Go Down

Topic: Flow Meter Hack (Read 793 times) previous topic - next topic

ron_sutherland

Nov 09, 2017, 10:18 pm Last Edit: Nov 12, 2017, 09:34 pm by ron_sutherland
An Orbit 52212 has a Holtek HT67F30 MCU that reads a Magnetic Reed Switch.



I removed the electronics and replaced the reed switch with an Allegro A1190LUBTN Hall sensor which has two states, one where it is a 4.5mA current source and another where it is a 14mA current source. It is used to control the current in a CAT5 pair. The pair terminates on a 100 Ohm resistor. At 4.5mA the voltage on the 100 Ohm resistor is not able to bias an NPN transistor that pulls down the ICP1 pin of an ATmega328p, but when it switches to 14mA the NPN is biased and pulls down the ICP1 pin. So I can use the capture times to accurately measure flow rate and count the pulses from the flow meter.





Such a setup can allow the hall sensor to be placed very far away from the AVR.

More notes at:
https://github.com/epccs/Document/tree/master/FlowMeter

Watcher

I wanted to measure the house water consumption using the  water flow in the water softener system.

The water softener had a built in water flow sensor were i was able to tap on it's outputs.

I then used a schmit trigger stage to shape the pulses before driving an interrupt pin of  the AVR.
I also used cat5e to place the AVR far away from the sensor.


ron_sutherland

Seeing the water flow has been helpful, I was way over watering one zone (I had a brain fart when setting the runtime). Looking at the data for each zone most nights has shown that irregular readings are valid, e.g. a stuck valve, a drip emitter blew open (clueless how that happened), and the pressure regulator was not doing its job (which I refused to believe but a new one confirms the old was junk). I want to add a temperature and pressure sensor at some point but am spinning the [control] board (again), so those sensors will wait.

[control]: https://github.com/epccs/RPUno/tree/master/Hardware

I have some Schmitt trigger IC's but these reed switches are just to glitchy, I think a pulse extender circuit would be needed to shape the pulse into something usable. The hall sensor signal looks perfect so I don't think a Schmitt trigger is needed.

Go Up