How Do You Select an INA240/INA180-Class Current-Sense Amplifier for a Motor or Switching Power Stage?

Detailed Explanation

A standard high-side current-sense amplifier's CMRR specification describes how well it rejects a common-mode voltage that's essentially DC or slowly varying, the shunt sitting at a fixed supply rail while the amplifier measures a small differential voltage across it. That's an accurate model for many applications (battery current monitoring, a steady supply rail), but it breaks down for a motor phase leg or a switching converter's high-side switch node, where the shunt's common-mode voltage doesn't sit at a fixed level at all: it swings from near ground to the full rail voltage on every PWM edge, often in tens of nanoseconds.

An amplifier's ability to reject a common-mode voltage that's changing this fast is a different specification from its DC CMRR, and a part with excellent DC CMRR can still perform poorly here if it wasn't specifically designed for it. This is exactly the gap INA240/INA180-class parts address: they're specified and optimised for common-mode voltage that changes rapidly, not just for a common-mode voltage that's simply large.

PWM Common-Mode Rejection vs Ordinary CMRR

The practical symptom of inadequate PWM rejection is a current reading that glitches in sync with the switching frequency, a spike or dip appearing at every PWM edge rather than a clean signal tracking the actual current. Because this glitch is correlated with switching, not random noise, it can't be filtered away without also delaying or distorting the genuine current signal, and in a fast current control loop (motor field-oriented control, a synchronous buck's peak or average current mode control), that distortion feeds directly into the control loop's stability and response.

Texas Instruments' INA240 datasheet specifies an "Enhanced PWM Rejection" figure separately from the part's standard CMRR, quantifying rejection specifically against fast common-mode transients rather than a slowly varying common-mode level. When comparing candidate parts for a switching application, check for this kind of transient-specific specification rather than relying on the standard CMRR number alone, since a large standard CMRR figure doesn't guarantee good behaviour at PWM edge rates.

Selecting Between Common Options

A Worked Example: BLDC Phase Current Sensing

Consider a three-phase BLDC motor driver using low-side shunts in each phase leg, switched at 20 kHz, with a 48 V bus. Each shunt's common-mode voltage sits near ground for most of the low-side switch's on-time, but every time the high-side switch turns on and the low-side switch turns off, the switch node (and with it, the shunt's common-mode reference) slews rapidly toward 48 V before returning, tens of nanoseconds later, back toward ground. This happens 20,000 times a second, on every phase, for the entire time the motor runs.

An amplifier without adequate transient rejection produces a current-reading glitch synchronised to that edge on every single switching cycle. Because the glitch repeats at the switching frequency rather than appearing as random noise, a simple low-pass filter on the amplifier output has to trade off between filtering the glitch and preserving the actual current waveform's own frequency content, since both can occupy an overlapping frequency range depending on the motor's electrical speed. Selecting an amplifier with genuine PWM/transient rejection avoids this trade-off entirely, since the glitch is suppressed at the source rather than filtered out afterward.

For a field-oriented control loop that samples current synchronously with the PWM cycle (a common technique, sampling in the middle of a switch's on-time when ringing has settled), a glitchy amplifier's error is at its worst exactly during the switching transition, less so at the sampling instant itself, which is part of why this failure mode can be easy to miss on a bench scope trace and only shows up as control-loop performance issues (torque ripple, current-loop instability at certain operating points) once the motor is under load.

Design Considerations

• Treat PWM/transient common-mode rejection as the primary selection criterion for any switching-leg current measurement, not an optional extra to check after choosing on DC CMRR or price alone.
• Keep the shunt and amplifier input traces short and symmetric, using a Kelvin connection directly at the shunt pads. PCB layout parasitics that couple switching-edge noise asymmetrically into the two input traces can degrade real-world rejection below the part's datasheet figure even with a well-chosen IC.
• Match the amplifier's bandwidth to the control loop it feeds, not just to the switching frequency. A current control loop typically needs the sense chain's bandwidth to exceed the loop's target crossover frequency by a comfortable margin, not just be faster than the PWM frequency itself.
• Confirm the bidirectional output's reference/mid-point voltage is compatible with the target ADC's input range, since a bidirectional current-sense amplifier's zero-current output typically sits at a mid-supply reference rather than at 0 V.
• Motor and power-stage current-sense design: Zeus Design selects and lays out current-sense amplifiers for motor drive and switching power designs, including PWM-rejection verification during bring-up.

Common Mistakes

• Selecting a current-sense amplifier by DC CMRR alone for a switching-leg application, then discovering current-reading glitches synchronised to the PWM edges during bring-up.
• Routing the shunt's Kelvin sense traces asymmetrically or with a shared ground return that picks up switching noise, degrading rejection below what the chosen IC's datasheet promises.
• Choosing a unidirectional current-sense amplifier for a motor phase current that actually reverses direction, clipping the output whenever current flows the "wrong" way.
• Ignoring the amplifier's bandwidth relative to the control loop's needs, then finding the current control loop's achievable bandwidth is limited by the sense chain rather than by the power stage itself.