Logging pump use

Added by marmotte over 7 years ago

I am planning to monitor how often and how long a water pump is triggered by its floater arm, in a house used only on holidays. The pump looks like it prevents a sizable amount of water from infiltrating from below the house.

The room where it is installed is separated from the house by an old stone wall about a meter thick, with no opening between the pump room and the rest of the house.

There is no separate fuse for the socket where the pump is plugged, so I think measuring from inside the house would be tricky.

I thought of 2 solutions :
- a Jeenode with a CT sensor ( and an interface to it (, on the pump mains cable.
- something like the AC current node ( and

Once I get the sensor working, I think locating the Jeenode near the thinest part of the wall (a niche on the other side) will help it transmit to a Jeelink in the house.

Do you think it has a chance of working ?

The main Jeenode/Jeelink setup will be logging temperature, etc… inside the house, helped by a RaspPi.

Replies (2)

RE: Logging pump use - Added by martynj over 7 years ago

The most inobtrusive solution is to detect something about the pump running. Monitor the current draw (use the current clamp method around a single conductor) so there is no need to disturb any of the mains wiring. Full isolation from dangerous voltages is important. Clamping near the pump provides a clean ON/OFF signal, but then using that to trigger a transmission through the stone wall might be too much signal loss.

The alternative is “signature analysis”. Use the clamp at the power distribution board on a circuit that has the least parallel loads. Then sample the clamp output continuously. The signature of a large motor is very distinctive - there is a peak draw of 5-10 x the running current at startup (which may saturate the CT, but no matter). Some signature detection logic that looks for that event should be enough to start the ‘PUMP ON" timing. Power the detector through a good quality USB plugin from the same circuit - this will provide a general reset function to recover correctly from brownouts. The detecting node can then send status packets for remote display or integrate into other house monitoring functions.
If you relax the "do not disturb wiring’ constraint, then there is no need for the CT. Just put the USB supply in parallel with the motor input terminals (AFTER the float switch). Then node alive = pump running. But there is still the issue of sending packets through that wall.

If you want to avoid mains wiring completely, use a different side effect. For example, the water is pumping somewhere through a discharge pipe - there are plenty of sensors to detect flow and/or pressure change (you can even recycle one from a junk washing machine). Most modern pump controllers do this anyway and lockout if the pump is on, but no discharge appears. How old is the pump/control box?

RE: Logging pump use - Added by marmotte over 7 years ago

dates in the concrete in the pump room and what I know of the history of the house tell me end of 80s, the pump may be that old, it was there before we bought the house.
the electric wiring has been entirely redone at the same period.

I’ll go for the CT, the mains cable is the most accessible part.

Now I wait for packages deliveries :)