Sump Pump False Alarms: Solution Needed

It may be possible to filter the Input signal through hardware (RC circuit). There may be a bug in the way the PLC is programmed. The PLC may need to be reset.

You may also need to look at your level sensor. Perhaps it is situated in a way as to be exposed to too much turbulence.

Yes... the suggested course of action from JRaef is where you should start.

Programming should only be used as a fix for faulty or non-performing I/O as a last resort. Granted, the need for a de-bounce timer is a given, but, a 1 second delay should be lots, unless of course the action of the water in the sump is so chaotic that the high level sensor is constantly being thrown around.

The real question is why is the I/O point changing state and putting the process into alarm. If all avenues of investigation as to why the I/O does not perform as expected have not fixed the problem, then a software fix is appropriate.

Try putting several sensors in series to reduce falls alarms.

Good luck,

James

Uh, how do you know it is a false alarm? Maybe it is a real alarm and instead of fixing the alarm you need to fix the system?

That's it!

Next time there is a 'false' alarm go into the pit and see for yourself!

You can always repair the alarm cotact if it is sticking./shorting/bridging -by-filth.

I think you are using contact type level sencing that some times gives wrong signals due to various reasons. (Like dirt on probes, probes too close, RFI/ EMI, etc). Try to use non contact type float less level switch.

<use non contact type float less level switch.>

But you do need a float here too.

And this is not a non-contact one.

Increasing the debounce time is a good suggestion, also make sure you don't have any intermittent connections. Most logic inputs will float to a high or logic 1 state if they are open. The solution to that is to put a high value resistor (say 1-10 megaohms) between the switch contact input at the PLC and ground. That will draw just enough current to pull the input low. then when the switch is energized, the voltage drop across the resistor will still be high enough to register as a logic 1.

Another more complicated way to skin the cat would be to use an analog output level sensor instead of a float switch and average or integrate the value to eliminate false positives.

Another approach is to use multiple level switches in series set at different levels so that the pump will only engage if all of the switches are activated as someone else suggested. the problem with that approach is that your reliability becomes a problem because it only takes one switch to fail open to disable the entire system.

The solution to that issue is to put them on separate inputs and "vote" them so that if any two of three indicate a high level you turn on the pump. This is the scheme that NASA used on it's Hydrogen sensors in the shuttle.

I would try the resistor first. It is cheap and requires no re-programming.

I think your multi-level float switches are excellent. Redundant sensors will lesson false alarms. Good job. "You the man". Prepare a wiring diagram for "The Man". I see your idea clearly.

Good answer.

James