Sensing physical phenomena and preserving the fidelity of the resulting, often-tiny, sensor signals is a craft of its own within the broader discipline of analog circuit design. One of the most common and, counter intuitively perhaps, more challenging sensing tasks is that of measuring electric current.

Why counter intuitively? Well, for starters, the parameter of interest is already in electrical form: I = dQ/dt, and Q is a charge quantum—a number of electrons: no transduction required. Transduction, however, is only the beginning of any sensor-based task and, with the remainder in mind, current sensors face many of the same challenges as transducer-based sensors.

One of the challenges is dynamic range. In the case of current measurement, there are applications that range from pA to kA—a 15-order-of-magnitude range. Thankfully, no single application requires a span greater than a small fraction of that range, but currently sensible (excuse the pun), collectively exhaustive techniques exist to satisfy our industry’s many and various but thankfully non-simultaneous demands.

In addition to the potential range of measurement magnitudes, other issues differentiating sensing methods include accuracy, linearity, drift over temperature, bandwidth, isolation, DC coupling, size and cost. The simplest current sensor is the lowly resistor—the most readily available of passive components. The resistor’s definition is that of a linear I-V converter—a heck of a good start for a current sensing and measurement system (Figure 1). A Kelvin-connected shunt resistor provides precisely located contacts for the low-current measurement path, isolating it from cabling parasitic resistances in the high-current path (red).

Many current measurements are of off-ground feeds but must connect to ground-referenced circuits. Isolation amplifiers can bridge the two domains (Figure 2), but some sensing methods have the added advantage of being self-isolating. For AC current-measurement applications such as mains monitoring, current transformers bring multiple advantages: Low insertion impedance, high output, and high isolation voltage. Many current transformer manufacturers wind turns on toroidal cores. Torroids provide high ratio accuracy because a turn is counted only when the wire passes through the core’s center.

For example, consider a 100:1 toroidal current transformer monitoring a 50-A, 240-V power feed. The power lead passes straight through the core’s center, forming a single “turn”. Connecting a sense resistor across the 100-turn secondary yields a ground-based voltage proportional to the feed’s current. The 100:1 turns ratio reduces the 50-A full scale to 500 mA. A 2-Ω sense resistor gives a 1-V full-scale output and dissipates only 500 mW. The sense resistance reflects back to the measured feed through the turns ratio squared as a negligible 200 μΩ insertion impedance.

Current transformers offer narrower operating bandwidths compared to sense resistors alone—typically to about 1 MHz—and do not offer, by themselves, a spectrum that includes DC. Current transformers also become nonlinear if an application exceeds their specified maximum signal magnitude.

Hall-effect current sensors use a gapped magnetic core, typically with a single primary turn and a Hall-effect element located in the gap. Closed-loop versions operate the core at zero flux, eliminating the magnetic material’s nonlinear component. The bandwidth of such sensors may include DC but often only extends to 150 kHz; often less.

Several companies offer complete Hall-effect current-measurement components that integrate signal conditioning circuitry, greatly simplifying design in. In-line and contactless versions are available from companies such as Allegro MicroSystems, Infineon, and Melexis, to name just three.

Additional methods include SenseFETs, but these are usually limited to unidirectional current measurements. These devices take advantage of the fact that many power MOSFET designs are cell based. Connecting the source contacts of, say, 1 in 1000 cells together and providing a separate sensing source connection, the device operates as an accurate high-ratio current divider provided that the sensing source and main source connections operate at the same potential.

Like many sensing methods, SenseFETs require a small amount of signal processing (Figure 3). In this case the amplifier A1 servos the Sense lead to the same voltage as the Kelvin lead—a requirement for proper ratiometric operation of the SenseFET as noted above. A1’s output is −ISense ∙ RSense. Amplifier A2 negates A1’s output to provide a positive output voltage to the following stage—typically an ADC.

As with virtually all sensing applications, numerous other sensing methods exist. They vary primarily by sensing magnitude range, accuracy, variation with temperature, linearity, and dissipation.