4-20mA Sensor Current Sensing: AD8211 Shunt Amp vs Direct Resistor, Temp Drift Issue

Hello all,
I have been designing a PCB with a current sensing circuit in order to sense the output coming from a 4-20mA industrial level sensor. The design that I have now incorporates a power supply that powers both the amp and the sensor at 12V and a current shunt amp whose output is then fed to an ADC (the ADS1115, datasheet linked below) and then read by an MCU.  The main issue that I am observing is that temperature swing affects the reading an extremely large amount. This issue could be a combination of a few things: the power supply, the shunt amp, and the sensor. The main issue that I am focusing on is the shunt amp.The shunt amp that I selected (the AD8211, datasheet also linked below) says that the drift for temperature is ~15ppm/Celsius, however when I tested the amp by itself (on a breakout to isolate it from the rest of the components) I observed a swing of over 100 ppm/Celsius. I am wondering if it might be possible to remove the shunt amp completely and just add a resistor and then adjust the gain of the ADC to match.  The temperature swing that causes this is a daily temperature swing, from say 25 Celsius to 50/55 Celsius. The other avenue, rather than removing a component, would be to select a different amp that has a smaller temperature swing. One that I have found is the AD8217 (datasheet also linked below), whose ultra-low drift over the operating temperature is attractive. If anyone has any insight it would be much appreciated, as I am trying to solve this problem piece-by-piece. If I have missed anything in my description or if you require further information I will do my best to provide answers. I appreciate any advice that you can offer, and I respectfully look forward to conversing with everyone!

Here on power there are a lot of articles about that, such as https://www.powerelectronicsnews.com/400-khz-hall-effect-current-sensor-with-5-kv-isolation/ please check 

SaraI have several questions that would probably be answered by a schematic with the values you are using. However, let me start by saying that I have designed 4-20mA current loops for nearly 30 years and I have never seen a current shunt amp used. The current shunt amp is intended for much higher currents and different circumstances (like voltage outside the supply rails) and I suspect your circuit is falling into a marginal area of operation of the device.When interfacing to an ADC I have never seen a circuit that doesn't use a resistor (just like your solution does). Because the 4-20mA concept is so old, it most often is a 250R resistor (or 249R9 if you system allows for calibration) that gives a 1-5V output that is fed directly to the ADC. Of course depending on your ADC input voltage range you can scale the resistor down.It goes without saying that you should use a low tempco resistor (at 20mA input).  For very low input voltages, I have followed the resistor with an instrumentation amplifier, but it doesn't sound like this is a problem that exists for you at the moment. Some designs may include some low pass filtering although a current loop is inherently noise resistant. In some systems, like 4-20mA with HART (an AC signal imposed on the 4-20mA loop) the resistor is expected to be in the range of 250R.So go ahead, use the resistor connected directly to the ADC which should be configured for a single ended input. You may want to add current limiting resistors as input protection as described in 10.1.3 of the ADS1113 data sheet.

By the way, what is the tempco of your present shunt resistor?

To get the whole thing into perspective, using a sixteen bit converter and you are talking about 16ppm, and that has to be over 30 degrees C, so something has to give or you are talking about some expensive parts and quite a bit more design work. It might help to know what the current loop sensor  at the other end is, does  it justify such extreme measures at the receive end?cheers,Richard