Hello everyone, I have a problem with the AD8420 in-amp in a new electromagnetic flow meter design. Please refer to the attached image of a section of the circuit diagram.
The AD8420 is powered from a 3 V supply with the REF pin connected to mid-rail at 1.5 V. The inputs are AC coupled and there are 150k resistors to the 1.5 V supply to provide a path for the input bias currents. The gain is set at 151 with a 150k feedback resistor and a 1k resistor (R17) between the FB pin and the 1.5 V reference voltage. A 1k resistor is placed between 1.5 V and the REF pin. With this configuration one would expect the output to be at 1.5 V, but instead it sits at about 2.8 V. I measure (and calculate) about 8 mV across R17 which implies a corresponding 8 mV differential voltage at the input. However, the inputs are at exactly the same voltage.
I have tried to tie the two 'bias' resistors (R25 & R26 in the circuit) to ground; I have tried to remove the RF filter capacitors; I have tried changing the gain resistors; I have even tried cutting the trace between the output of the AD 8420 and the LTC6910. Nothing helps. The output stays near the upper rail. It is not a faulty device - three different prototype boards have the same problem.
I am now stuck, and I hope someone can shed some light on this problem.
See the input range specification in the AD8420 datasheet. The inputs (as well as REF and FB) require 2.2V headroom from the positive rail. 1.5V is out of range for a 3V supply.
You might be able to bias the inputs to 0V instead of 1.5V, but it may be more difficult to bring REF down into a working range and still get the output swing that you want. If that will be a problem, you might consider switching to the pin-compatible AD8237, which is recommended by our in-amp tool, along with AD8236 (not pin-compatible but notably for EM flow, it has very high input impedance).
Another option that would work for the AD8420 would be to pull REF down to an intermediate voltage such as 0.5V and then use a pull-down resistor from FB to 0V in order to level-shift the output range back up so the output is centered around 1.5V. A similar idea is described in this article: How to Stay Out of Deep Water When Designing with Bridge Sensors
Here's a sketch of this option:
Thanks for your advice. I will try it, but I suspect that the problem is the 2.2V headroom requirement on the inputs as well as REF and FB.
Thanks for your feedback. I will try the circuit you gave, but I think the easiest solution now is to use the AD8237. I already have 10 trial boards and I cannot afford to change the layout at this stage in order to use a device that is not pin compatible.
I now have the opportunity to make some changes to the electromagnetic flow meter board because of another requirement, so I can use the AD8236 instead of the AD8237. Which would you recommend?
In my experience with EM Flow sensors, the output resistance of the sensor can be very large, more than 1MegΩ in some cases, and a high input impedance amplifier can improve the performance in such cases. AD8236 has a very high input impedance, but it is up to you if the noise and bandwidth of the AD8236 will be good enough in your design.
I hope this helps.