The schematic shows 2 transimpedance amplifier circuits on the first stage, for reference and sample photodiodes. The outputs are then connected to sallen-key butterworth circuits for low pass filtering. The output of these 2 are then connected to a differential amplifier to obtain its difference.

When testing out stage 3, instability arose from differential amplifier output and the problem was determined to lie with the FB pin of AD8237. I had tried to switch between different capacitors in order to improve stability but nothing worked. Could anyone help to suggest a solution?

If more information is required ill be more than happy to share!

I have attached .asc file for simulation but I had to download OPA333 spice model in order to run the component in LTSpice.

Some additional information that may help: Im using a single lithium ion battery of 3.7V connected to a voltage regulator to output 3.3V and bpw21r photodiodes to replace input sources. I used a single battery as this circuit design is meant to be placed in a watch and thus is the most practical battery option.

 

When testing out stage 3, instability arose from differential amplifier output and the problem was determined to lie with the FB pin of AD8237. I had tried to switch between different capacitors in order to improve stability but nothing worked. Could anyone help to suggest a solution?

To clarify, the testing had been done on a breadboard.

 

hi otters,
Cleaned up the circuit layout.
To test the Frequency response, I have set both the first OPA's for a voltage gain of 1.
Running an AC analysis shows a very unstable response.
Note used OPA340 for the 1.65V

The second plot shows the result when injecting the AC1 analysis into bot inputs.

E

 

If the problem is that the final output voltage is unstable when it should be very stable, the problem quite apparent to me. Consider that there are a total of four amplifier stages between the voltage of interest and the differential input, AND almost 20 resistors in the signal portion of the circuit. Also, there is one power supply that may not be "Perfectly stable and totally noise-free powering everything. I did not calculate the gain in each channel, but I am suspecting that the single final differential amplifier might be able to provide all of the required gain.That will probably reduce the noise quite a bit.