ADI explains PGIA in detail, a device that front-end designers regret not knowing earlier!
In these applications, a programmable gain instrumentation amplifier (PGIA) is a suitable solution for the front end, adapting to the sensitivity of various sensor interfaces while optimizing the SNR. Integrated PGIAs can achieve good DC and AC specifications. This article discusses various integrated PGIAs and their advantages. It also discusses the associated limitations and guidelines that should be followed when building a discrete PGIA to meet specific requirements.
Integrated PGIA
There are many integrated PGIAs in ADI's product portfolio. Integrated PGIAs offer the advantages of shorter design time and smaller size. Digitally adjustable gain is achieved through internal precision resistor arrays. These resistor arrays can be trimmed on-chip to optimize gain, CMRR, and offset, resulting in good overall DC performance. Design techniques can also be applied to achieve a compact IC layout, minimize parasitics, and provide excellent matching, resulting in good AC performance. Due to these advantages, if a PGIA is available that meets the design requirements, it is highly recommended to select such a device. Table 1 lists the available integrated PGIAs along with some key specifications.
Table 1. Programmable Gain Instrumentation Amplifier Specifications
The choice of PGIA depends on the application. The AD825x is very useful in multiplexed systems due to its fast settling time and high slew rate. The AD8231 and LTC6915 use a zero-drift architecture and are suitable for systems that require precision performance over a wide temperature range.
There are also many devices that integrate the multiplexer, PGIA, and ADC to form a complete DAQ solution. Examples are the ADAS3022, ADAS3023, and AD7124-8.
Table 2. DAQ System Specifications
The choice of these solutions depends mainly on the specifications of the input signal source. The AD7124-8 is designed for slow applications that require very high accuracy, such as temperature and pressure measurement. The ADAS3022 and ADAS3023 are suitable for relatively high bandwidth applications, such as process control or power line monitoring, but their power consumption is higher than the AD7124-8.
Implementing a Discrete PGIA
Some systems may have one or two specifications that cannot be met by the above integrated devices. Typically, users need to build their own PGIA using discrete components if the following requirements exist:
-
Multiplexed systems requiring higher bandwidth, very high scan rates
-
Ultra-low power consumption
-
The system requires custom gain or attenuation
-
Low input bias current for high impedance sensors
-
Very low noise
One common approach to designing a discrete PGIA is to use an instrumentation amplifier with the desired input characteristics, such as the low noise AD8421, and a multiplexer to switch gain resistors to change the gain.
Figure 1. AD8421 and multiplexer for switching gain.
In this configuration, the on-resistance of the multiplexer is actually in series with the gain resistor. This on-resistance changes with the voltage on the drain, which presents a problem. Figure 2, taken from the ADG1208 data sheet, shows this relationship.
Figure 2. On-resistance vs. drain voltage for the ADG1208.
The series combination of the on-resistance and the gain resistor causes a nonlinear error in the gain. This means that the gain will change with the common-mode voltage, which is bad. For example, the AD8421 requires a 1.1kΩ gain resistor to achieve a gain of 10. For the ADG1208, the on-resistance changes by as much as 40Ω when the source or drain voltage changes by ±15V, resulting in a gain nonlinearity error of about 3%. With larger gains, this error becomes more pronounced, and the on-resistance may even become comparable to the gain resistor.
Alternatively, a multiplexer with low on-resistance can be used to reduce this effect, but at the expense of higher input capacitance. Table 3 illustrates this by comparing the ADG1208 and ADG1408.
Table 3. On-resistance and capacitance trade-offs in multiplexers
The input capacitance of the switch causes another problem with the configuration shown in Figure 1, because the RG pin on any given three-pin op amp instrumentation amplifier is very sensitive to capacitance. The switch capacitance can cause peaking or instability in this circuit. A bigger problem is that the capacitance imbalance on the RG pins degrades the ac common-mode rejection ratio (CMRR), which is a key specification for instrumentation amplifiers. The simulation plot in Figure 3 shows the degradation of CMRR when different multiplexers are used on the gain pins of the AD8421. The plot clearly shows that the CMRR degrades more as the capacitance increases.
Figure 3. Simulated CMRR using different switches.
To minimize the degradation in AC CMRR, the best solution is to ensure that the RG pins have the same impedance. This can be achieved by balancing the resistors and placing the switch element between the two resistors, as shown in Figure 4. In this case, the multiplexer does not work due to the inherent capacitive imbalance across the switch. In addition, since the drains of the multiplexer are shorted together, only one resistor can be used on one side of the RG pin, which still causes an imbalance.
Figure 4. Discrete PGIA using a balanced configuration.
In this case, a quad SPST switch, such as the ADG5412F, is recommended. In addition to the flexibility of the switch to use balanced resistors, the capacitance of the drain and source is also balanced, which reduces the degradation of the CMRR. Figure 5 compares the ac CMRR of using a multiplexer on the gain pins of the AD8421 versus using a quad SPST switch.
Figure 5. CMRR simulation for SPST switch and multiplexer configurations
The ADG5412F also features a low on-resistance that is very flat over the drain or source voltage range, as shown in Figure 6. It is specified to have a maximum of 1.1Ω over the drain or source voltage range. Returning to the original example, with a gain of 10 for the AD8421 and a gain resistor of 1.1kΩ, the switch introduces only 0.1% gain nonlinearity. Nevertheless, there is still a drift component that becomes more pronounced at higher gains.
Figure 6. On-resistance vs. common-mode voltage for the ADG5412F.
To eliminate the parasitic resistance effect of the switch, instrumentation amplifiers with different architectures can be used to achieve arbitrary gain. The AD8420 and AD8237 use an indirect current feedback (ICF) architecture and are excellent choices for applications that require low power and low bandwidth. In this configuration, the switch is placed in a high impedance sensing path, so the gain is not affected by changes in the switch on resistance.
