across parallel branches is equal ($V = I_1R_1 = I_2R_2$):

$$30mA \times R_1 = 20mA \times R_2 \implies R_2 = 1.5 \times R_1$$

We select standard 1% E96 series values: $R_1 = 200\Omega$ and $R_2 = 300\Omega$.

Compliance Voltage Check: The equivalent parallel resistance is $120\Omega$. At 50mA, the voltage drop across the splitter network is $V = 0.05A \times 120\Omega = 6V$. Your constant current source must have a compliance voltage of at least 6V plus its own internal headroom (usually 2V to 3V). If your power supply is only 5V, this circuit will fail to regulate, and the split ratio will collapse.

Next, we calculate power dissipation to pick the physical resistor size. Branch 1 dissipates $I^2R = (0.03)^2 \times 200 = 0.18W$. Branch 2 dissipates $(0.02)^2 \times 300 = 0.12W$. Standard 1/4W (0.25W) through-hole resistors are sufficient, but they will run warm to the touch. For better thermal stability, step up to 1/2W Yageo MFR-50 resistors.

Behavior Matrix: What Breaks at the Extremes?

Understanding failure modes is where bench experience separates from textbook theory. In a voltage-fed parallel circuit, an open branch simply turns off. In a current-fed splitter, an open branch is a destructive event.

Fault ConditionBranch 1 (Target 30mA)Branch 2 (Target 20mA)Physical Consequence
R1 Shorted (0Ω)50mA (All current)0mASource compliance voltage drops to near 0V. Circuit survives, but Branch 2 is starved.
R1 Open0mA50mA (All current)R2 now dissipates $0.05^2 \times 300 = 0.75W$. A 1/4W resistor will overheat, smoke, and fail open.
R2 Shorted (0Ω)0mA50mASource compliance drops. Branch 1 starved.
R2 Open50mA (All current)0mAR1 now dissipates $0.05^2 \times 200 = 0.50W$. A 1/4W resistor will burn out.

This is the golden rule of current splitters: The surviving branch absorbs the fault current. If you are driving sensitive loads (like laser diodes or bare LED die) instead of robust resistors, you must add Zener clamps at Node B to prevent overcurrent destruction if a parallel branch opens.

Breadboard Verification: Step-by-Step Testing Protocol

Do not just plug it in and hope. Follow this verification sequence to protect your multimeter and your components.

  1. Verify the Source First: Before connecting the splitter, wire your 50mA constant current source to a dummy $100\Omega$ load. Measure the voltage across the dummy load. It should read exactly 5.0V ($0.05A \times 100\Omega$). If it does not, fix the source before proceeding.
  2. Build the Splitter Network: Insert the $200\Omega$ and $300\Omega$ resistors into the breadboard. Connect Node A to the source, and Nodes C/D to ground.
  3. Measure Voltage, Not Current: Set your multimeter to DC Volts. Probe across $R_1$ and $R_2$. Both must read exactly 6.0V. If they differ, you have a breadboard contact resistance issue or a miswired node.
  4. Calculate Before Breaking: Use Ohm's law on your measured voltage. If $V_{R1} = 6.02V$, then $I_1 = 6.02 / 200 = 30.1mA$. This non-invasive check confirms the split without risking your DMM fuse.
  5. Series Ammeter Verification: If you must measure current directly, power down. Break the Branch 1 path and insert your DMM in series. Ensure your DMM leads are in the correct mA/uA jacks (e.g., on a Fluke 117, use the mA jack, not the 10A jack, for 0.01mA resolution). Power up and verify ~30mA.
Warning: Never place your multimeter in current-measurement mode (ammeter) directly across Node B and Ground. An ammeter has near-zero internal resistance. You will instantly short the current source, dumping the full 50mA (or more, if the source fails) through your meter's internal shunt, likely blowing the DMM's internal fuse.

The Final Verdict: Which Topology Should You Build?

For 90% of maker, sensor-biasing, and LED-driving applications where the load impedance is fixed or highly predictable, build the passive topology using Yageo MFR-25 series 1% metal film resistors. It costs pennies, introduces zero high-frequency switching noise, and avoids the thermal drift inherent in discrete semiconductors. Just ensure your power supply has the compliance voltage headroom to support the $I \times R$ drop at Node B.

However, if your load is dynamic—such as a varying thermistor, an active IC, or a component with a wide manufacturing tolerance—abandon the passive splitter immediately. Instead, use a Nexperia BCM847BS matched dual PNP transistor configured as an active current mirror. As detailed in the BCM847BS datasheet, the monolithic silicon die ensures the two transistors are thermally coupled and Vbe-matched to within 2mV. This guarantees a precise 1:1 (or scaled) current split regardless of what the load does, provided you stay within the 45V Vceo and 100mA Ic limits of the SOT-363 package.