First determine whether the measurement disappeared or only the transmission did. If the local reader sees a stable frequency, look for the problem in the logger, time, buffer, or telemetry. If the local reading also fails, check the correct excitation band, a known good reader, resistance according to the manual, the cable, connectors, shield, and sources of interference. Do not open a factory-sealed sensor in the field. Document every test before you move any wires or change the configuration.
In brief
- "No data" has at least three classes of causes: no wire measurement, no record in the logger, and no delivery to the platform.
- A stable but wrong frequency can be more dangerous than an empty field. The reader may lock onto interference in the wrong band.
- Amplitude, signal-to-noise ratio, dominant noise frequency, and signal decay tell you more than the Hz result alone.
- Resistance is compared with the manual for the specific sensor and the length of both cable conductors. There is no single universal value.
- The platform may show silence, a jump, and raw data exchange. It cannot remotely decide whether the cause is moisture in a joint, a broken conductor, or a damaged wire.
The most expensive service starts with the wrong diagnosis
A NO_DATA alarm arrives on Sunday at 03:10. The team sends a technician 180 km to the site. On site it turns out that the sensor and logger had been operating the whole time, while the modem lost its network configuration. Another time the operator restarts the gateway, although the real cause is a cable cut during drilling. The third case is more deceptive: the data arrive regularly, but the reader tracks a motor harmonic instead of the wire resonance.
The travel cost is visible on the invoice. The cost of a wrong but believable reading appears later, in a bad decision, a false alarm, or an unnoticed structural change. A good troubleshooting procedure minimizes both costs.
It starts from the symptom, not from the assumed cause.
Five symptoms, five different trees
| Symptom | What may still work | Main directions to check | First proof |
|---|---|---|---|
| no online record | sensor and local storage | transmission, buffer, time, mapping | last frame and local memory |
| logger reports no reading | telemetry | excitation, band, cable, sensor | manual reading and resistance |
| frequency jumps | part of the chain | noise, shielding, band, loose connection | spectrum, amplitude, SNR |
| frequency is stable but foreign | transmission and reader | wrong spectrum peak, configuration, mechanical range | comparison with expected band |
| temperature is extreme or blank | wire measurement | thermistor circuit, cable, moisture | thermistor resistance |
The general guide to the wrong-reading cause tree covers many technologies. Here we focus only on the vibrating wire chain: the excitation pulse or sweep, coil, wire, decaying signal over time, conductors, shield, thermistor, and frequency selection algorithm.
The first 15 minutes, without going to site
1. Check the scope of the failure
Did one point disappear, all points on one logger, or the entire site? One point points to the input, cable, and sensor. A group on one multiplexer points to common power, switching, or a connection. The whole site more often means a gateway, network, clock, or service.
This is not proof, only a reduction of the hypothesis set. A shared surge event can damage many inputs, while a mapping error can hide a single point even when transmission is correct.
2. Separate measurement time from reception time
Check the time of the last observation, not only the moment it appeared in the app. The logger may have sent delayed records. A wrong clock can place a record in the past or future, making it look like current data are missing.
Compare the time on the logger, gateway, and server. Do not correct history by changing labels without recording the reason.
3. Open the last correct and first failed exchange
Look for differences in the identifier, number of values, status, time format, and server response. If the device stopped sending the entire message, the problem lies before the platform. If it sends data and receives a validation error, inspect the payload or configuration.
4. Check the logger’s auxiliary data
Battery voltage, link quality, last contact time, enclosure temperature, and memory usage can tell you whether the node is alive. Missing wire values with present system data point to a different class of failure than total silence.
5. Do not restart before securing evidence
Save logs, times, configuration, and a status snapshot. A restart may help, but it removes part of the context. If the site is critical, act according to procedure and after agreement with the responsible person. Troubleshooting does not authorize disconnecting a working alarm system.
When the local reading also fails
Field work should be prepared. The technician needs a wiring diagram, the manual for the specific model, the expected frequency range, the previous reading, cable length, a meter, and a reader tested on another sensor. For an automatic installation, before connecting a handheld analyzer, safely disconnect the correct chain according to the device instructions. The Campbell Scientific analyzer manual warns against connecting to an active system without disconnecting power.
Step 1. Known good reader and known good sensor
Test the reader on a control sensor. Then connect the suspect sensor to a second, working reader. This simple cross-test separates tool failure from the field chain.
Do not start by opening the joint box. A discharged reader battery and the wrong device mode happen more often than simultaneous damage to several wires.
Step 2. Correct excitation band
The reader must excite a range that covers the expected sensor frequency. A band that is too narrow or a wrong sweep may fail to excite the wire or may lock onto a foreign peak. The GEOKON 4420 troubleshooting manual says to check the instrument type setting and the swept-frequency excitation parameters. The welded strain gauge manual from RST identifies the wrong excitation band as a cause of no reading and unstable reading.
The documentation conclusion is simple: the expected band must be recorded for the point. "Auto" does not replace knowledge of the sensor. Before service, also make sure whether the documentation uses Hz, period, or digits; the relationships and control test are described in the guide to vibrating wire sensor calibration.
