The analog Instrumentation Amplifier (INA) has been a solid building block for a variety of applications where the differential voltage between two signals is of interest. An application example that requires this type of signal conditioning is found in the front-end electronics of an electrocardiograph with the two inputs of the INA connected directly to a person through electrodes. Pressure-sensing circuits offer another opportunity to use an INA in the design. In these circuits the difference between the voltages on the legs of the Wheatstone Bridge indicates the amount of force being exerted on the sensor.

The INA differentiates two signals and provides an optional gain, while rejecting most of the common-mode signals present. This analog function is easily achieved with two or three operational amplifiers, as shown in the figure below. These classical configurations of amplifiers not only take the mathematical difference of the two signals, but their combination offers a high input impedance with a low output impedance. Bear in mind that the INA block is one of many signal-conditioning stages that usually lead toward the final analog-to-digital conversion. Consequently, the features of the INA are synergistic with the sensor and the following signal-conditioning electronics, which all reside in the analog domain. It is no wonder the INA has dominated the market for so long.

Click here for Image

The Battle of the Domains: Analog versus Digital

But will the INA continue to dominate the market? In recent developments with integrated solutions, it seems like this pure analog signal-conditioning solution is experiencing a degree of market erosion. More recent developments of instrumentation quality ADCs would make one think that the pure analog solution will soon be phased out. Maybe so, and maybe not, you be the judge.

In terms of performance, the INA has good low-noise accuracy, fairly good common-mode rejection, and high input impedance. Features include adjustable gain and a wide common-mode input range.

In the INA circuits above, the input terminals of the devices have the equivalent input impedance of an operational amplifier. Most typically this impedance is in the range of a few G-ohms or higher. This high input impedance reduces signal loss between the driving circuitry and the input of the INA. The input range of these amplifiers is limited by the fundamental topology restrictions of any amplifier: It is not unusual for the input range of the analog INA be limited from 0 V up to 1.5 V from the positive-power supply-rail. Although this limitation is slight, it does affect the dynamic range of the INA function. And finally, a differential signal gain can be implemented with the front end of the INA by adjusting Rg. Both of the applications mentioned above, electrocardiograph and bridge sensing, require some amount of gain. This gain is implemented to ensure that the signal-to-noise ratio for the following A/D conversion process is good enough to maintain signal integrity. After the signal is gained by the INA, a 12-bit to 16-bit converter is usually more than adequate.

Common-mode rejection is a key characteristic of the INA. This feature ensures that any high level voltage (several volts) is removed from the resulting signal (usually millivolts.) This function is not limited to the dc domain. Removal of extraneous common-mode noise, such as the mains frequency, is extremely important as well. With the three-operational-amplifier INA circuit, it would seem intuitively obvious that the input stage of the INA (A1 and A2) would also dominate the common mode signal rejection. This is not necessarily true. The over-bearing factor in common-mode rejection for this INA is in the output stage, specifically the matching of the resistors R1 to R2 and R3 to R4. If these resistors are adequately matched the common-mode rejection of A1 and A2 will start to have an influence. Finally, the output amplifier serves the purpose of being capable of swinging as close to the rail as possible in order to maintain a good dynamic range. The output impedance should be low in order to keep the signal intact through the following signal stage.

The Paradigm Shifts

As always, the digital gremlins are encroaching into analog territory. The most obvious step is to pull the INA function into the ADC. This has been attempted with two types of instrumentation quality converters, the Successive Approximation Register (SAR) converter and the Delta-Sigma (DS) converter. Both have their merits in the marketplace: The SAR converter is faster, the delta-sigma converter is more accurate. But the question is, "How do they fair in this type of application where common-mode rejection and accuracy are critical?"

Can the SAR Converter do the Job Today?

In the SAR converter arena there are several devices on the market claiming true differential performance. It turns out that after close inspection, you will find that many of these devices operate with a subset of the differential capability of the INA, while others are fully equivalent. An example of a differential-input stage that is a subset of the true INA differential stage is shown below. In this ADC input stage, the positive input has a full swing from rail-to-rail but the non-inverting input does not. The non-inverting input is capable of rejecting common-mode signals as long as they are plus or minus a few hundred millivolts. This SAR converter is 12-bit accurate with an adjustable-voltage reference. As the voltage at the reference pin is decreased, the LSB size is also decreased.

Click here for Image

Despite all of these features, this SAR converter is not capable of tackling the tough INA problems. Since the inverting input is limited to a few hundred millivolts of range, neither the electrocardiograph or bridge-sensor application can use this ADC directly.

A second example of an SAR ADC that differentiates two signals is in the next diagram. This SAR input stage differentiates, but the sampling of the two signals is asynchronous. With this converter the non-inverting input is sampled one full clock cycle ahead of the inverting input. This strategy is not all bad if this slight delay is taken into account.

Click here for Image

The third example of a SAR ADC has a true differential-input stage and is shown below. In this circuit both inputs are sampled at the same time. Additionally, both inputs have slightly better input ranges than its analog counterpart, the INA. With this circuit, gain is also achieved by reducing the magnitude of the voltage reference to the chip. Specifically, an overall gain improvement of approximately 2.5 is possible with a peak-to-peak LSB size of 488 ýV. The maximum conversion speed of this type of converter is 200 kHz.

Click here for Image

This third example has an input stage that is the closest to the INA specifications, so close that it is included in a specification comparison table at the end of this article. Both input ranges are good (0 to 4 V) and the common-mode rejection is respectable (80dB), as is the input impedance (15 pF.) The only thing lacking is the differential accuracy. In low-end applications, this shortcoming is less critical and for high-end applications, the accuracy can be improved in two ways: