You've got a 40-node field network running Modbus-RTU over RS485, and the first 72 hours look clean. Then node 17 starts reporting EC values that creep upward by 200–400 µS/cm over the next two days. The VWC and temperature channels stay rock solid. You check the cable, the terminator, the ground loop—nothing obvious.
This isn't a sensor failure. It's a signal integrity issue that shows up specifically in EC measurements on multi-layer probes. Here's what's happening and how to catch it before it corrupts your dataset.
The EC channel is the canary
The OHTS1021 measures VWC using a TDR-based dielectric method with non-contact electrodes. That approach is inherently resistant to the electrochemical corrosion and polarization that plague resistive sensors—the electrodes never touch the soil solution. But the EC option is a different story. It's an electrical conductivity measurement, which means it depends on the ionic content of the soil water. And that's where drift sneaks in.
After 72 hours of continuous operation, three things typically contribute to EC drift:
Temperature coefficient mismatch — The EC reading is temperature-compensated, but the compensation curve assumes a uniform thermal profile. In a 3-layer configuration with 10cm spacing, the top layer can be 8–10°C warmer than the bottom during midday. If your logger applies a single compensation factor to all layers, the deeper readings will drift relative to the surface.
Micro-voltage offsets on the RS485 bus — The OHTS1021 draws 0.7W in the 3-layer configuration. That's low, but over a 2000-meter bus with multiple nodes, the cumulative ground potential shift can create small DC offsets at the sensor's analog front-end. These offsets are invisible in the VWC channel (which uses frequency-domain detection) but directly bias the EC measurement.
Soil pore water redistribution — This isn't a sensor error. After 72 hours of continuous measurement, the probe's presence can alter local soil moisture distribution, especially in clay soils. The EC reading changes because the soil around the probe is changing.
What to check before you suspect the sensor
Here's a field procedure that has caught more false drift alarms than anything else:
Step 1: Log the raw EC and temperature simultaneously. Don't rely on the compensated output. The OHTS1021 provides temperature at each measurement point—compare the raw EC vs. temperature curve for each layer. If the drift correlates with a temperature change of more than 2°C, your compensation curve is the issue, not the sensor.
Step 2: Check the supply voltage at the sensor. The OHTS1021 accepts 10–30V DC. If you're at the low end of that range and the bus has 15+ nodes, the voltage drop over 2000 meters can push the sensor below its regulation threshold. The VWC channel will still work, but the EC front-end may start to saturate.
Step 3: Run a 24-hour soak test with the probe in a known solution. Use a calibration standard (e.g., 1413 µS/cm) and log the reading every 10 minutes. The OHTS1021 has a response time of ≤60s, so you should see stable values within the first hour. If the reading drifts more than ±3% over 24 hours, you've got a hardware issue—refer to the datasheet for the exact accuracy spec.
Step 4: Verify the tilt sensor isn't the culprit. If your unit has the optional tri-axial tilt sensor, a 1–2° shift in probe orientation can change the soil contact pressure around the EC electrodes. This shows up as a slow, monotonic drift—exactly what you'd see after 72 hours.
A practical wiring note
The RS485 bus is rated for 2000 meters, but that assumes proper termination and a daisy-chain topology. In practice, I've seen more EC drift issues from star topologies than from any sensor defect. If you're running a multi-level bus network, keep the following in mind:
| Parameter | Recommendation |
|---|---|
| Termination | 120Ω at both ends of the bus |
| Grounding | Single-point ground at the logger |
| Cable | Twisted-pair, shielded, drain wire grounded at one end |
| Node spacing | Keep unterminated stub lengths under 1 meter |
The OHTS1021's low power draw (0.7W for 3-layer, 0.96W for 5-layer) means you can run it on a 12V battery pack for extended periods—but that doesn't excuse sloppy grounding.
The takeaway
EC drift after 72 hours is almost always a system-level issue, not a sensor defect. The OHTS1021's TDR-based VWC measurement is solid—that's the channel you can trust for long-term trend analysis. For EC, treat it as a relative indicator unless you're logging temperature at the same depth and checking the compensation curve.
If you're designing a network and want to avoid this class of problem, look at the OHTS1021 datasheet for the exact temperature characteristics and EC range (0–20000 µS/cm). And if you're already in the field, add a 24-hour soak test to your commissioning checklist—it's cheaper than a site visit.
For questions about bus topology or calibration procedures, reach out to the engineering team—they'll give you the same answer I just did, but with better formatting.
Related Products & Next Steps
- OHTS1021 Tube Soil Moisture Monitoring Sensor — full specifications and datasheet
- OHTS1020 Isolated Multi-Parameter Soil Sensor
- OHTS1022 Soil Moisture and Temperature Sensor
Contact our engineering team for application-specific deployment guidance.