Step 3. Coil resistance and cable continuity
Disconnect the chain according to the procedure and use an ohmmeter in the way described by the manufacturer. Very high or infinite resistance may indicate an open circuit. Very low resistance may mean a short or leakage. An intermediate result must be compared with the nominal coil resistance plus the resistance of both cable conductors.
The RST VW2100 manual gives this model a nominal coil resistance of about 180 Ω with tolerance and requires adding both cable runs. The GEOKON manual for model 4420 also gives values specific to its cables and sensors. Do not treat these numbers as universal. Check the manual for your model, cable length, cross-section, and temperature.
Step 4. Thermistor as an independent clue
If the wire reading disappeared and the thermistor also shows an open circuit, suspicion of the cable or a shared connector increases. If the wire works, but the temperature is extremely high or low, the problem may concern only the thermistor pair.
In many sensors, the thermistor has several kilohms at room temperature, but the exact relation depends on the type. High resistance may indicate an open circuit, and low resistance may indicate a short or moisture. Also establish resistance to shield and between pairs according to the manufacturer’s table.
Step 5. Connectors, joint box, moisture, and mechanical damage
Look for corrosion, green deposits, water, crushing, sharp bends, signs of pulling, and nearby construction work on the cable route. Moisture can create a partial leakage rather than a full short. Then the reading becomes unstable, amplitude drops, and the thermistor shows a seemingly plausible but variable value.
Repair requires a method and a kit intended for the specific cable. Twisting conductors together and wrapping them with insulation tape in a wet pit is not a measurement splice. Before repair, record the resistances, the location of the damage, and the reading. After repair, repeat the same test set.
When frequency jumps: look at the signal, not only the result
A classic reader may return one number. A spectrum analyzer shows why it chose it. The Campbell Scientific VWAnalyzer publishes five useful parameters:
- sensor frequency,
- signal amplitude in mV RMS,
- the ratio of sensor amplitude to the largest noise component,
- frequency of the largest noise component,
- decay ratio, meaning the signal decay rate.
There is no single threshold valid for every sensor. The analyzer manufacturer gives default thresholds that warn about weak signal and extreme noise, but leaves the assessment to the user. What matters is comparison with the history of the point, the expected band, and field conditions.
Amplitude drops, frequency remains stable
This can be an early sign of rising resistance, a deteriorating connection, a longer chain, or incorrect excitation. It does not automatically mean wire failure. Record the amplitude trend and compare it with another point on the same reader.
SNR drops and the frequency jumps between two peaks
Suspicion falls on an interference source or incorrect band. Check whether the alternative peak frequency matches a motor, generator, welder, power line, or radio. Take a measurement with the source switched off, if that is safe and possible. The change confirms environmental influence, although you still need a lasting solution.
The signal has correct amplitude, but the wrong range
A configuration error, selection of another sensor in the multiplexer, or a mechanical excursion beyond range is possible. Manufacturer instructions describe wire instability when it is too loose near the travel limit. Compare the result with the initial reading and the limits of the specific model. Do not try to mechanically "retune" an inaccessible sensor in the field.
Two good readers show the same unusual trend
The likelihood increases that the change is real or lies in the sensor and installation, not in the readout electronics. Compare temperature, other points, events, and an independent method. Technical troubleshooting should not automatically discard an observation just because it is unexpected.
Shielding and grounding without random trials
RST and GEOKON instructions point to shielding, grounding, and keeping the cable away from motors, generators, antennas, welders, and power conductors as typical actions for instability. This does not mean that any extra grounding helps. A ground loop can make the situation worse.
Record the existing shielding arrangement. Change one thing at a time and compare the spectrum before and after. If the manufacturer recommends isolating the reader from earth in a test, do it according to the device instructions. A lasting solution should be designed by someone who knows the site’s electrical installation and surge protection.
Do not route a new cable path based on a single measurement. First check whether the interference appears cyclically with the operation of a specific device. A series of short measurements in different states gives better evidence than "it seemed quieter".
Example: stable reading at fan frequency
Illustrative example. A load sensor worked for three years in the 2100 to 2350 Hz range. After a new ventilation system was started, the logger began to report 3000 Hz with very little scatter. The converted result immediately went into alarm. The technician sees a stable number and assumes a sudden load change.
A portable analyzer shows two peaks. The correct sensor signal is at 2268 Hz, but it is weaker. The largest component in the excitation band is at 3000 Hz and appears only while the inverter is running. The old algorithm chooses the largest amplitude.
The team does not raise the alarm threshold. First it secures the data, confirms the behavior with the ventilation switched off, checks the shield and cable route, then narrows the band according to the sensor manual and implements filtering in the reader. After the change, frequency, amplitude, and SNR are compared over a full installation duty cycle.
The alarm is closed only after the engineer assesses that the load did not increase and after the technical cause is documented. The stability of the number was not proof of its correctness.
