The Anderson loop is a fundamental measurement circuit topology. Its unique feature is the six-terminal dual-differential subtractor illustrated below. The subtractor is a continuously operating analog element, not a digital function. The ideal subtractor's output is uninfluenced by any voltage that may exist between each differential input terminal pair. Stated another way, the two differential inputs of the ideal subtractor each exhibit infinite common mode rejection.

The Anderson loop employs a continuous, multi-differential Kelvin measurement based on the same excitation current with pairs of elements observed by dual-differential subtractors.

The added capabilities of the Anderson loop as compared to the Wheatstone bridge arise from the fact that the Wheatstone bridge accomplishes a PASSIVE subtraction while the Anderson loop accomplishes an ACTIVE subtraction of electrical potential differences.

The benefits of employing active analog subtractors in measurement applications are enormous. Many things previously considered impossible become routine - for example, the continuous separation of thermoelectric signals from IR drop signals. A strain gage can be connected in an Anderson loop with thermocouple wire and, with DC excitation, strain and temperature outputs are available, each essentially uninfluenced by the other.

The Anderson loop circuit topology is universal in that resistances, inductances and capacitances can be used as gauge and reference elements with dynamic loop excitation waveforms.

In its simplest form, with a single sensor resistance and a resistance reference element in a constant DC current loop whose voltage drops are observed by a subtractor, the circuit looks like this:

Note that, unlike the Wheatstone bridge, (1) Anderson loop circuit equations are fundamentally linear for any change in circuit elements and (2) wire voltage drops are simply ignored, not somehow compensated for by depending on actual wires to behave exactly alike in the measurement environment.

The key element in the Anderson loop is the dual-differential subtractor. The goodness of the resulting measurement depends in large part on the goodness of the subtractor. Fortunately microvolt-stable subtractors are not difficult to obtain as standard, off-the-shelf components.

Using only 25% of the excitation power, the Anderson loop can deliver the same sensor output voltage level as a Wheatstone bridge. This tremendous increase in efficiency happens because the Anderson loop output voltage is double what can be obtained from a Wheatstone bridge having the same excitation voltage across each of the various sensor elements.

Ratiometeric operation of the Anderson loop is accomplished by simply using V_ref, the voltage drop across the reference element as the reference input to the system observing V_out, for example, an A-to-D converter. With ratiometeric operation, regulation of the excitation current becomes theoretically unnecessary.

Multiple sensing elements can be observed with respect to one or more reference elements by using multiple subtractors in the same Anderson loop. For example, changes in the four resistance elements in a typical Wheatstone bridge transducer can be independently observed when they are wired in an Anderson loop. Small differences developing between widely separated sensor elements can be reliably observed.

Large variations in the various lead wires do not significantly affect practical Anderson loop measurements. Connecting wires can be of any conducting material and only need to be large enough to survive in the measurement environment.

When the sensing elements are physically near each other a savings in the quantity of connecting wires can result. For example, a strain gauge rosette (three co-located sensors sensitive to strain at various angles) can be observed with only six connecting wires. Even with only two wires per sensor, connecting wire resistance can vary randomly due to environmental effects without introducing a significant error in strain rosette measurements.

Any active element in a measurement system will add some noise to the measured result. However, the overall noise floor of an Anderson loop measurement will be lower than the noise observed using Wheatstone bridge circuits as long as the noise in the measurement environment is significantly greater than the noise added by the subtractor. The maximum available improvement in signal-to-noise ratio is six decibels when compared to the Wheatstone bridge circuit.