AC-coupling instrumentation amplifier improves rejection range of differential dc input voltage
The need for conditioning low-level ac signals in the presence of both common-mode noise and differential dc voltage prevails in many applications. In such situations, ac-coupling to instrumentation and difference amplifiers is mandatory to extract the ac signal and reject common-mode noise and differential dc voltage. This situation typically occurs in bioelectric-signal acquisition, in which metallic-electrode polarization produces a large random differential dc voltage, ranging from ±0.15V, which adds to low-level biological signals. Input ac-coupling is one approach to removing the differential dc content. But this technique requires adding a pair of capacitors and resistors to ac-couple the inputs of the difference amplifier. The manufacturing tolerances of these components severely degrade the CMRR (common-mode-rejection ratio) of the amplifier. If cost is not an issue, you could perform an initial trim, but this operation is useless for biological applications plagued by wide variations in electrodes and tissue impedances. The differential topology in Figure 1 addresses these problems (Reference 1).
The principle of this ac-coupled instrumentation amplifier is to maintain the mean output voltage at 0V. To do so, you insert an autozero feedback loop, comprising IC_{4}, R_{FB}, and C_{FB}, in a classic three-op-amp instrumentation amplifier. This feedback loop produces a frequency-dependent transfer function:
Consequently, the ac-coupled instrumentation amplifier behaves as a highpass filter with a –3-dB cutoff frequency from the equation f=1/2πR_{FB}C_{FB}. At first glance, you might think that the output-autozeroing behavior of the ac-coupled instrumentation amplifier is perfect. Unfortunately, the output autozeroing capability of this circuit is strongly limited. You can determine this limitation by expressing the output voltage as a function of the input signals and the integrator's output voltage, V_{Z}: V_{OUT}=(1+2R_{2}/R_{1})(V_{IN1}–V_{IN2})+V_{Z}=A_{D}(V_{IN1}–V_{IN2})+V_{Z}, where V_{OUT} is the output voltage. In this expression, A_{D}=1+2R_{2}/R_{1} is the differential gain in the passband. At dc, the output voltage is 0V as long as the integrator's output does not reach its saturation voltage, V_{Z(MAX)}. Therefore, setting the output voltage at 0V in the above expression yields the maximum differential-input dc voltage that this circuit can handle:
Consider, for instance, the typical performance and constraints of a portable biotelemetry system: differential gain of 1000, ±5V split power supplies, and op amps with rail-to-rail output-voltage swing. In this system, the application of the formula for ΔV_{IN} yields a maximum differential-input dc voltage of only ±5 mV. This limited performance is unacceptable for biological applications, in which you encounter differential-input dc voltages of ±0.15V. The ac-coupled instrumentation amplifier in Figure 2 overcomes this limitation, thanks to the addition of "active feedback," which includes voltage divider R_{3}-R_{4} and the associated buffer amplifier, IC_{5}. With this arrangement, the following equations give the new transfer function and highpass cutoff frequency, respectively.
The expression for the output voltage as a function of the input signal and the integrator's output voltage, V_{Z}, becomes:
In this expression, A_{D}=(1+2R_{2}/R_{1})(1+R_{4}/R_{3}) is the new differential gain in the passband.
At dc, the output voltage remains 0V as long as the integrator's output does not reach its saturation voltage, V_{Z(MAX)}. Therefore, setting the output voltage at 0V in the new expression for output voltage yields the new maximum differential-input dc voltage and differential gain. They are, respectively:
In the above equations, the additional term, 1+R_{4}/R_{3}, is the gain of the active-feedback stage.
The new expressions for ΔV_{IN(MAX)} and A_{D(MAX)} clearly show the advantages of Figure 2's ac-coupled instrumentation amplifier with active feedback: For an identical differential gain, you can extend the polarization-voltage range, ΔV_{IN(MAX)}, by a factor equal to the gain of the active-feedback stage. Conversely, for a given polarization-voltage range, ΔV_{IN(MAX)}, you can increase the differential gain by the gain of the active-feedback stage.
The only drawback of this topology is apparent in the expression for f_{C}, the highpass cutoff frequency.You multiply this frequency by the gain of the active-feedback stage. Therefore, to maintain a given cutoff frequency, you must multiply the time constant by a factor equal to the active-feedback stage gain. This factor can be an issue in processing signals whose spectrum includes low-frequency components. In such applications, R_{FB} and C_{FB} can reach prohibitive values. Consequently, you must make a trade-off between the time constant and the active-feedback stage gain. The component values in Figure 2 are a typical example of such a trade-off: The values are for an EEG (electroencephalogram) amplifier with ±5V split power supplies. The amplifier has a differential gain of 1000 and a highpass cutoff frequency of 2.3 Hz, and it can handle a differential-input dc-voltage range of ±0.34V.
To obtain this performance, you set the active-feedback stage gain and the differential-amplifier gain, respectively, to 67.6 and 15. With these gain values, the noise performance of the ac-coupled instrumentation amplifier of Figure 2 is similar to that of a classic instrumentation amplifier. This situation occurs because the autozeroing and active-feedback stages, IC_{4} and IC_{5}, are after the input differential stage, IC_{1} and IC_{2}. Consequently, the gain of the differential stage roughly divides their respective noise contributions, which are therefore negligible. You can use several low-noise op-amps for IC_{1} and IC_{2}. For portable biotelemetry applications, the LT1464 is a good compromise for input-noise density, noise-corner frequency, input-bias current and current drain. (Respectively: V_{NOISE}=26 nV/, f_{C}=9 Hz, I_{BIAS}=0.4 pA, and I_{CC}=230 μA.)
A theoretical analysis using the LT 1464's noise parameters shows that under worst-case conditions, the input-noise voltage should not exceed 11 µV rms. Tests on prototypes confirm this prediction; the tests effectively measure input-noise voltages of 3 to 6 µV rms. To sum up, an ac-coupled instrumentation amplifier with active feedback is well-suited for applications requiring high differential gain, a capability for handling large differential-input dc voltages, and low-noise performance.
Reference |
Check out our Best of Design Ideas section! |
System level design and integration challenges with multiple ADCs on single chip
Understanding the basics of setup and hold time
Product How-to: Digital isolators offer easy-to-use isolated USB option
Managing noise in the signal chain, Part 2: Noise and distortion in data converters
War of currents: Tesla vs Edison
Simple reverse-polarity-protection circuit has no voltage drop
Control an LM317T with a PWM signal
Start with the right op amp when driving SAR ADCs
Currently no items