Putting two resistors in parallel sounds simple. The current splits, each resistor takes its share, and life goes on. In reality, the current does not split evenly unless the resistors are perfectly matched and the layout is perfectly symmetric. Even a 1% mismatch in resistance causes one resistor to carry significantly more current than the other. Over time, that imbalance creates a thermal runaway loop — the hotter resistor drifts further out of spec, takes even more current, gets even hotter, and eventually fails. If you are running resistors in parallel for power handling, current sensing, or redundancy, you need to test the actual current sharing, not just assume it will work.

Why Parallel Resistors Do Not Share Current Equally

The ideal case is straightforward. Two 100 ohm resistors in parallel with 1 amp total current means each resistor carries 0.5 amps. But real resistors are not ideal. They have tolerance, they have temperature coefficients, and they sit at different temperatures depending on where you place them on the board.

A 1% tolerance mismatch means one resistor is 99 ohms and the other is 101 ohms. The 99 ohm resistor carries 50.5% of the current, not 50%. That sounds trivial until you realize the 99 ohm resistor is also dissipating slightly more power, which raises its temperature, which changes its resistance further. For a resistor with a 100 ppm/°C TCR, a 10°C temperature difference shifts the resistance by 0.1%. That shifts the current split by another fraction of a percent. The effect compounds.

At high currents, this compounding becomes dangerous. One resistor ends up carrying 60% or 70% of the total current while the other coasts along at 30%. The overloaded resistor overheats, drifts, takes even more current, and fails open or short. The second resistor then takes the full load and fails too, usually within milliseconds.

Setting Up a Parallel Current Sharing Test

Matching Strategy Before You Test

Do not grab random resistors from the bin and parallel them. If you need matched current sharing, sort the resistors by actual measured value, not by nominal value. Use a precision meter to bin them into groups within 0.1% or better. For power applications, also match the TCR. Two resistors with the same room-temperature value but different TCRs will diverge as soon as they heat up.

Place the resistors as close together as possible on the board. Ideally, they should share the same thermal environment — same copper area, same airflow, same proximity to heat sources. If one resistor sits over a copper pour and the other hangs over an empty area, the thermal resistance to ambient is different, and the current split will drift over time even if the initial values are perfectly matched.

Test Circuit Configuration

Drive the parallel pair with a constant current source, not a voltage source. A current source forces a known total current through the pair regardless of resistance changes. A voltage source lets the total current drift as the resistors heat up and their values change, which makes the data harder to interpret.

Use 4-wire Kelvin connections on each resistor individually. You need to measure the voltage across each resistor separately, not just the total voltage across the pair. The current through each resistor is then I = V_individual / R_individual. This gives you the actual current split, not an estimate based on assumed resistance values.

Instrumentation Requirements

Your voltmeter needs at least 6.5 digits of resolution if you are measuring millivolt drops across low-value resistors. A 3.5-digit bench meter will not resolve the small voltage differences that tell you whether the current is splitting 50/50 or 55/45.

Use a data logger that samples both voltage channels simultaneously. If you measure one resistor, then the other, then the first again, you are chasing a moving target because the resistors are heating up while you switch channels. Simultaneous sampling captures the true instantaneous current split.

Thermal Coupling and Its Effect on Current Sharing

The Hot Spot Problem

When two resistors are in parallel but thermally isolated from each other, the one that happens to carry slightly more current heats up faster. Its resistance drifts, it takes even more current, and the imbalance accelerates. This is thermal runaway in slow motion.

The fix is thermal coupling. Mount both resistors on the same copper pad or the same heatsink so they share the same temperature. When they are at the same temperature, the TCR mismatch becomes the dominant factor, and that is usually much smaller than the runaway effect from thermal isolation.

Do not rely on air cooling to equalize the temperature. Air is a terrible thermal conductor. Two resistors sitting side by side on a board can have a 10 to 20°C temperature difference if one is closer to a heat source or has a larger copper pad underneath it.

Measuring Temperature at the Resistor Body

Do not measure board temperature with a thermocouple somewhere else on the PCB. That number is useless for current sharing analysis. You need the actual resistor body temperature.

