Shunt Resistor Selection: Kelvin, Heat, Accuracy & RFQ

This is not a “what is a resistor” page. It’s a decision guide for engineers and buyers who need a shunt resistor that behaves in real hardware: stable at temperature, measurable through noise, routable with Kelvin sense, and scalable to production without surprise failures.

Jump to what you need

Answer Box Intent Choose R Accuracy budget Power & thermal Kelvin layout Protection RFQ checklist Recommended models FAQ

Low-ohm shunt resistor with Kelvin sense routing on a PCB

One-Screen Answer (Selection + Procurement)

A shunt resistor decision is a trade between measurable voltage and lost power/heat. The “right” part is the one that hits your accuracy target after temperature rise, TCR, layout parasitics, amplifier offset, and ADC noise are counted.

Fast pick rules

Pick R so Vshunt(max) fits your front-end range with headroom.
Ensure P = I²R stays below rating with margin, at your real PCB temperature.
Use Kelvin (4-wire) sense for low mΩ parts, or your “mΩ” becomes “marketing.”
Choose low TCR (ppm/°C) when accuracy matters over heat.
Specify package/mechanics (metal element vs metal plate vs foil) based on surge and thermal cycling.

Most common failure mode

Picking a tiny “1 W” shunt for a current pulse that lasts “only a second,” then discovering the copper pour and solder joints become the heater. The measurement drifts, the board browns, and the “sensor” becomes a space heater with opinions.

Decision shortcut

If you need highest accuracy → prioritize low TCR + Kelvin layout + stable thermal design. If you need highest current → prioritize low R + metal-plate/large package + short, wide copper. If you need low BOM cost → limit accuracy promises, and define a realistic drift budget upfront.

Search Intent: What “Shunt Resistor” Really Means

People searching shunt resistor usually want selection (which value/package/TCR?), application (how to route Kelvin sense?), and troubleshooting (why does my reading drift or saturate?). So every technical point below ties back to a decision outcome: what to choose, how to validate it, and what to tell procurement in an RFQ so you can actually build it twice.

Answer Box Visual (Selection-First Checklist Scene)

How to Choose Shunt Resistance (mΩ Math That Doesn’t Lie)

Start with your measurement chain limits. The shunt creates a voltage: V = I × R. Bigger R makes measurement easier (more volts), but wastes more power and creates more self-heating. Smaller R saves power but pushes you into microvolts—where amplifier offset, ADC noise, and layout parasitics can dominate.

Step 1 — pick your Vshunt target

Many current-sense front ends are happy around 25–100 mV at full-scale (FS), but your number should come from: the amplifier input range (including common-mode), ADC full-scale, and overload headroom.

If your chain saturates at 50 mV, don’t “borrow” extra headroom from hope.
If you want better SNR, move toward the upper end (without cooking the shunt).

Step 2 — compute R from Imax

Choose R = Vshunt(FS) / Imax. Example: you want 50 mV at 20 A → R = 0.05 / 20 = 2.5 mΩ.

Then check power: P = I²R → 20² × 0.0025 = 1.0 W (and that’s before temperature and copper are considered).

Step 3 — check “minimum measurable” current

For low current resolution, compute V at Imin: with 2.5 mΩ at 100 mA → V = 0.1 × 0.0025 = 250 µV. If your amplifier offset + drift is comparable to 250 µV, your low-end reading will wander.

Practical tip

If you’re stuck between “measurable” and “too hot,” you can: (1) increase gain and use a lower shunt voltage target, (2) use a larger shunt package/metal-plate style, (3) measure in two ranges (dual gain paths) if your system truly spans decades of current.

Accuracy Budget: Where Shunt Measurements Actually Drift

Shunt accuracy is not just the tolerance printed on the reel. Your real-world error is a stack: shunt tolerance, shunt TCR, self-heating, amplifier input offset and drift, ADC reference error, and layout-induced error (trace resistance and ground bounce).

Tolerance vs TCR

Tolerance is the room-temperature starting error. TCR (ppm/°C) is how resistance changes with temperature. If your shunt runs 60°C hotter than ambient and TCR is 200 ppm/°C, the drift is 200 × 60 = 12,000 ppm = 1.2%—often larger than the original tolerance.

