Isolating a signal input

Community Projects
1

I’m trying to monitor the signal to a anti-rust chemical injection system to a lawn irrigation system. Seems the controller to this system has flaked out and it simply runs all the time. So we have to unplug it when its not needed. I have a micro controller wired to a Powerswitch Tail 2 and it seems to work with a button I currently have for the input signal.

The problem is sensing when the water is flowing. The broken system has a flow sensor that is from what I could glean a 3-wire hall effect sensor that plugs into a special plumbing union that must spin some magnets around when there is flow through the union. From a volt meter, I see one line is about 14V DC, there is a ground line, and what I assume is the signal. A line that when water is flowing is about 7 Vdc and about 1.4 Vac. So I assume the AC is the signal caused by the moving magnets and the signal is DC biased to about half the input voltage.

I am going to use a Resister based voltage dividor to bring the signal down to a MCU friendly voltage range. I’m going to software handle the signal to detect when there is meaningful AC.

My problem is I don’t now how safe it is to connect the Microcontroller Ground to the injector’s controller’s ground. Could be bad for both, idk. Also the Injector controller and pump are AC powered so there’s probably noise, and the potential for spikes. So I was hoping to find something to isolate them from each other.

At first an opto-coupler sounded like the correct thing I wanted. But the more I read the more they sounded like switches of some kind. I need communicate enough of the signal to monitor the AC portion to detect flow.

Is an opto-coupler what I want or is there something else?

Thx.

2

Most flow meters provide DC pulses. If you look at the signal with a DVM it may very well appear to be AC. I’d say look at it with a scope to verify that. Once you know just what the signal is then you’ll know how to proceed.

3

lyndon:
Most flow meters provide DC pulses. If you look at the signal with a DVM it may very well appear to be AC. I’d say look at it with a scope to verify that. Once you know just what the signal is then you’ll know how to proceed.

As long as I can detect some AC behavior, I will be happy. I figured I'd sample for about 100 ms. As long as the period is within that window, I will be able to detect activity. Then I have some de-bounce like logic in place to make sure I see signal for a solid 5s, then the injector will be switched on.

I’m more concerned about protecting the microcontroller from shocks and/or a different ground. I guess one solution for the later would be to leach power for the micro from the sensor power. or confirm the sensor ground is house ground, so when I plug in a supply for the micro, it will be common. Then spikes and noise would be my only concern.