I finished and tested with buffers this little differential pH probe on a perfboard, it works, 3 points calibration is linear. The ground electrode is a stainless steel rod wired to the GND of the circuitry, it's dipped in the liquid.
Also, it's a mix of my initial design (not differential) and some reference design found on the web. I am not from the EE world...
- My first concern is about the RC filter on the opamp inputs (LMC662), I use classic resistors 4.7Mohm, useful or useless as the probe is now differential, should I use other kind of resistors? 2.2nF capacitors are ceramic.
- The Analogic-Digital-Converter does the differential measurement between the two opamp outputs. Maybe the RC filter could be improved for the opamp outputs? I kept the value I used for non-differential measurement.
- Having the GND of the circuitry connected directly to the liquid can cause issues? like ground loop if I put another sensor?
(-Vs) comes from a LT1054C voltage pump, with tantalum capacitors.
Any criticism welcome.
Edit 1: schematics updated.
Thank you for your comments @Brethlosze
I am fine with 0.05 resolution and so far, it has been working well. I use an infinite impulse filter to smooth the data with a small period. Truth is that differential mode is little more noisy than non-differential (aka when reference electrode is connected directly to ground). I chose a 16-bit ADC because it is very cheap (check ADS1115 chip) and widely available. The electrode is industrial rated but basic with BNC connector, impedance <=200Mohm.
OK for the ESD worry. Until now, mine has been the EMI/RFI generated by the motor starting/stopping. For ESD, I could replace a plastic valve with a stainless valve and connect it to earth ground – and liquid is overflowing from the tank continuously through an outlet. In a previous experiment with the classic pH probe (classic means reference electrode connected to GND), all has been working fine for 2 hours then I got a static error: pH jumped to 8 instead of 5 and oscillated between 7.5 and 8.5. Using a sample bucket, moving the probe inside and using a conductor, I figured out it was a ground loop problem but I never found the origin. Maybe the AC motor upon heating? So, I start researching for differential reading.
As you said, I need to amplify the voltage difference between the reference and the pH probe. Does a RC filter makes sense if I am reading a difference between the two potentials (and not in respect to GND)? I guess they have the same noise? the noise component cancels itself while subtracting? I see here a risk of producing additional noise with RCs instead of cancelling it.
Differential mode is actually a little more noisy than classic mode. A few screenshots:
note: for ph4-10 range, readings go from -5000 to +5000
On additional comments:
- by ground electrode, I mean a stainless-steel rod which is immersed in the tank in the neighborhood of the probe (1 to 2 in)
- EMI generated by starting a AC-motor produces a spike in pH reading.
Edit 2: August 9th.
Noise is electrical, I wouldn’t expect more as long the probe is in the buffer solution. Extra noise comes when it’s put inside the tank with motor running after a couple of hours. I won’t be able to test before a couple of weeks.
The same for measuring GND/neutral lines of the motor. The cable came with the probe, a flexible EMT conduit is a good idea in my context. The length of the cable has already been reduced as much as possible. In order to make the problem as simple as possible, I use a simple lead battery as power source for the pH probe circuitry. I read your paper about chopper circuit; I am afraid it’s way beyond my EE skills.
W.r.t FIR filter, I prefer to stay with IIR with require low computational power. I am not displaying the results but they are processed by the microcontroller to start/stop caustic/acid dosing pump in my system. If I am not wrong, the electrode input have a FIR actually (RC filter + opamp = active lowpass). W.r.t RC filter, you are suggesting an anti-aliasing filter like this one on each electrode inputs?
From previous comment, I modified my schematics with sort of RFI filter at the input (and output in the 2nd) with capacitance between electrodes and a small amount with GND. Not sure if it has to be done on both inputs and outputs of the opamps. I can find some mylar polester film capacitors for the inputs and test.