pH measurement, using glass electrode, is a technique which we read an electric potential difference without consume electric current. In fact, it is impossible not consume current. Every measurement system consumes some electric current. In this case, we need select an operational amplifier for a measurement system that consumes the smallest current possible. But, this is not enough to have good measurements. There are other sources that interfere with the process of measurement. For example, a bad PCB layout.
Let's suppose an environment with a high relative humidity. Water vapor will adsorbs on the PCB surface. If it happens in a high impedance area of the PCB, the water will disperse the charges like an anti-static bag. It will interfere at the measurement process.
To avoid this situation, we need to reduce the path of this high impedance terminal. One suggestion is sold the high impedance terminal from the operational amplifier closer to the central terminal of the BNC connector. A extreme situation is cut the central terminal from the PCB and sold it directly to the high impedance terminal of the operational amplifier.
Take a look inside of the red circle in the figure above. This seems like a mess, but this is the better way to connect a high impedance terminal. The surface is the smallest and it reduces the disperse process of the charges. The air is a better insulator than the PCB surface.
How to do this?
It is simple. First, remove the operational amplifier and rise the terminal. Second, rise the central terminal of the BNC connector. Third, reconnect the operational amplifier. Then, adjust the length of the terminals, cut the excess. Finally, make a neat sold without tips. Done.
Is this really necessary?
It depends the quality of you PCB. In my case, the PCB layout has a short path to connect the central terminal of the BNC connector to the high-impedance input terminal of the operational amplifier. In addition, I covered the copper surface with a protective varnish to avoid the copper oxidation. In my tests, I didn't notice any difference between the two process during the measurement process. I didn't make a test with a PCB without varnish or with the copper oxidized.
In this project, I used a J-FET operational amplifier (TL082 or TL072) with 1012Ω of input impedance (or resistance for a DC measurement). This suggestion of soldering process could make difference for higher impedances.