Figure 7. Discrete PGIA with indirect current feedback instrumentation amplifier.
The gain of these amplifiers is set by the ratio of external resistors in the same manner as noninverting amplifiers. This provides the user with greater flexibility, as the gain-setting resistors can be selected based on the design requirements. Standard thin-film or metal-film resistors can have temperature coefficients as low as 15ppm/°C, which results in better gain drift than standard instrumentation amplifiers that use a single external resistor to set the gain, where the mismatch between the on-chip and external resistors typically limits the gain drift to around 50ppm/°C. For the best gain error and drift performance, resistor networks can be used for tolerance and temperature coefficient tracking. However, this comes at the expense of cost, so discrete resistors should be preferred unless absolutely necessary.
Another solution, and the one that offers the most flexibility, is a three op amp instrumentation amplifier architecture using discrete components, as shown in Figure 8, with gain resistors switched via a multiplexer. There is a much greater range of op amps to choose from than instrumentation amplifiers, so designers have more choices, which allows them to design around specific design requirements. Special functions such as filtering can also be built into the first stage. A difference amplifier in the second stage completes this architecture.
Figure 8. Discrete PGIA
The choice of input amplifier is directly dependent on the DAQ requirements. For example, a low power design requires the use of an amplifier with low quiescent current, while a system that anticipates high impedance sensors at the input can utilize an amplifier with ultra-low bias current to minimize errors. Dual amplifiers should be used for better tracking over temperature.
It can be noted that when using the configuration shown in Figure 8, the on-resistance of the switch is also in series with the high impedance input of the amplifier, so it does not affect the gain. Recalling the trade-off between on-resistance and switch input capacitance, since the constraint on on-resistance no longer exists, the design can choose a low input capacitance switch, such as the ADG1209. In this way, instability and ac CMRR degradation are avoided.
As with the previous design, gain accuracy and drift will be determined by the resistors. Discrete resistors can be selected with appropriate tolerance and drift to meet the application design requirements. Again, using a resistor network can achieve higher accuracy, better tolerance, and temperature tracking, but at an increased cost.
The second stage of the three op amp instrumentation amplifier is responsible for rejecting common-mode voltages. A difference amplifier with an integrated resistor network is recommended for this stage to ensure the best CMRR. For single-ended outputs and relatively low bandwidth applications, the AD8276 is a good choice. If differential outputs and higher bandwidth are required, the AD8476 can be used. Another option for the second stage is to use the LT5400 as the gain setting resistors around a standard amplifier. This may take up more board space, but on the other hand gives more flexibility in the choice of amplifier, and the user can design more around specific design requirements.
It should be noted that the layout of the discrete PGIA requires care. Any imbalance in the board layout will cause the CMRR to degrade with frequency.
Discrete PGIA Design Example
Figure 9 shows an example of a discrete PGIA built for a specific design specification. In this design, the PGIA built should have very low power consumption. The input buffer is selected as the LTC2063, which has a low supply current of 2μA maximum. The switch element is selected as the ADG659, which has a low supply current of 1μA maximum and low input capacitance.
Care must also be taken when selecting the passive components in the circuit to meet the low power requirements. Improper passive component selection will result in increased current consumption, which will offset the effect of using low power components. In this case, the gain resistor needs to be large enough not to consume too much current. The selected resistor values (to provide gains of 1, 2, 5, and 10) are shown in Figure 9.
Figure 9. Low-power PGIA design.
For the second stage difference amplifier, the LTC2063 is used with the LT5400 quad matched resistor network (1MΩ option). This ensures the lowest current consumption and the precise matching of the resistors preserves the CMRR performance.
The circuit was powered from a 5V supply and evaluated using different common-mode voltages, differential input voltages, and gains. Under the optimum condition of the reference voltage and inputs being held at midsupply, the circuit consumed only 4.8μA.
Some increase in current is expected when the differential input changes, as current will flow through the gain resistors, with a value equal to |V OUT –V REF |/(2MΩ||1MΩ). Figure 10 below shows the current consumed for different gains. Due to the gain, the data is measured with respect to the output.
Figure 10. Supply current vs. output voltage
When different common-mode voltages are applied to the input, the current is also expected to increase. The applied voltage will cause current to flow through the resistors in the second stage, causing additional current consumption equal to |V CM – V REF |/1MΩ. The 1MΩ resistors were chosen specifically to minimize this current in the LT5400. Figure 11 below shows the effect of common-mode voltage on current consumption at different gains:
Figure 11. Supply current vs. common-mode voltage.
The quiescent current of the circuit in shutdown mode was also measured. When all devices are shut down, the circuit consumes only 180nA. This does not change, even if variables such as common-mode voltage, reference voltage, and differential inputs change, as long as they all remain within the supply range. All devices have a shutdown option in case further power savings are needed and the user wants to power cycle. This circuit is very useful in portable battery-powered applications; otherwise, the key specifications cannot be achieved with the integrated PGIA.
Programmable gain instrumentation amplifiers are key components in the data acquisition space, enabling good SNR performance even when used with sensors of varying sensitivity. Using an integrated PGIA reduces design time and improves the overall DC and AC performance of the front end. If an integrated PGIA is available that meets the requirements, such a device should generally be used in the design. However, when the system requires specifications that cannot be achieved with existing integrated devices, a discrete PGIA can be designed. By following the correct design recommendations, an optimal design can be achieved even with a discrete approach, and various implementation methods can be evaluated to determine the best configuration for a specific application.
ADI website updated and upgraded again