SX10504: About water level detection
The SX10504 has five channels of water level switching. These work on the principle that if two electrodes have a voltage applied between them, and they are submersed in an electrically conductive fluid, a detectable current will flow. The amount of current that flows is proportional to the surface area of the electrodes (larger electrodes = more current) and inversely proportional to the distance between the electrodes (more spacing = less current). The current also depends on the conductivity of the fluid (very pure water = small current, very salty water = much more current). For a given electrode geometry and fluid the overall conduction can be defined as a resistance value.
The conductivity circuit on the SX10504 is designed to switch on and off at a certain resistance. The board has an adjustment for setting this resistance sensitivity. The adjustment range is approximately 2.2K to 27K. The circuit applies an AC voltage to the electrodes (DC can cause electrolysis problems).
Each of the six conductivity channels has a 2-wire connection, for 2 electrodes. On each channel one electrode is connected on the board to 0V or ground, while the other electrode is the "active" one. If all channels are used in a single tank to indicate up to 5 levels, then only one ground electrode is required. If you are monitoring several isolated (not interconnected) tanks, then each will require its own ground electrode.
It is important to realise that the grounds are all connected together and therefore at the same potential. The SX10504 senses the resistance between each active electrode and ground. If you were to try and have multiple ground electrodes all within the one fluid mass (say several interconnected tanks), the board could get confused. It can also get very confused if you have any other device in the fluid that generates AC or DC voltages or currents.
You have to design and provide your own electrodes. In the main this requires mainly common sense plus a bit of experimentation.
There are two main classes of electrode design, which are governed by the nature of the vessel or pipe work in which you want to detect the presence or absence of fluid. If the vessel is made of an electrically insulating material you will need two electrodes. If the vessel is made of metal, you will very possibly be able to use the vessel itself as one electrode. In the latter case you have to make sure the vessel is connected to the 0V (ground) electrode connection because the vessel will usually already be at ground potential.
It is usually advisable to use stainless steel for the electrodes. In many cases something as simple as two stainless steel machine screws through the side of a plastic vessel will do, with wires connected on the outside. In a metal vessel, a single screw with an insulated mounting arrangement, plus a wire from the screw and one from the body of the vessel (don't rely on ground conduction for the return path).
There are two parameters that affect the performance of your electrodes: How much they conduct when they are in fluid (wet) and how little they conduct when they are not in the fluid (dry).
How could dry electrodes conduct? Because they may not be dry! Wetting, residue build-up, even solid matter or foam, can all form unexpected and unwanted conduction paths when the electrodes are supposed to be dry. You need to apply some common sense to your own situation to come up with a suitable design that will avoid such effects.
You also need to apply some of that common sense to the wet situation.
Is there any residue in the fluid that might give the electrodes an insulating coating? (Greasy or oily water from a restaurant or garage could have this problem).
Can the conductivity vary a lot, so the tests you performed with tap water are invalidated when the customer uses distilled water?
When you are testing electrode geometries, it is a good idea to use a multimeter to measure the resistance between the electrodes. Providing you get a reasonably steady reading for a minute or two, this is quite valid (in some fluids electro-chemical effects may make a DC resistance reading meaningless).