CN0355

The circuit shown in Figure 1 is based on the AD7793, 24-bit, Σ-Δ. It has three differential analog inputs and has an on-chip, low noise PGA with gain that ranges from unity gain to 128, making it ideal for multiple sensor interfaces. The AD7793 consumes a maximum of only 500 μA and is therefore suitable for low power applications. It has a low noise, low drift internal band gap reference and can accept an external differential reference. The output data rate is software programmable from 4.17 Hz to 470 Hz.

The AD8420 low power in-amp with a supply current of 80 μA maximum, can operate up to 36 V single-supply and is used to remove the common voltage at the bridge transducer. It can also provide gain, if needed, to the small differential signal output of the transducer.

The ADA4096-2, dual channel op amp, with a typical supply current of 60 μA per amp and a wide operating input voltage range of up to 30 V, drives the sensor bridge. The other half of the ADA4096-2 is used as a buffer for the reference voltage.

There are a wide variety of pressure sensors requiring a voltage drive between 5 V and 15 V. The circuit shown in Figure 1 provides a complete solution for bridge type transducers and has four critical sections; the transducer voltage drive, the in-amp, the reference buffer, and the ADC.

Bridge Type Transducer Voltage Drive

The ADA4096-2 is configured as a noninverting amplifier with configurable gain set by the feedback resistor, as shown in Figure 2.

The gain is set by configuring the jumpers indicated in Table 1.

The transfer function can be computed as follows:

Where R_F can be 40.2 kΩ, 91 kΩ, or 140 kΩ, and R8 = 10 kΩ.

An NPN transistor is use to boost the current needed to drive the bridge sensor. Feedback to the inverting input of the ADA4096 makes the inverting input voltage equal to the noninverting input voltage, thereby ensuring a constant voltage of V_DRIVE across the bridge circuit.

Transistor Q1 is a BJT that has maximum breakdown voltage of 80 V, capable of dissipating 0.35 W at 25°C. The maximum collector current is 500 mA.

Instrumentation Amplifier

The AD8420 rejects the common-mode voltage generated at the bridge and only amplifies the differential bridge voltage, as shown in Figure 3. The AD8420 has rail-to-rail output voltage swing that is completely independent of the input common-mode voltage. This feature exempts the AD8420 from the restrictions caused by the interaction between the common-mode input and output voltages associated with most conventional instrumentation amplifier architectures. The gain of the in-amp is set at unity.

The input to the AD8420 has a differential mode noise filter (20 kΩ/1 μF/100 nF) with a 7.6 Hz bandwidth and a common-mode noise filter (10 kΩ/100 nF) with a 150 Hz bandwidth. Traditional instrumentation amplifier architectures require the reference pin to be driven with a low impedance source. Any impedance at the reference pin degrades both common-mode rejection ratio (CMRR) and gain accuracy. With the AD8420 architecture, resistance at the reference pin has no effect on CMRR. The transfer function of the AD8420 is

where:

The AD8420 differential input voltage is internally limited with diodes to at ±1 V from −40°C to +85°C. If the input voltage exceeds the limit, the internal diodes start to conduct and draw current. The current is limited internally to a value that is safe for the AD8420.

Reference Buffer

An excitation current of 210 μA, generated by the AD7793, passes through the 5 kΩ resistor, as shown in Figure 4. This generates the 1.05 V reference voltage that is buffered by the ADA4096-2. The output of the buffer drives the reference of the AD7793 and the AD8420. The circuit is ratiometric; therefore, the error due to the variation in the voltage across the 5 kΩ resistor, caused by the 5% tolerance on the 210 μA excitation current from the AD7793, is minimized. The buffered voltage reference also drives the amplifier that sets the voltage drive for the bridge transducer (Figure 2).

ADC Channel 1 Configuration: Bridge Type Sensor

Channel 1 of the AD7793 measures the bridge sensor output from the AD8420. The external V_REF (1.05 V) is used as a reference and therefore, the input range of the AD7793 is ±1.05 V centered on a common-mode voltage of +1.05 V.

ADC Channel 2 Configuration: Temperature Sensor

The second channel of the AD7793 monitors the voltage generated across a resistance temperature detector (RTD), driven by the 210 μA excitation current, as shown in Figure 5.

