High-performance sensing applications often use discrete analog components to interface a sensor with a microprocessor or microcontroller. Understanding how the sensor and the analog components interact is a critical element in developing an effective analog design.
This article reviews a sensor signal path solution developed for a typical pressure sensor used in industrial, medical, and high-end consumer applications. We discuss critical analog-to-digital (ADC) and operational amplifier (op amp) parameters and review a new online Web-based tool—WEBENCH Sensor Designer—that matches sensors with analog components and enables the designer to create a sensor-analog signal path that can be modified to a specific application need.
Pressure Sensor Overview
Figure 1. Bridge pressure sensor (Courtesy of GE Sensing and Inspection Technologies)
While performance of a MEMS-based silicon pressure sensor will vary among manufacturers, a maximum nonlinearity of ±0.1% (BFSL) at 25°C is available from a number of suppliers, including All Sensors, GE NovaSensor, and Measurement Specialties. A signal path solution provides the interface between the pressure sensor and a processor or microcontroller. The sensor and signal path components must provide an overall performance that meets the needs of the application. The analog signal path error is governed by a number of parameters. The following discussion will review major op amp and ADC error parameters in a signal path solution and explain how the WEBENCH Sensor Designer enables optimization of the signal path performance to meet the needs of the application.
Error over the full operational temperature range is much larger than the error at 25°C. This is common for most bridge sensors. The sensor in Figure 1 has thermal offset and thermal sensitivity errors of ±0.5% (each) over a temperature range of 0°C–60°C. Offset is trimmed to ±2.0% of F.S. span and the output voltage of this sensor is between 75 mV and 150 mV at F.S. pressure, at the rated supply. Companies may also add a gain set resistor as part of the sensor assembly to correct for the output when the sensor is combined with a differential amplifier. This configuration is shown in Figure 2. The output tolerance after the amplifier stage is ±1.0%.
Figure 2: Pressure sensor with gain set circuit
Since the MEMs bridge sensor provides a differential output, one possible amp-ADC configuration is to take the output of the sensor into a dual op amp and then into a differential input ADC. An alternative is to use an instrumentation amplifier configuration and a single-ended input ADC. In this case the second stage provides common mode rejection. Our sample pressure sensor also requires a constant current supply. The schematic in Figure 3 shows an instrumentation op amp configuration with a constant current drive. (This is a typical schematic that can be generated by the WEBENCH Sensor Designer Web tool.)
Figure 3. A sensor signal path generated by WEBENCH
A MEMs pressure sensor has excellent repeatability and very good performance over a narrow temperature range. Depending on the application, the sensor may have a performance variation of ±0.1% of F.S. span to as much as ±1.0% and this is a factor in selecting the appropriate amplifier and ADC to use with the sensor.
In general, the amplifier parameters we need to consider in sensing applications are: closed-loop bandwidth and gain; offset voltage (Vos); noise (flicker, flatband, and thermal); common-mode rejection ratio (CMRR); and power supply rejection ratio (PSRR). For the ADC we need to consider signal-to-noise plus distortion (SINAD), which is also expressed as effective number of bits (ENOB); integral nonlinearity (INL); differential nonlinearity (DNL); offset error; gain error; number of codes used for calibration; and sample rate. Evaluating how these parameters affect signal path performance is a central part of the WEBENCH Sensor Designer tool.
In dynamic ADC applications performance is usually governed by SINAD. However, in our pressure sensor example, the output frequency of the sensor is near DC and the governing performance specification is the total unadjusted error (TUE).
TUE is the combination of offset error, gain error, and INL. Since the sensor, amp, ADC, and processor signal path will be calibrated together, the offset and gain errors are eliminated as part of the calibration process, leaving INL as the governing parameter. INL is specified over temperature and the INL of a discrete 12-bit ADC is usually ±1 LSB or ±1 part in 4096, equivalent to ±0.024% signal accuracy over temperature.