Amplifier offset matters at mΩ

With a 1 mΩ shunt, 1 A produces 1 mV. If your amplifier has 50 µV offset, that’s a 5% error at 1 A before anything else. Selection impact: if you need low-current accuracy, shunt choice and front-end choice are a single decision, not two independent line items.

Layout resistance is a stealth resistor

A few milliohms of copper in the wrong place can add to the shunt and create a current-dependent error. If your “sense” points aren’t true Kelvin, you measure shunt + copper + solder—then wonder why production units disagree.

Procurement note: if your spec says “±1% current,” you must specify shunt TCR and the allowed temperature rise, otherwise suppliers will optimize for unit price and you’ll pay later in calibration and returns.

Kelvin vs Non-Kelvin Layout (Good vs Bad Side-by-Side)

Power, Pulses, and Thermal Reality

Shunts fail (or drift) when the thermal model is treated like a footnote. Power is P = I²R. The temperature rise depends on package, copper area, airflow, and where the heat can go. Many “high power” shunts assume a certain pad size—ignore that and your rating becomes fiction.

Continuous current check

Compute P at Imax and compare to rating with margin (common practice: keep steady-state below ~50–70% of rating, depending on environment). If your design runs in a hot enclosure, derate harder.

Pulse current check

A “short pulse” can still be destructive if it exceeds the element’s surge capability. Procurement impact: request pulse/surge curves and define your waveform (amplitude, duration, repetition, duty cycle). “It’s only 10 ms” is not a spec.

Thermal drift = measurement drift

As the shunt heats, R changes (TCR), and the reading changes—even if current is stable. If you can’t reduce heat, compensate in firmware using measured board temperature, but only after validating with real airflow and copper.

Design move that often fixes everything

Step up one mechanical class: from a tiny chip shunt to a larger metal-plate or 4-terminal part. You usually gain lower TCR, better heat spreading, and better Kelvin geometry—often cheaper than weeks of “why does it drift?” debugging.

Kelvin (4-Wire) Sense: The Layout Pattern That Separates Lab from Production

With low-ohm shunts, the pads, solder joints, and copper traces can be comparable to the shunt itself. Kelvin sense means your measurement taps the shunt element directly, not the current-carrying copper. If you route sense from the wrong point, you measure “whatever is in series,” which changes with assembly and temperature.

Kelvin routing checklist

Sense traces start at the dedicated Kelvin pads (or at the inner edge of current pads if no Kelvin pads exist).
Sense traces are thin/quiet; current traces are short/wide (don’t mix jobs).
Route the sense pair as a tight pair to the amplifier inputs.
Keep sense away from switching nodes and high dI/dt loops.

Ground strategy (avoid “Kelvin…ish”)

Many errors come from where the amplifier reference returns. If the amplifier ground shares a noisy power ground, your measurement becomes a “shunt + ground bounce” detector. Give the sense amplifier a clean reference path, and join grounds intentionally.

Production repeatability

Kelvin is not just “nice.” It reduces unit-to-unit variation caused by solder volume, pad wetting, and copper tolerance. If you plan alternates, Kelvin geometry also makes replacements less risky—because you’re measuring the element, not the PCB.

Red flag: if your design review shows the sense traces connected to the high-current pour “somewhere near the shunt,” expect drift and mismatch in production.

Thermal Reality (Heat Spreading Demonstration)

Protection and Survivability: Don’t Let the Shunt Become the Fuse

Shunts live in power paths. That means they see inrush, short-circuit events, inductive kick, and sometimes reverse polarity. Even when the shunt survives electrically, the measurement chain can be damaged by transient common-mode shifts or input overvoltage.

Front-end input protection

Use input series resistance and RC filtering (set bandwidth intentionally).
Consider clamps/TVS where the harness is long or the environment is noisy.
Validate amplifier common-mode range during faults (not only normal operation).

Short-circuit and surge

If your system can hit a hard short, define the fault duration and protection trip time. Procurement impact: specify surge/pulse capability requirements for the shunt, not only steady power.

EMI and switching converters

In buck/boost systems, the shunt is often near high dI/dt loops. If your reading looks like “random glitter,” reduce loop area, route Kelvin properly, add filtering, and check sample timing vs switching phase.

RFQ-Ready Shunt Resistor Checklist

Copy/paste this into an RFQ so suppliers quote comparable parts (and you don’t discover “missing assumptions” at EVT).

