Electronic – Interfacing a flow control sensor with a microcontroller

These problems are hard to diagnose. The first thing I would do is swap the level and pressure ADC inputs to make sure that there isn't some sort of software issue. Even if you've looked at the code a thousand times I would do this.

The impedance of your sensor is pretty high to feed directly to the ADC input. I think you should add a unity gain, non-inverting buffer between the sensor and the ADC input.

Bypass capacitors may be needed. There's not enough information to tell where they should be added or what their values should be. I would be sure to bypass the 5V supply for the sensor. You could add a small capacitor to the output of the sensor as well, assuming that the pressure is not changing rapidly.

You could also try slowing the ADC clock rate to give the sample-and-hold more time to charge. Sometimes putting the processor to sleep during a conversion can reduce noise as well.

Looks good to me. The inverse diode D1 is a good idea. If you have a minimum of 12V available you may wish to reduce R2 somewhat. This circuit has a threshold of maybe 2V, you could easily halve R2 or double R1.

In the case of momentary extreme over-voltage, the base-emitter voltage (forward biased) will not rise above a volt or so, even with 100mA. It looks like another diode in inverse parallel to D1. One of the advantages of a BJT in this application. The limitation is more likely to be the voltage rating of R1.

If you want to consider sustained overvoltage, you may have to consider the power rating of R1. If some idiot connects it to the mains (we can usually assume that about 240VAC is the most voltage idiots will have access too- idiots with access to higher voltages are sort of a self-eliminating problem) then R1 would dissipate almost 6W, so it would have to be a physically large part. You could solve that issue by increasing the value of R1 so that a smaller part could be used.