Hello,

I am designing a circuit to measure sea water conductvity. The measure of conductivity is done using a 4 electrodes sensor. A constant current should be applied to 2 electrodes and the voltage is measured on 2 others.
The signal driving the current pump is a sine wave at 100Khz 700mV peak to peak, oscillating around 0V (generated by a circuit using AD9833).

The howland pump is as follow:

It is powered by +5 / -5v symetrical power supply.

If the green part is replaced by a resistor (100ohms for example), the oscilloscope connected at blue point shows me a nice sine wave, centered on 0V with an amplitude almost proportional to resistor value (constant current). No problem there, it is the expected behavior.

If the green part is replaced by the water sensor, the oscilloscope still shows a sine wave but not centered on 0v anymore but rather centered on -3V (exact voltage depends on sensor contacts area)! The problem is that I need a sine wave centered on 0V to avoid electrolysis of sea water.

I have tried to add a capacitor in series with the sensor to filter out the DC part but it fails completelly; it even modifies the input signal.

Any idea would be greatly appreciated.

Best regards,

Olivier

 

What is the "water sensor"?

 

The center voltage will depend on the supply voltages to the opamp.
Instead of R6 connecting to analog ground, connect it to an adjustable voltage so that you can vary where the sine wave center is.
And since you do not show the op-amp supply , and you get negative 3 volts, I am guessing it runs on + and - 5 volts.
Let us know, it matters.

 

What is the "water sensor"?

Hello, the water sensor is made of 4 titanium conductors of a few millimetres, evenly spaced by a few millimetres. Only 2 of the 4 conductors are used at this step; the 2 others will be used to measure the electric potential in water created by the 2 firsts…

 

The center voltage will depend on the supply voltages to the opamp.
Instead of R6 connecting to analog ground, connect it to an adjustable voltage so that you can vary where the sine wave center is.
And since you do not show the op-amp supply , and you get negative 3 volts, I am guessing it runs on + and - 5 volts.
Let us know, it matters.

Yes, the opamp is supplied with +5 -5. But as said, with a resistor in place of the sensor, the center voltage is on 0v.

 

Hi olilam,
Please post a type number or link for the sensor, so that we can determine a possible cause of the Vout offset.
E

 

Hi olilam,
Please post a type number or link for the sensor, so that we can determine a possible cause of the Vout offset.
E

Thanks for your interrest. It is not a commercial sensor. It is a home made design:

The grey parts at the bottom are titanium plates, the orange part is PETG plastic including the cables, filled in with epoxy for insulation. The titaniums parts are 5mm width, 1mm thick and are extending 2mm appart from PETG. They are 7mm appart from each others.

As soon as I experienced the issues described before, I tried with other designs. I tried with a 4 pins header in line, a PCB part with a SOT foot print, etc...

When changing the sensor, the measured voltages changes too but the phenomenon stays the same: the resulting signal is offset and never centered around 0V.

 

Hi oli,
The problem most likely due to the high resistance of the sensor.
Have you tried adding a 10k resistor in parallel with the sensor.?

Try it and let us know the result.
E

I will try your circuit in the LTSpice simulator.

 

Hi oli,
The problem most likely due to the high resistance of the sensor.
Have you tried adding a 10k resistor in parallel with the sensor.?

Try it and let us know the result.
E

I will try your circuit in the LTSpice simulator.

I will try adding the resistor as you suggest and will keep you posted.

 

hi oli,
Using LTSpice shows the OPA Vout is clipping at the peaks.
With and without a 10K simulated sensor load
E

 

Hi oli,
Using the simulator suggests that there could be a common ground referencing problem, please post a complete circuit showing the AD9833 and power supplies interconnection.

Using various simulation ideas when using your basic OPA circuit, no way can the OPA Vout be at -3V

E

 

Thanks for your time.
The signal is clipped because of high value of R7. If R7 is lower (like 100ohms as I explained in first message), everything is fine. The problem is that I suspect the sensor in water is not purely reactive... I will take a few screenshots of the scope to better illustrate what happens.

 

Here is the completed circuit with wave generator part:

The C5 capacitor and R1 resistors are used to center the sine wave around 0V.
The scope screenshots will follow soon...

Thanks!

 

Here is a screenshot of scope graph showing the signal at the sensor in blue and the signal before the howland in yellow when using 1K resistor in place of the sensor:


The same with the sensor in place of resistor, showing the issue:

The offset is indeed not as big as the one I reported (I don't remember what were the test condition when it was so big as -3V) but still, the signal is centered around -1.5V...

And as suggested by Eric, the same with a 1K resistor in parallel to the sensor:

This seems not too bad! No offset anymore! I will investigate further with diferent resistor value to achieve best results.
Thanks a lot!

 

hi oli,
Your circuit is super sensitive to any DC offset of the 100kHz signal
The left side circuit has no DC offset on the input 100kHz, the right side has a -10uV offset, note the effect on the Vouts.
I value of 10uV can easily be created by incorrect grounding of the common nodes on the circuit.
E

 

Because of surface dioxide layer, titanium is not good as sensor electrodes.
They may be made from platinum, gold, stainless steel, nichrome or
other metals or alloys, which have bare metal surface.
Then equivalent circuit of sensor is resistance of water R1, connected in series
with double layer capacitance C1 between water and bare metal surface
(about 20 μF at 1 sq. centimeter of electrodes area).

Measurement circuit:

 

Because of surface dioxide layer, titanium is not good as sensor electrodes.
They may be made from platinum, gold, stainless steel, nichrome or
other metals or alloys, which have bare metal surface.

Thanks for info! Very interresting...
I am using titanium because of its very good resistance to sea water corrosion. I will try finding another material then.