hi Jules,
The simulation shows that it does work, if the Q1 Base voltage swings above and below the trigger threshold.!
E
Added:
Note the Sensor is NOT connected for the POT test.

 

Even if you add a load simulating the pump, the simulation shows there is nearly 3V on the pump when it is supposedly 'off'. That may be enough to keep the pump running.

 

OK, something in the circuit is not as in the schematic.
Disconnect the sensor and the motor (no point in it running for this).
Measure the voltage at the wiper of the pot or the base of TR1, whichever is easier, and adjust the pot for the lowest voltage (should be close to zero).
Now measure these voltages:
TR1 collector and emitter.
TR2 base, collector and emitter.

Ok, may have to be tomorrow.

 

Even if you add a load simulating the pump, the simulation shows there is nearly 3V on the pump when it is supposedly 'off'. That may be enough to keep the pump running.

There is a relay driving the pump, so it is the relay drop-out voltage which is the determining factor.

 

There is a relay driving the pump, so it is the relay drop-out voltage which is the determining factor.

The current with the pump full on is 2.7A so it might be possible to run the system without the relay? Most FETs would be fine with that load.

 

Yours is a completely different circuit. The original circuit is a Schmitt trigger comparator. Your circuit has no hysteresis, and will not work well in this application.

It has hysteresis on the back end and since the TS has already made the pcb it would be an easy mod. Does it really need hysteresis on the front end? I guess that would depend on how long it takes for the fluid level to drop below the terminal.
Steve G

 

I guess that would depend on how long it takes for the fluid level to drop below the terminal.

And how much the liquid sloshes around especially while being pumped perhaps.

 

There is a relay driving the pump

Ah, missed that.

 

Even if you add a load simulating the pump, the simulation shows there is nearly 3V on the pump when it is supposedly 'off'. That may be enough to keep the pump running.

That is the problem with the circuit. I have the exact relay used and the dropout voltage is 1.5volts.
Steve G

 

That is the problem with the circuit. I have the exact relay used and the dropout voltage is 1.5volts.

If that is the problem it will become clear when TS has measureed the voltages.

 

I have breadboarded JulesP circuit and it does have a constant 3.4 volt offset at the emitter of Q3. With VR1 set at midrange the delayed start time is about 4 seconds. The problem is the relay holding current is too low, but if you change R4 to 470 ohms the offset drops to 1.1 volts. Increase R2 to 22K and you can get the 4 second delay back if needed.
Steve G
Edit: The delay is dependent on the resistance of the sensor. The 4 second delay was using a sensor resistance of zero ohms.

 

Hi all,

I have built a water level feed control for an electrolyser using the attached circuit but it doesn't quite work as I would like it to.

The sensor consists of two Allen screws in the side of the electrolyser body (test rig) acting as the sensor and when the water level falls below the top screw the motor starts up as expected. However, when the water level rises back to the top screw the motor doesn't switch off.

There is a 10k pot in the circuit that doesn't seem to make any difference to how the device works and what I'm asking is how can I make the motor stop immediately once the water (0.05M KOH) rises to the top screw.

I'm thinking that perhaps the 330uF cap is holding the base of the first transistor on for too long? So maybe a smaller value like 50uF or a 1k bleed resistor across it?

Thoughts appreciated.

Jules

This circuit is an example of how to do something that is relatively simple, with fewer parts, into something difficult with unnecessary complexity. Here is a much simpler version:

Sensor closed, motor off:



Sensor Open, motor on:

 

That looks ok for a low resistance type of sensor, but will it work where the sensor resistance is either high (water between two electrodes) or very high (no water)?

 

The circuit does not work because it has terrible sensitivity.
It depends on electrode size, but their wet resistance is nowhere low enough for a couple milliamperes flow to pull down point "A" and shut off Tr1. The circuit seems designed for huge electrodes.
I would increase the value of R2 from 1k to say 100k-1MEG, and ditch the 10k pot. Just put in a 1MEG-10MEG resisitor there, from point A to GND. The time-delay capacitor C2 not ideal there but should still give some anti-slosh coverage.

Ideal water-level sense circuits use AC excitation to the electrode so electrolysis and corrosion do not happen, and they are more sensitive than DC circuits. It would be a better way to go but I see you already have a PC board that could work. Try the two changes and it should work.

 

The circuit does not work because it has terrible sensitivity.

It's not the sensitivity causing the problem. It's the 3 volt offset and the relay (see post #51). Changing R2 and removing the pot will not change the offset.
Steve G

 

OP's original circuit, from the free-energy crowd making hydrogen with an electrolyte like NaOH. I say it can't work with water alone. In all likelyhood the electrode acts like a half-cell and has a large (negative?) voltage on it, so Tr1 switches harder. So we are all on a goose chase. It depends on the metals and chemical added to the water what the wet electrode voltage really is.

 

From post #1 it's 0.05M KOH.

 

It's another silly circuit that works only due to side effects. This is not operating in water, the level sensor is operating within essentially a battery- but should be null potential at rest due to all the same metal SS electrodes.
The cell has ~2V applied to it to generate hydrogen gas, and when the salt electrolyte touches the sensor electrode, there is an applied voltage at Point A. It depends on the power supply's common-mode voltage wrt the sensor board's power.
I say this doesn't work with water alone, our simulations are wrong as well because of the other electrodes and current being injected from another power source driving Point A to another voltage.

 

It's another silly circuit that works only due to side effects.

The circuit works it's just the relay used is to sensitive
Steve G

 

I see the relay coil on/off voltages are not great. Songle SRD-12VDC relay will stay pulled in down to 1.2V and I played around with the circuit and it's not flexible at all.
To fix, I changed the relay transistor to PNP and moved the hysteresis up one stage and the circuit works much better, although still yucky IMHO because the sensing is not ideal, I think not going to work with OP's small carriage bolt vs a rod or plate electrode.

edit: I just realized the electrodes are in the presence of hydrogen gas so you would need an IS barrier. Something to keep voltage/current/capacitance below the ignition limits of IEC 60079. I've played with hydrogen and the smallest spark makes it go boom. It's not difficult to limit the energy available, for explosion safety.

I would ditch OP's circuit completely and go to an AC excitation/sensor system, these are far better.