Salvage sensor, no datasheet, not cheap

Ok so first i’ll give you a little background before I get into this.

I’ve been into the world of microcontrollers for maybe 6 years. I am a diy addict, and I have have a minimum of 1 raspberry pi and 1 arduino in any given room in my house. I don’t post on forums hardly ever – I know how to use a search engine. I am a huge lurker. Sometimes I just can’t find an answer or the information I do find is way over my head. Seldomly I give up. The following problem I have picked up and given up more times than I can count. Now I’m at the point of either putting it to use or getting rid of it.

The product I am referring to can be found on ebay or anywhere else, I won’t link to it. It’s an APEX controller by Ecolab. It is a industrial chemical dosing and monitoring system for a commercial dishwasher. I have seen it in use several times passing through various locations, and I always thought of all things that are in there that I could repurpose. Then one day I needed an accurate peristaltic pump that needed to be very particular, not something I could buy prebuilt. I purchased this apex machine for next to nothing, the seller barely made enough to cover his shipping. I went on to complete my project with lots of leftovers from the device. Over the course of a couple years I have used the other various sensors and switches and actuators for various things. Now I have a use for a probe that came with this device. I need a longer term conductivity sensor, and this product has one with it. It is a toroidal conductivity sensor. If I can use it great! If I can’t well then I won’t continue this any further. An EC probe for continual use is in the hundreds and they need to be replaced every year or two. At that rate, I will just continue using a refractometer. The Apex is able to detect total dissolved solids and electrical conductivity using this probe. As I understand, that information with a little math can be used to determine salinity.

Here lies the problem. I cannot find any brand names or part numbers on the probe. So no data sheet. All i have is a hermetically sealed probe with 7 wires coming out. I measured the resistance and found that 2 pairs of wires have continuity. I snipped the connector off and peeled back some insulation and found that those two wires are shielded. So, my assumption is that (please see photo below) the red and black go to one toroid, and the green and white go to the other. The ground being obvious and the other two wires I would assume are for a temperature sensor. The information I could find about toroidal sensor implementation mostly touches on a broad overview of how they work. I did not find any hardcore examples, or small projects/ diy blogs (lol - or instructables) with one of these probes being implemented. That doesn’t surprise me given the high cost of these probes. This gives me more reason to use it.

My question comes down to how would I go about reading from this probe, and am I correct to assume this pinout?

cj: ...I won't link to it...

Please give links to it.

The bare wire is likely just a runner for the foil shields. The power wires are usually the red/black pair. Because they're shielded, that assumption is questionable. If you value the device, do nothing until you find a spec sheet. If you don't care then put 5v, and later 12v, on the red/black pair and look at all others with a scope or meter while changing the environmental conditions around the probe.

This looks something like the sensor you might have. If so, then the two shielded pairs are the toroids and the brn/yel pair are likely the thermo sensor. How you drive the coils and interpret the reading is unknown.

Koepel: Please give links to it.

Arctic_Eddie: If you don't care then put 5v, and later 12v...

Should I be inducing ac into red/black? Also that was a lot of information on that patent, nice find. So if I throw 12vac into one toroid, the other (when in salt water) should have something happening on the green/white? How much voltage, how much current?

No idea what voltage or frequency you put on the coils, one or both, or what the sensor shows. I would send an email to EcoLab with model number info. That's the first time I've ever heard of such a device.

If you have an LCR meter, you could find the coil inductance and take a guess as to what 60Hz voltage would be safe to try. I didn't read the patent description, just recognized the shape and possible wiring.

You could try putting a very low AC voltage on one coil and see what you get out of the other and the sensor. If you read anything in air then dunk it in strong soap water and try again.

PS The Yun I2C lines are on pins D2 and D3 not A4 and A5 as is the Uno.

That patent drawing looks interesting. I suspect the LCR meter is going to give a value that suggests using 10kHz or higher. The center handle looks to have a nonconductive tube that extends just out of both toroids and thus forces current flow in the fluid to go through the center of the toroids. To simulate the conductive fluid use a wire through the center (e.g. coupling both toroids) and a resistive shunt to see how much of a signal from one toroid couples to the other at that resistance. Unfortunately, the conductive path in the fluid is complicated, I think it would look like the donut shape of a bar magnet's field lines, but it would be current flow in the fluid. If there is a fluid sample with known conductivity that could be used for calibration, perhaps a variable resistor (shunt) could then be used to match with it. It is unlikely that the coils have a linear coupling, so they need to be mapped with that shunt from the calibration point, but the ratio between the shunt and calibration fluid resistance should hold.

The frequency is probably that which gives the best coupling between the two coils via the fluid. It could even be in the ultrasonic region and perhaps 40KHz. A variable frequency signal generator could be used for testing or even a 555 timer circuit with an NPN driver.

By my understanding the math is Where

T is the temperature of the sample,
Tcal is the calibration temperature,
σT is the electrical conductivity at the temperature T,
σTcal is the electrical conductivity at the calibration temperature Tcal,
α is the temperature compensation slope of the solution.

For sodium chloride that means α = 2.14

Also to quote wikipedia:

Industrial conductivity probes often employ an inductive method, which has the advantage that the fluid does not wet the electrical parts of the sensor. Here, two inductively-coupled coils are used. One is the driving coil producing a magnetic field and it is supplied with accurately-known voltage. The other forms a secondary coil of a transformer. The liquid passing through a channel in the sensor forms one turn in the secondary winding of the transformer. The induced current is the output of the sensor.

This seems to be mostly what I find, but don't actually know what the units are for reading. Is this the voltage from toroid 2 in volts, the difference, I don't know.

The induced current is the output of the sensor.

Ok that helps, so the fluid resistance is proportional to the induced current. Had to make a very simple simulation to understand what was going on. I have no clue what the inductance of the cores @cj has (sim is set at 1mH) or the turns (sim set at 40). I think the one turn from the fluid could have a good coupling coefficient (.99 set in the sim), but that is just a hunch. I'm not sure how to measure the current output in a real circuit without burdening. That is to say, if I put a sense resistor in place of the current meter (which is a short circuit) it changes the current flow and there are lots of unknowns, so the simulation is not going to help with that. I wonder if a hall sensor can measure uA's. Anyway if you have the Chrome browser on a PC then the values of things in the simulation can be modified.

In a real circuit, you can measure the max power transferred to the load. I used this scheme when designing solar panel controllers/chargers.

Calculate the source impedance by dividing open circuit voltage by short circuit current. Then load the output with a resistor of that value. The power dissipated in that resistor is half of the total generated by the secondary. A resistor decade box confirmed the results.

Well I finally have an example. I have never worked with multipliers before, but this publication pretty much gives a complete example.

50kHz, 10 volts, and a multiplier.

However it gives a value of 2.2 for α when previous research I have done shows NaCl having a value of 2.14.