SBOA551 June   2022 INA240

 

  1.   Abstract
  2.   Trademarks
  3. 1Introduction
  4. 2One, Versus Two Reference Pins
  5. 3Bidirectional Current Sense Amplifier Topologies
    1. 3.1 Single-Stage Difference Amplifier
    2. 3.2 Difference Amplifier Input Followed by Noninverting Output Buffer
    3. 3.3 Voltage Feedback Multi-Stage Difference Amplifier
    4. 3.4 Single-Stage Current Feedback
    5. 3.5 Current Feedback Multi-Stage Difference Amplifier
    6. 3.6 Isolated Bidirectional Current Sensors
  6. 4Options for Driving Reference Pins and Input Referred Reference Error
  7. 5Resistor Divider as Reference
    1. 5.1 Resistor Divider and Equivalent Circuit
    2. 5.2 Reference Source Impedance Error in Difference Amplifier
    3. 5.3 Reference Source Impedance Error in Voltage Feedback Multi-Stage CSA
    4. 5.4 Reference Source Impedance Error in Current Feedback Multi-Stage CSA
    5. 5.5 Reference Source Impedance Error in Difference Amplifier with Output Buffer
  8. 6Examples
    1. 6.1 Calculating Reference Source Impedance Error in Difference Amplifier
    2. 6.2 Calculating Reference Source Impedance Error in Voltage Feedback Multi-Stage CSA
    3. 6.3 Calculating Reference Source Impedance Error in Current Feedback Multi-Stage CSA
  9. 7Summary

7 Summary

For high accuracy measurement, the resistor divider should be avoided. Instead, consider using either a reference IC or buffered voltage source.

When a resistor divider is used to set the reference voltage for a bidirectional CSA, differential measurements with respect to the reference pin should be taken when possible. With differential measurement, the output error due to reference source impedance is canceled.

When differential measurement is not an option, the equivalent source impedance can cause both common-mode error and gain error. The coefficients of the error terms are approximately proportional to the ratio of the equivalent source impedance to the difference amplifier total resistance (for example, Ri + Rf). Larger output error can be expected with larger source impedance, except for the special situations where Vcm and Vref are equal in magnitude, then their resulting errors cancel.

It should be noted that regardless of how the measurement is taken – differentially or single-ended, the encroachment on output dynamic range is present. The issue can be quite serious with large source impedance.

The output error can be estimated. It is straightforward if the CSA takes on single-stage difference amplifier topology. Sometimes the resistance values are not listed in the data sheet, but an inference can be made from a simple bench measurement. For multi-stage topologies, bench measurement is not straightforward, some background information about the design of the chip is necessary. Such information may include the gain of each stage, the common-mode voltage of the output stage, as well as the resistor network value.

Spice simulation must be used with care. For single-stage difference amplifier, it is generally not an issue. However for multi-stage CSA, the source impedance effect is typically not modeled.

Some may ask - Should I use a large divider to drive the reference pin directly, then calculate and subtract its errors based on the error equations? The answer is “no”, and here are the reasons why, even if the loss of output dynamic range is not an issue:

  • The internal resistor values can be different by up to 15% among devices due to process variation
  • For multi-stage topologies, the gain of each stage may not be constant even though the overall gain is
  • The input common-mode voltage of the output stage is not strictly, accurately defined. Variation from the nominal value is expected.

Therefore, this error estimate should only be used as a guideline, helping system designers choose a proper reference driving method in the early design stage. It should not be used to subtract from actual measurements, in an attempt to cancel the error.