Use a fine-wire thermocouple taped directly to the resistor body with Kapton tape. Or use an IR thermometer aimed at the resistor surface — but account for emissivity, which varies by resistor coating color. A black-coated resistor has an emissivity around 0.95. A beige-coated one might be 0.85. Get this wrong and your temperature reading is off by several degrees.

Log the temperature of each resistor individually alongside the current and voltage data. The correlation between temperature difference and current imbalance is the key metric you are looking for.

Interpreting Current Sharing Data

The Current Sharing Ratio

Define the current sharing ratio as the percentage of total current carried by the higher-current resistor. A perfect 50/50 split gives you 50%. A 60/40 split gives you 60%. For most power applications, you want this ratio below 55% at full load. Above that, the hotter resistor is dissipating significantly more power than its rating was designed for, and long-term reliability drops fast.

Plot the sharing ratio versus total current. At low currents, the ratio should be close to 50% because self-heating is negligible. As current increases, the ratio will drift. The slope of that drift tells you how well-matched the pair is. A flat line means good thermal coupling and close resistance matching. A steep upward slope means one resistor is hogging the current and thermal runaway is starting.

Time-Dependent Drift

Run the test for at least 30 minutes at full rated current. The current sharing ratio will change over time as the resistors heat up and stabilize. The initial reading at 10 seconds is not the same as the reading at 30 minutes. You need the steady-state value, not the transient.

If the ratio keeps drifting upward after 30 minutes, you have a thermal runaway problem. No amount of better matching will fix this — the layout is wrong, the thermal coupling is insufficient, or the resistors are too close to their power limit.

Testing at Different Current Levels

Low Current Baseline

Start at 10% of rated current. At this level, self-heating is negligible and the current split should match the resistance ratio almost exactly. If it does not, your resistors are not matched well enough for parallel operation. Bin them tighter or buy a matched pair.

Record the baseline sharing ratio. This is your reference point. Every measurement at higher current should be compared to this baseline to see how much thermal effects are shifting the split.

Rated Current and Beyond

Ramp the current up to the rated value in steps — 25%, 50%, 75%, 100%. Hold each step for 5 minutes and log the data. At 100% rated current, the sharing ratio should still be within your acceptable range (typically 55% or better).

If you need to test beyond rated current for qualification purposes, go to 125% and hold for 1 hour. Watch for any sudden jumps in the sharing ratio — a jump means one resistor is starting to fail. Stop the test immediately when that happens.

Common Mistakes That Skew Your Results

Using Two-Wire Measurement on the Pair

If you measure the total resistance of the parallel pair with a 2-wire meter, you get a single number that tells you nothing about current sharing. A 50/50 split and a 70/30 split can give you almost the same total resistance if the individual values are close. You must measure each resistor separately with 4-wire Kelvin connections.

Ignoring Lead Resistance in Low-Value Resistors

For milliohm shunt resistors in parallel, the lead resistance and solder joint resistance are a significant fraction of the total. If each shunt is 10 milliohms and the lead adds 2 milliohms, your current sharing data is dominated by lead mismatch, not resistor mismatch. Use Kelvin connections with the sense points as close to the resistor body as physically possible.

Testing at Room Temperature Only

Room temperature testing tells you about initial matching. It tells you almost nothing about how the pair will behave at operating temperature. A pair that shares 50/50 at 25°C might shift to 65/35 at 85°C if the TCRs are not matched. Always test at the actual operating temperature, or use the temperature coefficient data to extrapolate.

Assuming Matched Nominal Values Mean Matched Performance

Two resistors from the same batch with the same nominal value can differ by 0.5% or more in actual resistance. For parallel current sharing, 0.5% mismatch is already pushing the limits. Do not assume — measure every individual part before you parallel them.

When Parallel Resistors Make Sense and When They Do Not

Parallel resistors work well for increasing power handling when the resistors are well-matched, thermally coupled, and the current is shared within 55/45 or better. They also work for redundancy — if one fails open, the other takes the full load temporarily.

They do not work well for precision applications. Even a 0.1% mismatch creates a measurable current imbalance, and the imbalance drifts with temperature. If you need precision current division, use a single resistor with the correct value, not two in parallel. The parallel approach trades precision for power handling, and you need to know which one you are optimizing for.