Although 100 Ω platinum RTDs are popular, other resistances (200 Ω, 500 Ω, 1000 Ω, etc.) and materials (nickel, copper, nickel-iron) can be specified. For this application, a 100 Ω DIN 43760 Class A RTD is used.

A 4-wire (Kelvin) connection, as shown in Figure 5, eliminates the effect of the RTD lead resistance. Note that 2-wire, 3-wire, and 4-wire configurations are also available using Links P3 and P4, as shown in Table 2.

Link P9 bypasses the RTD if temperature compensation is not required.

Output Coding

The output code for an input voltage on either channel is

where:

AIN = AIN(+) – AIN(−) = AIN(+) – V_REF

Gain is the PGA gain setting, and N = 24.

Voltage Supply Requirement

For the circuit to operate properly, the supply voltage, V_CC, must be greater than 6 V to provide a minimum 5 V drive for the bridge type transducer.

System Calibration

There are several methods for carrying out pressure sensor temperature calibration. For this application, a four-point calibration procedure was used. A good reference for the calibration procedure is Active Temperature Compensation and Calibration for MEMS Pressure Sensors with Constant Voltage, AN13-01, Silicon Microstructures, Inc., Milpitas, Ca.

Test Data and Results

System Noise

All data capture was taken using the CN-0355 Evaluation Software.

Two setup measurements were taken to capture the noise of the board. The first measurement, shown in Figure 6, was taken with the input to the AD8420 shorted, thereby measuring the peak-to-peak noise of the AD8420 and the AD7793. A sample of 1000 was taken; this resulted in a code spread of approximately 100 codes, which translates into a peak-to-peak noise of 12.5 μV, or 17.36 noise free bits for a full-scale span of 2.1 V.

The second measurement was made with a Honeywell NSCSANN600MGUNV gage pressure sensor connected to the evaluation board. This is a board mounted pressure sensor that is unamplified and uncompensated, and the voltage driver was set at 10.1 V. This test effectively shows the noise contributed by the whole system, including the transducer noise, as shown in Figure 7. A sample of 1000 was taken; this resulted in a code spread of approximately 120 codes, which translates into a peak-to-peak noise of 15 μV, or 17.1 noise free bits for a full-scale span of 2.1 V.

System Current Consumption

Table 3 shows the total current consumption of the system, excluding the current consumed by the pressure transducer.

The Honeywell NSCSANN600MGUNV pressure sensor has an impedance of about 3 kΩ, which adds approximately 3.36 mA to the total current in Table 3.

The current consumed by the system can be further reduced by driving the RTD with a lower current, such as 10 μA, and at the same time use a higher RTD resistance value, such as 1 kΩ.

Error Analysis for Active Components

The maximum and root-sum-square (rss) errors, due to the active components in the system for AD8420 and the ADA4096-2, are shown in Table 4.

Total Circuit Accuracy

A good approximation of the total error contributed by the resistor tolerances, is to assume that each of the critical resistors contributes equally to the total error. The two critical resistors are R8 and any of the Resistors R19, R20, or R21. The worst case resistor tolerance build up of 0.1% yields a total resistor error of 0.2% maximum. If rss errors are assumed, the total rss error is 0.1√2 = 0.14%.

Combining the resistor errors with the component errors from Table 4 yields the following results:

• Offset error = 0.365% + 0.1400% = 0.505%
• Gain error = 0.050% + 0.1400% = 0.190%
• Full-scale error = 0.415% + 0.1400% = 0.555%

These errors assume that calculated resistor values are selected, and that the only errors are tolerance errors, and that the voltage drive for the transducer is set to a gain of 10.1. The linearity error was tested over an input range of −500 mV to +500 mV, using the setup shown in Figure 10. The total nonlinearity was approximately 0.45%. The nonlinearity is primarily caused by the input transconductance (gm) stage of the AD8420.

The total output error (%FSR) is calculated by taking the difference between the measured output voltage and the ideal output voltage, dividing by the FSR of the output voltage, and multiplying the result by 100. The results of this calculation are shown in Figure 8.

Figure 9 shows a photo of the EVAL-CN0355-PMDZ evaluation board. Complete documentation for the system can be found in the CN-0355 Design Support Package.