Gain-of-two instrumentation amplifier uses no external resistors

An instrumentation amplifier offers precise gain without feedback resistors, and, at any value of gain, it provides high input impedances at its noninverting and inverting inputs. In a typical IC instrumentation amplifier, a single resistor that connects across two gain-adjustment pins determines the circuit's overall gain. Integrated versions of most instrumentation amplifiers allow the pins to remain open for unity gain but require finite-value gain-setting resistors for gains exceeding one. Although the gain-adjustment resistor might comprise a tiny surface-mounted device, its electrodes and internal resistive layer extend the conductive surface connected to the IC's gain-adjustment pins. The extended surface acts as an antenna and thus makes the amplifier more susceptible to stray external electromagnetic fields.

Figure 1 shows an instrumentation amplifier that offers a gain of two without using any external resistors. The circuit comprises a cascade of asymmetrical, differential-output amplifier, formed by two channels of IC1; an Analog Devices (www.analog.com) AD8222 instrumentation amplifier; and a difference amplifier comprising one half of IC2, a second AD8222. All three instrumentation-amplifier sections in the circuit provide a stand-alone gain of one. Because the differential outputs of the first stage have opposite signs, their difference is twice that of the difference of the input signals.

The circuit's worst-case gain error does not exceed the value of δ2=3δ1, where, at a gain of one, δ1 represents the maximum gain error of one section of the AD8222. For B-grade ICs, you calculate the value of δ2 as δ2≤ 0.06% (Reference 1). Typically, the value of δ2 rarely reaches its maximum value. Given the reasonable assumptions that all three amplifiers' gain errors are independent and obey a gaussian distribution, the probability of occurrence of δ2=3δ1 is about 1/20 the probability of encountering a single amplifier that has a maximum gain error of δ1.

Illustrations:

Figure 1