Advertisement

Blog

Technique accurately calibrates multichannel bridge-sensor configuration

When a data acquisition system has a large number of slow- or moderate-speed sensors, it often makes sense to use a high channel-count multiplexer to minimize component footprint, cost and power. But if each of these sensors needs unique calibration (and they often do), this can cause difficulties in effectively using that multiplexer front end. By using a carefully defined algorithm, the calibration issue can be efficiently resolved.

For example, the Exar XR10910 sensor-interface IC includes an internal 16:1 multiplexer, offset-correction DAC, instrumentation amplifier and voltage reference, Figure 1. It can be used in many different applications which include pressure, temperature and strain gauge sensors (often used for weigh scales). The most common circuit for these applications is the bridge sensor.

Figure 1

XR10910 typical application

XR10910 typical application

Each bridge sensor has its own unique characteristics and has an offset voltage between its outputs at a nominal state. This offset needs to be corrected before an accurate reading can be made. With the bridge sensors outputs connected to a set of inputs on the XR10910, this offset voltage will be gained up and can be easily measured at the output of the XR10910, decreasing system sensitivity and performance. A calibration based on the attributes of the device must be performed to remove this offset.

The XR10910 offers 8 selectable gains from 2V/V to 760V/V. For applications that use only 1 gain setting, a simple DAC offset correction can be made to bring the offset to a desired value.

  1. Calculate Offset needed for desired output voltage (DVOUT)

For applications where multiple gains will be used “on the fly”, a DAC offset can be applied to the XR10910 so that its output at any gain will be as close to its internal reference voltage as possible. The XR10910 has an internal reference of ~1.5V, but cannot be directly measured, so we must mathematically find the reference voltage (VREF ) by performing a two-condition test.

  1. Find VREF :
    1. Condition1: With the bridge sensor outputs connected to the XR10910 inputs, set the XR10910 to Gain=2 and the DAC offset = 0x000. Measure the output voltage (VOUTG2 ) of the XR10910.
    2. Condition2: With the bridge sensor outputs connected to the XR10910 inputs, set the XR10910 to Gain=20 and the DAC offset = 0x000. Measure the output voltage (VOUTG20 ) of the XR10910.
    3. Calculate VREF :

    Next, find the voltage value of the XR10910’s LSB.

  2. Find the LSB
    1. With the bridge sensor outputs connected to the XR10910 inputs, set the XR10910 to Gain=2 and the DAC offset = 0x3ff. Measure the output voltage (VOUT Max ) of the XR10910.
    2. By using the output voltage VOUTG2 from the prior measurement, calculate for LSB:

    LSB=(VOUT Max-VOUTG2)/GAIN/1024

    Example:

    VOUTG2=1.459V, VOUT Max=2.673V

    LSB=(2.673-1.459)/2/1024=592.77uV

    Now determine if a calibration is necessary.

  3. Calculate the DAC offset needed to bring the output as close to VREF as possible at any given gain.
    1. Calculate VOFFSET: Subtract VOUTG2 from VREF
      1. If VOFFSET is less than 1 LSB, then no calibration is required.
      2. If VOFFSET is greater than 1 LSB, a DAC offset can be applied equal to VOFFSET (within 1 LSB). The DAC offset is applied prior to the gain stage of the XR10910. By doing this, the resulting output voltage will be as close as possible to the reference voltage at all gain settings.

      Below are examples of calibrated and uncalibrated ±10mV offsets across gain, Figure 2 and Figure 3. As you can see, the uncalibrated offset quickly starts to dominate as the gain increases and the output starts to clip the rail. Performing the calibration brings the output voltage as close to the reference voltage as possible across all gain settings.

      Figure 2

      Figure 3

0 comments on “Technique accurately calibrates multichannel bridge-sensor configuration

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.