Electronics > Projects, Designs, and Technical Stuff
Soil moisture sensors and ground loop issues
aberco:
Dear all,
I am facing ground loop issues as I am designing a soil moisture logging device, and are looking for some advices to circumvent the problem.
Setup:
- I have designed a logging station that is powered from a main switching power supply (56VDC).
- And 16 sensor units are daisy-chained to the logging station using ethernet cables with 56VCD power and RS485 signaling.
- Each sensor unit contains a NE555 front-end to interface to a soil humidity sensor (resistive type).
Everything works very well, but I forgot to consider ground loop current while doing the design: once the sensors are installed into the ground, readings are getting all over the place. I ran the setup from a battery UPS and it works well as long as only one sensor is in the ground.
Since I have already designed the logger hardware, made custom PCBs etc., I am looking for a solution that can be adapted without too much troubles to the existing design. It seems that the only way to avoid problem is to isolate the power supply and signal of elements connected to the ground. I came up with the possibility to isolate the NE555 frontend + sensor using a isolated DC/DC converter for power and an optocoupler for the signal output (see attached image). Would this be a sufficient solution?
I would like to run the sensors in parallel as well, they should not interact as they are separated by at least 5m.
Ultimately I could power them up in sequence one after another to take readings, but I would rather avoid doing this if possible as it would involve some major changes in the code.
Any help greatly appreciated!
floobydust:
RS-485 is a 3-wire system (Rx, Tx, Gnd.)
If you are only running two wires between nodes, then there is a problem.
Seeking ground, there will be a lot of common-mode noise at each node due to RS-485 imbalance. Do you have a termination resistor with bias network at the master station, which is earth-grounded? Somewhere the RS-485 needs an earth-ground reference and I don't see it in your block diagram.
The possible fix using an opto-coupler should work if you are using a second DC-DC converter, to isolate the 555 and MCU/RS-485 circuits' +5V rails. I don't know the 555 frequency or sensor capacitance/resistance, as the DC-DC converter can add its own noise and give you more trouble.
aberco:
The RS485 network does work really well, it has its own ground. It is also properly terminated (120 ohms and 680 ohms to ground and V+ for A and B).
I have not included much details on this part of the circuit because it all work. The base station is earth grounded and powered through mains with a switching power supply.
I have read other feedback on project involving soil moisture sensors (and not related to a RS485 line) that give crazy readings when plugged into the ground (and powered from mains voltage).
For practical reasons, I would like to isolate the front end exclusively (orange frame on my schematics), by powering only the NE555 from an isolated DC/DC converter and feeding the signal to the MCU trough an optocoupler.
555 is between 50Hz and maybe ~10kHz depending of soil moisture.
floobydust:
The problem seems to be common-mode noise upsetting the 555 oscillator.
In reality the 2-wire soil sensor has a third RC influence to soil earth ground.
If the master station is earth-grounded, then it's not making sense why adding node 2 affects node 1. Did you confirm where the noise is getting in ? Is it low freq. AC mains hum or HF from the DC-DC converter, or a noisy 56V PSU?
A building's powerline ground is always noisy compared to nearby soil. Connection to the utility grid, a corroded ground rod etc. Underground pipelines with cathodic protection having pulsed DC are another source of noise in dirt.
You can test it by measuring ACV between the RS-485 ground and a local ground rod stuck in the soil at a sensor, assuming it's far away.
A new DC-DC converter will add its own noise, the little 1W modules run at about 300kHz and have lots of internal stray capacitance. So the 5V side would have 300kHz noise there. I add a small CM choke and Y-cap to get rid of this. There is a good chance galvanic isolation with a DC-DC does not work as expected.
Your firmware could filter out mains hum and other periodic noise, assuming the 555 is still able to pick out soil sensor. I would add a small speaker and listen to the 555 timer's output, to hear what is modulating things, as a poor man's test.
ogden:
Today there's no point of wired soil moisture sensors. 20 years ago it would be good solution indeed.
Navigation
[0] Message Index
[#] Next page
Go to full version