Why Does Your Conductivity Reading Drift Every Afternoon?

Question-format technical deep dive · Instrument engineer perspective · Published: September 29, 2026

Key Takeaways

• A smooth afternoon drift is usually thermal physics, not failure. Most natural waters change conductivity by about 2% per °C (typical range 1-3%/°C), so an 8 °C afternoon temperature swing can shift the apparent reading by roughly 16% before any instrument fault exists [1].
• Readings are normalised to a 25 °C reference, and a mismatched compensation coefficient turns a small real effect into a double-digit reporting error [1].
• Fouling is a first-order error source, not a nuisance. Continuous-monitoring protocols routinely evaluate sensor-fouling and calibration-drift error before trusting the data [4].
• Random afternoon spikes point to variable-frequency drives, large motors and grounding problems. A ten-minute equipment-off test settles it [5].
• Supply water follows daily rhythms too. Diurnal cycles in specific conductance are documented in streams and in groundwater near streams [3].
• A one-hour diagnostic sequence — capture, classify, paired handheld check, clean, settings review — separates water chemistry from instrument behaviour before anything is replaced.

Every plant has one. The conductivity trace sits flat and honest all morning, then starts climbing around 13:00, peaks in mid-afternoon, and slides back down by evening. Every single day. Maintenance swaps the sensor, the pattern returns, and the instrument gets blamed for lying. In most cases the instrument is telling the truth. A reading that drifts on a daily rhythm is usually responding to something real: the water itself, the temperature cycle, the electrical environment, or a fouling layer that grows and shrinks with the sun. Five causes follow, each with a verification experiment, plus a one-hour diagnostic sequence you can run without pulling the probe out of the line.

First, Sort the Signature

Before diagnosing anything, classify the drift. Periodic drift comes in three signatures, and each points to a different family of causes:

• A smooth daily curve tracking sun and ambient temperature is physics, not failure. The electrical conductivity of water rises with temperature — for most naturally occurring waters the change is about 2% per degree Celsius, with a typical range of 1% to 3% per °C [1].
• A step change at a fixed clock time — a jump at 14:00 sharp, every day — is almost never physics. Look for a scheduled event: chemical dosing, a filter backwash, a regeneration, a second production line starting up.
• Random jumps and spikes clustering in the afternoon point to the electrical environment: drives, motors and welding equipment switching on.

The afternoon timing in the title is therefore the best diagnostic gift your process can give you. Random faults do not keep a schedule; causes that repeat every afternoon are physically coupled to the daily cycle. Standard field practice reinforces this. USGS measurement protocols treat specific conductance as a temperature-sensitive quantity that must be normalised to a 25 °C reference before readings from different hours can be compared at all [2].

Cause 1: The Daily Temperature Cycle

Temperature is the leading suspect, because water conductivity is strongly temperature-dependent. Instrument references commonly normalise readings to 25 °C, and the temperature-compensation gradient for most naturally occurring water samples is about 2%/°C — but it can range between 1% and 3%/°C depending on the ionic mix [1].

The arithmetic is unforgiving. A plant whose process water climbs 8 °C between morning and mid-afternoon — common in roof tanks, outdoor pipework and shallow wells in summer — will see the apparent conductivity rise roughly 16% even if the dissolved-salt concentration has not changed at all (8 °C × 2%/°C). If the transmitter is compensating with the wrong coefficient, or its temperature element senses a different stream than the conductivity cell, a small real effect becomes a double-digit reporting error [1].

Verification experiment: log temperature and conductivity at the same timestamp for three days and plot conductivity against temperature. A tight linear cloud with a slope near 2%/°C tells you the afternoon drift is thermal [1]. Then compare the compensation coefficient configured in the transmitter with the coefficient appropriate for your water. In high-purity and mixed-effluent waters the temperature relationship becomes non-linear, so a fixed linear coefficient will systematically mis-report even when every component works as designed.

