Too often, even though we have a great analog circuit design, the system specification changes before the product goes into production. Programmable-gain amplifiers (PGAs) can provide the adjustability needed to deal with such last-minute changes. These tried-and-true tiny devices have now become essential in data-acquisition systems largely for three reasons: the level of performance they provide at low cost, the space savings they bring to the system, and the overall design simplicity they offer.
First, let's address the matter of performance. For data-acquisition systems that must handle a wide range of input levels, PGAs provide the ability to change the gain on-the-fly, which effectively increases the dynamic range of the system. In fact, placing a PGA in front of an analog-to-digital converter (ADC) can expand that converter's input amplitude range by over 60 dB while still operating from the same voltage supply rail. Applications that can benefit from this wide dynamic range include thermocouple amplifiers, electronic scales, optical sensors, laser power adjustments in medical imaging, strain-gauge amplifiers, and other multirange, high-resolution data-acquisition systems.
To see how this is possible, consider a typical example, as depicted in the figure. Suppose a PGA that has a gain setting of 10 V/V is driving an ADC that is operating at 250 ksamples/s. In this case, a 10-kHz input signal at 60% of full scale shows a total harmonic distortion (THD) of −87 dB at the ADC's digital output. Under the same conditions, 100-kHz input signals produce THD values of around −75 dB. Noise effects (both random and quantization) in the ADC are divided by the gain of the amplifier when referred to the input voltage (VIN). Because of this, the circuit can acquire a signal that's 40 dB down from the full scale of 5 V peak-to-peak with a signal-to-noise ratio (SNR) of over 70 dB. Such performance from an ADC alone (70 + 40 = 110 dB of useful dynamic range at 250 ksamples/s) would be far more expensive, if available at all.
Second, the space-savings advantage offered by PGAs is essential in data-acquisition systems. PGAs save space by eliminating the need for those usual strings of feedback and/or gain resistors, as well as the need for control logic devices. Without a PGA, these extra parts must be external to the ADC. With a PGA, all of these additional components are replaced by a single device in a package having a footprint as small as 3 by 3 mm.
Third, simplicity of design is another key reason to get on the PGA bandwagon. Any op amp can be made "programmable" by adding digital potentiometers or gain switch matrices. However, these extra parts not only increase cost and complexity, they also cause your signals to be routed over longer distances and through multiple components, thereby raising noise and crosstalk. In fact, the internal logic used for today's PGAs actually minimizes thermal noise by determining the optimal combinations of input resistance and feedback resistance.
Another benefit is the extent to which error calculations are simplified. With discrete approaches, it can be tedious to estimate gain error and linearity over the entire input range, let alone over the full range of temperature variations. For many PGAs, these specifications can be found on the datasheet. Amplifier ac stability, which is effectively a "built-in" feature of most PGAs, can also be a real headache to accomplish using discrete solutions.
Increasingly, digital control is taking over system calibration and autoranging functions, as well as control loops. Using simple parallel or industry-standard serial interfaces like the serial-parallel interface (SPI) bus, PGAs form the critical link between the microprocessor and the signal path. While hard-core analog designers may be wary of losing signal integrity, the extra software control knob that PGAs provide can lead to higher signal-to-noise ratios and an increase in the number of usable bits provided by the ADC.