Decision item	Why it matters	What to specify
Current range	Defines R and power; affects accuracy at low end.	Imax continuous, pulse waveform, Imin resolution target.
Target shunt voltage	Sets R; impacts amplifier/ADC range and SNR.	Vshunt at full-scale + overload headroom.
Accuracy budget	Tolerance alone is not enough; drift dominates.	Tolerance, TCR (ppm/°C), allowed temp rise.
Package & construction	Determines heat spreading, surge, cycling reliability.	Size (e.g., 2512/3920), metal-plate/foil, 2T/4T.
Kelvin requirement	Controls layout error and repeatability.	4-terminal preferred; sense pad geometry constraints.
Supply continuity	Avoid requalification and calibration retuning.	Lifecycle status, alternates list, qualification plan.

“mΩ Math” Concept (Bench Measurement)

Recommended Shunt Resistor Models (Searchable Part Numbers)

Below are popular, searchable current-sense/shunt resistor part numbers used across power supplies, motor drives, battery systems, and industrial controls. No manufacturer names are listed here—only model numbers—so your team can cross-check availability and second sources.

Chip / metal-element style (common on PCBs)

WSL2512R0100FEA (example low-ohm current-sense in 2512 class)
KRL3264E-C-R012-F-T1 (2512-class current-sense part number)
KRL3216-M-R004-F-T1 (Kelvin-capable current-sense style model number)

Metal-plate / high-power shunts (better heat spreading)

PSR400-0L50 (metal-plate shunt family code format)
PSR100-1L00 (metal-plate shunt family code format)

Tip: metal-plate styles are often the best “first fix” when your current reading drifts under load, because the thermal gradient is smaller.

Bolt-on / busbar shunts (system-level current measurement)

BVN-M-R001-1.0 (busbar shunt model format used for higher current ranges) :contentReference[oaicite:5]{index=5}

Tip: bolt-on shunts reduce PCB heating but require careful sense wiring and mechanical mounting repeatability.

How to use this list safely

Treat the list as a shortlist, not a promise. For each candidate, confirm: resistance value, tolerance, TCR, power rating under your footprint, surge capability, and whether it supports true Kelvin sensing. Then validate in your exact mechanical and thermal environment.

Current Path vs Sense Path (Trace Width Emphasis)

FAQ: Shunt Resistor Selection

How do I choose the shunt resistor value (mΩ) for my current range?

Pick a target full-scale shunt voltage your front-end can handle (often 25–100 mV), then compute R = V/Imax. Next check power (P = I²R) and confirm the shunt won’t overheat in your real PCB environment. Finally verify low-current resolution: if V at Imin is near your amplifier offset/noise, you’ll see drift and jitter at the low end.

Why is Kelvin sensing important for low-ohm shunt resistors?

At milliohm levels, copper traces, pads, and solder joints can add meaningful resistance. Kelvin (4-wire) sensing measures the element itself rather than the high-current copper path. This improves accuracy and production repeatability, and reduces unit-to-unit variation from assembly.

What matters more: tolerance or TCR?

Tolerance sets the initial room-temperature error. TCR determines drift as the shunt heats. In many real designs, self-heating makes TCR the dominant term. If accuracy must hold over load and temperature, specify low TCR and limit temperature rise.

Why does my current reading drift upward during a constant load?

Common causes are shunt self-heating (TCR), amplifier offset drift, and non-Kelvin routing that makes copper resistance part of the measurement. Fixes include using a larger/higher-power shunt, improving heat spreading copper, enforcing true Kelvin sense routing, and validating the amplifier’s drift specs.

How do I prevent the shunt resistor from failing in surge or short-circuit events?

Define your fault waveform (peak current, duration, repetition) and ensure the shunt’s surge capability covers it. Also protect the measurement front-end with filtering and clamps where needed, and confirm amplifier common-mode behavior during faults. If faults are frequent, move to a more robust shunt construction (often metal-plate or busbar style).

What should I include in an RFQ for a shunt resistor?

Include current range (continuous + pulses), target shunt voltage, accuracy requirements (tolerance + TCR + allowed temperature rise), package/construction preference, Kelvin requirement, and supply continuity expectations (alternates and lifecycle status).