Cause 2: The Water Itself Changes with the Shift

Municipal and industrial supply systems are not chemically constant. Demand peaks in the afternoon pull water from different parts of the distribution network or from different wells. Reservoirs stratify and turn over. Upstream users discharge on their own schedules. Surface waters add a biological dimension, since photosynthesis and respiration follow the sun, and published field studies of streams and of groundwater near streams document how specific conductance tracks the daily exchange between surface water and the aquifer [3].

Verification experiment: take a grab sample at 06:00 and another at 15:00 for two days and have both analysed in the laboratory. If the laboratory values move in the same direction and by a similar magnitude as the online trace, the sensor is healthy. Your supply water really does change in the afternoon, and the fix belongs to operations planning, not instrumentation.

Cause 3: Biofilm and Fouling

Biofilm is the slower cousin of drift. It usually produces a gradual baseline climb over weeks rather than a clean daily curve. But in warm, sunlit, nutrient-rich water it can amplify the daily thermal effect, because the insulating layer changes the effective cell constant of the measuring electrodes. Monitoring agencies treat fouling as a first-order error source rather than a curiosity. USGS continuous water-quality monitoring guidance sets out routine evaluation of error from sensor fouling, whether from sedimentation or biological growth, and from calibration drift, then decides whether affected data need correction [4].

Verification experiment: inspect the electrodes visually. A tan, green or slimy film is diagnosis enough. Clean the cell gently, record before-and-after readings at the same water temperature, and recalibrate. If the afternoon pattern shrinks after cleaning, fouling was amplifying the thermal cycle. A cleaning schedule matched to your fouling rate — more frequent in warm months, less in cool ones — is cheaper than any diagnostic debate.

Cause 4: Grounding Loops and Drive Noise

If the drift is jagged rather than smooth, suspect the electrical environment. Field diagnostics rank variable-frequency drives, large motors, contactors, transformers and welding equipment among the first suspects when sensor readings become erratic, together with high-current power cables routed near signal wiring and damaged or ungrounded cable shields [5]. Conductivity cells are unusually exposed here, because the measurement itself is an ionic current in the water. A water system with inadequate grounding can couple noise directly into the reading.

The afternoon timing is the tell. If your drives ramp up with the second production shift, or the chiller compressors start at noon, the plant’s electrical noise floor rises exactly when the drift appears.

Verification experiment: power down the suspect equipment for ten minutes and watch the trace. If the spikes vanish, reroute signal cables away from power cables, ground shields at a single point, and consider an isolated 4-20 mA output. Record the signature before changing wiring: a ground loop typically shows a consistent offset or a frozen reading, while drive noise shows spikes — and the two fixes differ [5].

Cause 5: Cell Aging — and the Reference-Electrode Clue

Electrodes age. Coatings, corrosion and abrasion change the cell constant slowly, which appears as a one-way drift over months rather than a daily cycle. But if your probe is a multi-parameter unit, aging elsewhere on the probe is diagnostic gold. An aging pH reference electrode will not move the conductivity channel at all, yet it often shows the same afternoon signature on the pH trace. Comparing channels lets you rule the conductivity cell in or out. And when every parameter on the probe drifts together in the afternoon, the problem is usually the transmitter, the cable or the power supply — not the water.

Verification experiment: check the cell constant against a standard solution held at 25 °C. If a cleaned, recalibrated cell still tracks temperature with a slope far outside the expected 1-3%/°C band [1], the cell surface has changed permanently, and replacement — not another calibration round — is the economical fix.

The Diagnostic Table

Drift pattern	Most likely cause	How to verify
Smooth rise and fall tracking sun and ambient temperature	Daily temperature cycle; compensation-coefficient mismatch	Plot conductivity vs. temperature; a slope near 2%/°C confirms a thermal origin [1]
Step change at a fixed clock time	Scheduled plant event (dosing, backwash, regeneration, second line start)	Correlate the trace with event and dosing logs
Random spikes clustering in the afternoon	VFD, motor, or welder noise; ground loop	Switch the suspect equipment off for ten minutes and watch the trace [5]
Slow baseline climb over weeks	Biofilm, sediment fouling, or coating on the electrodes	Visual inspection; clean the cell and compare before-and-after readings [4]
Afternoon dip only in warm, dry months	Supply-water quality swing; biological uptake in the source	Laboratory analysis of grab samples at 06:00 vs. 15:00 [3]
All parameters drift together on a multi-parameter probe	Transmitter, cable, or power supply	Swap the transmitter or test with a known-good cable

