SBOA443 March   2021 INA293

 

  1.   Trademarks
  2. 1Introduction
  3. 2The SAR ADC Switching Model
    1. 2.1 Acquisition Time
    2. 2.2 ADC Resolution
    3. 2.3 Sample Rate
  4. 3The ADC Charge Bucket Filter
    1. 3.1 The Filter Capacitor, CFILT
    2. 3.2 Output Filter Resistor, RFILT
  5. 4Output Filter Discussion and Design
    1. 4.1 INA293 With the ADC Switching Model
  6. 5Summary
  7. 6References

Output Filter Resistor, RFILT

The immediate solution to the above problem is to isolate this capacitance from the output node of the amplifier via additional resistance. The interactions between the output impedance of the amplifier and this RC network are discussed in Output Filter Discussion and Design, but it is beneficial to explore the nature of this filter as it relates to the ADC input stage. Examining the voltage on the filter capacitor, recall that for a typical RC network in the time domain, the voltage charges as a function of the RC time constant, given as

Equation 8. GUID-20210223-CA0I-G8K9-MV0C-CJLRPRNDDGMG-low.gif

where VINIT is the initial charge on the filter capacitor prior to the beginning of the sample and hold cycle, VFINAL is the final voltage that the capacitor charges to, and τC is the total time constant for the system with which the node will charge that includes contributions from both the RC network as well as the amplifier. The addition of resistance on the output reduces the amount of the current that can be delivered by the amplifier, and also filters higher frequency components of the signal based on the cutoff frequency formed by the RC response with CFILT. The goal then, is to ensure that for the given CFILT, the system is still able to settle to within 1/2 LSB within the sample and hold cycle of the ADC. For the above expression, this equates to

Equation 9. GUID-20210223-CA0I-BSJJ-TG99-CXLZH5WQ6W7N-low.gif

It can also be observed that the maximum magnitude between VINIT and VFINAL is simply VDROOP, or

Equation 10. GUID-20210223-CA0I-DNR4-9BW0-43TL2ZT2TK42-low.gif

Rearranging and substituing these into the RC equation, it can then be expressed that

Equation 11. GUID-20210223-CA0I-CPQR-GJZL-09PK04BGFK9R-low.gif

Solving this expression for τC, the relationship of VFILT is then expressed in terms of settling as

Equation 12. GUID-20210223-CA0I-240X-GNCH-N2545XX2PW6S-low.gif

However, since the CSA also contributes to this charge, an expression relating τC to both of these is needed. Since the CSA is approximated as a second order system, the shared contribution of both of these can be approximated using an root-mean-square (RMS) approach, or

Equation 13. GUID-20210223-CA0I-SJM5-FJTF-QSDGWQ0C3V6Z-low.gif

And given the assumption that the CSA is 4 times as fast as the RC network,

Equation 14. GUID-20210223-CA0I-CZD3-0XC6-0LR8NFGR0K5P-low.gif

Solving this equation for τCSA yields

Equation 15. GUID-20210223-CA0I-JFPW-G3PZ-VC05CLNVGQ9Z-low.gif

And substituting this back into the relationship for the RC time constant gives

Equation 16. GUID-20210223-CA0I-MMFD-LJKR-JTXVRNHPV376-low.gif

This approximation again allows the rendering of a ratiometric relationship between the RC network, the CSA, and the total network allowing the system to be solved. RFILT is then given as

Equation 17. GUID-20210223-CA0I-XZGJ-V4VV-MWJBVDWZ8F4H-low.gif

However, given the nature of the approximations above, an optimal value of RFILT may fluctuate slightly from this calculated value, so a range of values is examined, typically from 0.25*RFILT < RFILT < 2*RFILT.

Finally, with values for RFILT and CFILT determined, the bandwidth needed by the amplifier can be determined. Expressing the bandwidth of the amplifier in terms of frequency, and relating this to the RC time constant, this is given as

Equation 18. GUID-20210224-CA0I-06F3-VBRS-Q4KXQHDKNMG8-low.gif

This reveals that if the bandwidth determined here is greater than the bandwidth of the current amplifier, adjustments must be made to ensure accurate settling.

Note:

The mathematics discussed to this point are those performed in the "SAR ADC Drive" section of TI's Analog Engineer's Calculator. Therefore, discussion to this point is mainly for the reader's benefit to understand the theory behind these values, and quick calculation of these values may be easily obtained by use of this tool. The tool can also be utilized to help trim these values to determine an optimal acquisition time and filter for the amplifier's given bandwidth.