I've designed a circuit around the TI LMP91000, a configurable Analog Front End (amperometric potentiostat, for use with electro-chemical gas sensors). In addition the circuit includes a 2.5V reference (for VREF on the LMP91000), and an 18-bit ADC.
Once the sensor is installed in circuit, it's very difficult to remove it without doing harm. I've experienced a significant number of failures late in the testing process (e.g. during calibration), that I don't understand the root cause of yet, but which I suspect are related to undetected manufacturing defects in the underlying circuit (i.e. not a problem with the sensors themselves). The upstream testing that I do right now is limited to just establishing I2C communication with the LMP91000 and the ADC before installing the sensor. If I could identify problems earlier, I could potentially fix them before installing sensors.
Is there a way to temporarily attach an external circuit in place of the electro-chemical sensor (i.e. using pogo pins connected to the working, counter, and reference electrode inputs) to inspire the LMP91000 to output an expected voltage so that I can demonstrate the analog front end is working correctly before installing a sensor? What would the composition of such a circuit need to be?
Best Answer
I'd suggest looking at the sensor output characteristics and coming up with a circuit that will mimic the output, and use Pogo pins to connect this test circuit up to your main board. Once you can excite your circuit with known good signals, you have a chance to troubleshoot it. Until then, it sounds like you're flying blind.