A One-Hour Diagnostic Sequence

1. Minutes 0-10: capture the signature. Export three days of data at one-minute resolution with temperature alongside conductivity. Never diagnose from a daily average — the pattern is the evidence.
2. Minutes 10-20: classify the curve. Smooth, stepped or spiky. Pick the matching row in the table above before touching anything.
3. Minutes 20-35: run the paired check. Place a verified handheld meter in the same flow as the online sensor at two different times of day, and record both water temperatures with the readings.
4. Minutes 35-45: clean and recheck. Inspect and clean the cell, then compare post-cleaning readings at matched temperature. Note whether the afternoon pattern shrinks.
5. Minutes 45-55: interrogate the settings. Compensation coefficient, compensation source (internal sensor vs. external input), and the 25 °C reference configuration [1].
6. Minutes 55-60: decide. Physics and process causes stay on the operations action list; a cell that fails the standard-solution check goes on the replacement list.

When the Sensor Is the Problem

If verification ends at the probe, replace deliberately rather than desperately. The highest-value upgrade is not a faster sensor but a synchronised one: a multi-parameter probe that measures conductivity, pH, ORP and temperature in a single body gives the transmitter one thermal reference, which is exactly what separates a real conductivity change from an apparent one. Four parameters. One probe. Zero alignment error.

It also pays to think one step beyond the 4-20 mA loop. Transmitters with Modbus RTU/TCP digital output stream synchronised multi-parameter data into SCADA or a cloud historian, so afternoon patterns become searchable, analysable datasets instead of anecdotes — sensors that feed your AI water model, not just your dashboard. An example of an online conductivity/resistivity/TDS controller designed around that kind of integration can be reviewed here [6].

References

1. Wikipedia — “Electrical conductivity meter”: temperature-compensation gradient for most natural waters is about 2%/°C (range 1-3%/°C), referenced to 25 °C. https://en.wikipedia.org/wiki/Electrical_conductivity_meter
2. USGS National Field Manual, Chapter A6.3 — Specific Conductance: field measurement protocols for specific conductance, including temperature dependence and normalization to 25 °C. https://pubs.usgs.gov/tm/09/a6.3/tm9-a6_3.pdf
3. Cox, M. H., Su, G. W., & Constantz, J., Ground Water 45(2), 187-195 (Wiley Online Library) — “Heat, Chloride, and Specific Conductance as Ground Water Tracers near Streams”: specific conductance and temperature as tracers of stream-groundwater exchange. https://onlinelibrary.wiley.com/doi/10.1111/j.1745-6584.2006.00276.x
4. USGS — Guidelines and Standard Procedures for Continuous Water-Quality Monitors (Wagner et al., Techniques and Methods 1-D3): record-computation procedures that evaluate and correct error from sensor fouling and calibration drift. https://pubs.usgs.gov/tm/2006/tm1D3/
5. UNITEC Industrial Procurement — “Troubleshooting Erratic Sensor Readings: EMI/RFI, Grounding, Cable Degradation, and Transmitter Diagnostics”: checklist naming variable-frequency drives, large motors, welding equipment, and grounding integrity as sources of erratic readings. https://www.unitecd.com/troubleshooting-erratic-sensor-readings-emi-rfi-grounding-cable-degradation-and-transmitter-diagnostics/
6. Shanghai ChiMay — online conductivity/resistivity/TDS controller product page (integration-oriented conductivity measurement and control). https://chimaytech.net/?p=931

About the author: Written by the Shanghai ChiMay Technical Editorial Team — specialists in online water quality analysis instrumentation for industrial and municipal water systems.