Determining input resistive loading error, especially for offset and drift is not currently a theoretical procedure and does require a bench verification due to the highly non-linear impedance of internal switching capacitors. The general procedure is straightforwad, in that the engineer can sweep V_OUT/V_IN over temperature without and with R_EXT (and a stabilizing C_DIFF) and calculate total gain, offset and drift errors using standard linear interpolation. The error generated solely from R_EXT is calculated by calculating the change in total errors.

There are two methods a designer can use to understand the realistic error to expect.

Method 1 - Sweep Input with Precision Differential Voltage Source

This method requires emulating the shunt resistor's voltage drop (V_SHUNT) with a precision DC voltage source (V_IN) capable of accurately driving milli-volts. Additionally, this source can have a 4-wire force and sense capability.

While this can seem the simplest approach, there is a downside concerning the inductance of long input wires going into the CSAs input pins (IN+ and IN-). As noted, a capacitively-coupled CSA can have complex capacitive switching network at it front end. These capacitors can be constantly charging and discharging with current supplied from the bus. The average current is small (nano-Amps), but the peak transient input bias current can be larger.

Any input inductance can load down these capacitors and create delays causing significant device error. Usually this is not a problem because the input traces from R_SHUNT to CSA input pins is not long enough to generate significant inductance. Additionally, inductive loading can be negated with a small input differential capacitor (C_DIFF) as shown in Figure 4-1.

Overall, if using this method make sure the V_IN source is a 4-wire force and sense connection and that CSA has a C_DIFF > 1-nF potentially more depending on how long cabling is and how high the ambient temperature is.

Method 2- Sweep Input with a Monitored Current Source and Calibrated Shunt Resistor

This method is the actual sensing of current across a soldered down shunt resistor with soldered down CSA circuit. It requires more effort up front, but can achieve more precise results. Also, it does not require negating input pin cable inductance because input traces can be small on realistic PCB.

The downsides with this method is it requires knowing the approximate shunt resistor (R_SHUNT) to be used in system and being able to solder it onto the CSAs EVM or prototype system PCB.

Another challenge is having an accurate ammeter to measure load directly. If testing requires larger currents, consider using precision shunt that is temperature controlled and monitored with accurate voltmeter.

Load current can be controlled simply with a variable resistor or rheostat pulling current from a stand voltage source emulating the bus voltage. Electric current sources, which can be more precise, can be used if desired and available

One convenient advantage with this method is that the shunt resistor does not need to be measured for gain calibration because external loading error (e_EXT) is delta in error from R_EXT=0-Ω to R_EXT>0-Ω.

If engineer chooses to measure R_SHUNT over temperature for gain calibration and accurate understanding of circuit error, then R_SHUNT can first be soldered down onto the PCB and measured with a precision 4-wire, Ohm-meter at the nominal, maximum, and minimum operating system ambient temperatures. Care needs to be taken for measurements to settle. Additionally, any resistance in parallel (including the CSA can be removed/negated) so the R_SHUNT measurement is not affected. Once R_SHUNT is calibrated, you can negate its tolerance by calculating .