False trigger of Water Level Controller with LM324 from talkingelectronics.com

[bigslo]

I am making a Water Level Controller project from talkingelectronics.com website:
http://www.talkingelectronics.com/projects/100%20IC%20Circuits/1-100_IC-Ccts.html#WaterLevel
PDF with more detail

I have tried several circuits I found on the Internet, but they are usually unreliable or doesn't suit my needs. I then found this circuit on Colin Mitchell's website which is exactly what I needed. I liked the design because:

• The signal for sensing water is AC, which prevents/reduces electrolysis on the probes
• The relay first turns on, pumping water and when the Medium sensor is triggered and then High level is triggered again, the relay turns off.
• If water falls below High level, the relay is not turned on again. Only when water falls below the Medium sensor, pump will be turned on again. This prevent the pump from turning on and off when water is used.

I tested this first on a circuit simulator from http://www.falstad.com/circuit/. Since I cannot get it working with a value of 47K for the R14 (connected to Pin 10 of LM324), I substituted it with a 680R (same as the R5 value connected to Pin 5 of LM324) which works as expected. I also removed the switch, directly connecting the R21 (10K) to the base of the first transistor S8050 and directly connecting the base of the second transistor to the collector of the first transistor. Please see attached image for full schematic of my project.

Schematic Diagram

I then build it on a breadboard which worked fine. Then I designed a board using EagleCAD, etched a circuit board and everything worked fine when tested on with a small amount of water in a bowl.

This morning I began installing it with the sensors on the overhead water tank on my house and the controller near the pump. I used an old Cat5 cable to connect the sensors with the controller which is about 10 meters apart. The individual wire is single cored about 24AWG. I checked the wires first with my multimeter and they were all good. I tied the ends of the sensor wires on a long piece of small plastic pipe so the the wires wouldn't float. When testing the controller before submerging in the water, it still worked fine. Please note while installing, the overhead tank is already half full, meaning that the Medium level sensor is already submerged.

Now the problem is that, when I turn on the controller with the sensors in the tank, the high level would be triggered even though it is way above the water level. Sometimes the high level would be triggered momentarily and sometimes it stays on. And because of the design, if medium level and high level are triggered, the relay (pump) cannot be turned on.

I then tried turning off and on again several times. Maybe once out of 6/7 tries, the high level is not triggered and the relay (pump) turns on, but the rest of the attempts are unsuccessful. After the tank is full, the relay is turned off as expected.

I really don't understand why the high level sensor is falsely triggered. I don't have any testing equipment or tools (other than a multimeter), and since I am a beginner in electronics I don't really know how to troubleshoot even if I have. So any help would be greatly appreciated.

[fourtytwo42]

I think the problem is probably the 10metres of cable between the circuit and sensor. One fix you could try is using a seperate pair for each signal and connecting the other wire of each pair to ground. This may reduce the crosstalk enough, if not see if you can reduce the cable length. If all else fails the circuit will have to be altered to cope with the cable crosstalk and that may not be easy.

[bigslo]

Thank you for the quick reply. But I have two question

One fix you could try is using a seperate pair for each signal

Does this mean I will put another 100n from the output of the first op-amp and use another wire and put the sensor probe near the High Level. (The setup I have right now is, the Signal probe is at the bottom of the tank. The Medium level probe is about 2 inches above that. And the High level probe is at the top of the tank, about 4 feet above the signal probe.)

and connecting the other wire of each pair to ground

I am not exactly sure how to do this (total newbie here  ). Can you please explain it a little more, perhaps with an image, if it is not too much to ask.

[phil from seattle]

I agree that wire length looks like a problem.  Definitely try the grounded pair approach fourtytwo42 suggested. If that fails, I would split the circuit up - power, relay and driver at the pump, analog circuitry near the sensors.  Run power to the sensor circuits in the same cat 5 with the relay signal coming back.  I know this complicates your LED indicator scheme.

Another approach to the sensors is to use floats that trigger microswitches. The signal could be higher current and thus more immune to noise, line drop or cross talk so you could keep your electronics near the pump.

Personally, I'd look at using a capacitive sensor - two flat, insulated, parallel metal plates. As water rises or falls, the capacitance changes and you can use the RC time constant to determine the water level. 

[edit] In the grounded pair approach, you take the 4 pairs of wires, connect one of each pair to ground and the other 4 wires are for your signals.[/edit]

[bigslo]

connecting the other wire of each pair to ground

I dont really understand this but I connected all the unused wires in the cable to ground but now the medium level is not triggered even though it is submerged. I will try again with new sets of wires to see if it is a problem with the cable