In the case of low-resolution ADCs, the ADC may have an INL less than ±1LSB. An 8-bit ADC (256 codes), for example, has an INL of ±0.3 LSB. As a result, the limitation on calculating signal accuracy for the 8-bit ADC is the resolution of the ADC itself, not INL. A good rule of thumb for calculating signal accuracy is the following: if the INL is greater than ±1LSB use INL. If INL is less than ±1LSB, use 1/2n where n is the number of bits in the ADC. The Web tool takes these considerations into account.
The amplifier parameters we need to consider for sensing applications are closed-loop bandwidth and gain, Vos, noise, CMRR, and PSRR. In our example, the sensor is a DC output device and gain bandwidth is not an issue. The sensor has a nominal output of 100 mV and the output after the last amplifier stage is 3–5 V, depending on the application. This makes the total closed-loop gain of both amplifier stages 30–50 V/V.
The sensor, amp, ADC, and processor will be calibrated as a single system. This will correct offset and gain errors due to component mismatches in the circuit and correct for error due to the Vos of the amplifier at 25°C.
On the other hand, the amplifier's Vos variation over temperature and noise will affect circuit performance. Vos over temperature for a typical precision op amp (e.g., the LMP7732) is ±5 µV/°C. The sensor has a temperature offset error of ±0.5% of span for a 30°C change in temperature. Since the sensor has a span of 100 mV, it will have an error over temperature of ±500 µV compared with an op amp Vos of ±150 µV over the same temperature range. This is <30% of the sensor's offset error over temperature and this may or may not end up being an important factor. We need to evaluate the overall signal path performance and if this does not meet the needs of the application, we may want to look at an amplifier with a lower Vos variation over temperature.
Amplifier noise can be a significant source of error in low-frequency applications. Total amplifier noise comprises three elements: 1/f noise, flat-band noise, and thermal noise. Total amplifier noise is expressed by the formula in Equation 1:
Calculating the value of these three elements can be tricky and time consuming. Depending on the closed-loop bandwidth of the amplifier, the 1/f noise contribution may or may not be negligible. The magnitude of the flat-band noise and the thermal noise can be limited by reducing the closed-loop gain and the bandwidth of the amplifier stage. Using lower-value resistors in the design can also decrease the impact of the thermal noise. WEBENCH (see sidebar "A Web-Based Aid to Sensor Signal Path Solutions") evaluates these factors along with the other amplifier error sources such as Vos variation over temperature. It evaluates overall circuit performance and provides a detailed list of error sources in the circuit. Specify a different overall circuit performance and the program will generate a solution based on the new requirements. Alternatively, the designer can use the detailed list of error sources to select a different amplifier or ADC and generate a solution based on the specified part(s).
Selecting the proper analog components and determining the overall signal path performance can be time consuming. Prototyping to verify that the design is robust and meets the desired performance requirements takes even more time. Understanding how a particular type of sensor and analog components interact and affect overall signal path performance is the critical factor in developing a superior design. Providing tools and support to be able to do this quickly both reduces design time and shortens product development cycles.
|A Web-Based Aid to Sensor Signal Path Solutions |
National's WEBENCH Sensor Designer is actually a set of Web tools designed to create signal paths for the most popular types of sensors. Currently the tool has solutions for popular pressure sensors, load cells, thermocouples, and optical sensors. By focusing on specific types of sensors, the tool can be configured to take into account critical sensor, ADC, and op amp parameters that are important for the particular sensing application.
Each WEBENCH Sensor Designer tool follows the same basic guidelines:
For a pressure sensor design, you select a sensor from a list of manufacturers and then customize the critical sensor parameters to the specific application (Figure 4).
Once the requirements are established, you can create a design. The Web tool evaluates critical op amp and ADC parameters and generates a design schematic (Figure 3) with a bill of materials. The program also provides a detailed summary of the analog signal path's error components (Figure 5) as well as overall performance at 25°C and over temperature (Figure 6). The design can then be modified by selecting a higher or lower performance amplifier or ADC (Figure 6).
After the design is completed, the "build it" option (includes board and components) allows you to validate it and move to prototyping.