INA105: A/D converter with unity Gain buffer (1V/1V)

Hey Richard,

Sorry for the delay in response, SOIC opens many doors. You could use an integrated difference amplifier, however, the outputs will not swing all the way to either rail, so you will begin to lose accuracy close to 0V. Additionally, there is not a 1/6 ratio difference amplifier, the closest you can get without exceeding the ADC input is 1/10th. Therefore, you would not have accurate sensing in the region from 3V-GND on the input. However, you would have great CMRR and offset error in the linear operating region of the device. The below picture is from the electrical characteristic table of the INA143.

You can use the existing resistor divider and use a rail to rail input and output op amp, however, these are not perfect either, but you will lose accuracy much closer to your V- rail, however, if you select a CMOS amplifier, you will only lose linearity in the region from about 600mV to GND, assuming a 1/6 voltage divider and that the region of high linearity goes to 100mV from the rail. You will still get an output with these amplifiers all the way down to a few mV from V- to tens of mV from V-, but the linearity will begin to decline.

This inaccuracy close to V- can be overcome with a small negative charge pump like the LM7705. This will allow the op amps to swing close to V-

In this case, I would recommend a device like the TLV9002, or the OPA310.

If you know the time of acquisition (dictated by sampling rate), the value of the sample and hold cap, you may use this information to calculate the bandwidth of amplifier that you need for this application. The Analog Engineers Calculator: https://www.ti.com/tool/ANALOG-ENGINEER-CALC has a calculator for this!

This had arbitrary values assigned for acquisition time and Csh, but based on your accuracy requirement and your full scale range, I'm estimating 8 bits of resolution on the ADC.

It's also worth noting that I would recommend selecting an amplifier with about 5x the bandwidth for comfortable margin. In production, there is a large variance in UGBW of amplifiers, so guardbanding this parameter heavily is recommended.

Best,
Jerry