A separate tree for telemetry
If the manual reading and local memory are correct, do not touch the sensor. Check in order:
- whether the logger creates a record at the expected time,
- whether the adapter fetches it and writes it to the buffer,
- whether the identifier matches the map,
- whether the time format and the number of values pass validation,
- whether the server confirms receipt,
- whether retries are blocking the queue,
- whether a wrong clock is placing the data outside the current view.
Total silence, a validation error, and a buffer backlog look similar on the dashboard, but they require different actions. A NO_DATA alarm should start a procedure that first classifies the failure and then sends the right service. This is described in more detail in the guide to setting a no data alarm.
What this looks like in Inclify
Inclify shows the history of values, temperature, and data freshness status. A separate NO_DATA alarm with its own window can be configured for each channel. Notifications can arrive by email, SMS, and in the app, and the user confirms takeover of the event or mutes it until a defined time.
For recognized HTTP/JSON exchanges, the platform stores the exact request and response in a compressed communication log. The list and details are available only to administrators. Retention is configurable and defaults to 168 hours. The log lets you determine whether the device sent a record and how the server responded, but it does not show the analog signal in the cable.
Charts help you see a jump and compare the result with temperature. The quality report calculates freshness, cadence, completeness, gaps, order, and payload consistency. Inclify does not automatically diagnose the cause of damage. It will not distinguish moisture in a joint, a wrong sweep, a loose terminal, and a broken wire on its own. Those hypotheses require field measurement and the sensor manual.
After repair, delayed data can be resent through the same channel with the original UTC time. Missing points should not be drawn in manually, and a gap should not be filled by interpolation if the system is meant to document real observations.
Service intervention card
Every visit should leave a repeatable record. The minimum is:
- [ ] time of detection and last correct record,
- [ ] failure scope: point, logger, site,
- [ ] battery, clock, memory, and transmission status,
- [ ] sensor model, reader, and configuration version,
- [ ] expected band and previous frequency,
- [ ] reading from the logger and from the control reader,
- [ ] frequency, amplitude, SNR, noise, and decay, if available,
- [ ] pair resistances and temperature during the test,
- [ ] cable length and type added,
- [ ] condition of connectors, shield, joint box, and route,
- [ ] active sources of interference,
- [ ] one change made at each step,
- [ ] result before and after repair,
- [ ] person approving the point’s return to decision use,
- [ ] limitations, if the sensor remained suspect.
A photo of the meter without a description of the wires and conditions has little value. The card should let the next person repeat the test. If the incident may have contractual significance, secure the frames, configuration, and protocol according to the principles for treating monitoring data as evidence. If the point is taken out of service, record the fallback method and the impact on the safety procedure.
When not to repair in the field
Closed wire transducers are usually not intended to be opened on site. Manufacturer instructions limit service to connections, cable, terminals, and tests. A damaged or out-of-range sensor should be discussed with the manufacturer or replaced according to the design.
Do not measure resistance on an active circuit. Do not disconnect multiple points without labeling them. Do not change the band and production coefficients without a backup of the configuration. Do not short the shield to an arbitrary conductor just because it temporarily stabilizes the reading. The troubleshooting procedure must respect the manual, electrical safety, explosion protection requirements, and monitoring continuity.
FAQ
Does infinite resistance mean a broken wire?
Not always. It may mean a broken conductor, a loose terminal, the wrong conductor pair, or a damaged coil. First check the connections and accessible cable segments, compare the thermistor, and use a known good reader. Interpret the value according to the manual for the specific model and with cable length in mind.
Does low resistance mean water in the cable?
It may indicate a short circuit, insulation damage, moisture in a connector, or a measurement error, but the number alone does not localize the cause. Check resistance between pairs and to shield, inspect the connectors, and divide the route into accessible segments. A factory-sealed sensor should not be opened in the field.
Why does the reading jump only when construction equipment is running?
A motor, generator, welder, inverter, antenna, or power conductors can introduce noise into the measurement band. Compare the spectrum, amplitude, and SNR while the device is operating and while it is stopped. Then check shielding, grounding, cable route, and sweep range. Change one element at a time, otherwise you will not know the cause.
Is a stable frequency always reliable?
No. The reader can stably track a foreign noise peak or harmonic, especially with the wrong band and a weak sensor signal. Compare the result with the historical range, spectrum, noise frequency, and a second reader. A sudden stable change also needs assessment of whether it is a real structural event.
What should I do when the reading exists locally, but not online?
Secure the local record and do not touch the sensor. Check the logger time, memory, adapter buffer, identifier mapping, server response, and link. After repair, send the backlog with the original time. In the platform, it is worth configuring NO_DATA so silence has an owner and a reaction time.
When does the sensor return to use after repair?
After repeating the control reading, comparing it with the historical range, checking temperature and signal quality, and getting approval from the person responsible for interpretation. One correct sample may not be enough. For a critical point, observe the full duty cycle and compare it with an independent measurement or neighboring sensors.
Sources and further reading
- Campbell Scientific, VWAnalyzer, frequency, amplitude, SNR, noise frequency, decay ratio, and diagnostic interpretation.
- Campbell Scientific, VWAnalyzer Manual, spectrum, measurements in noise, warning thresholds, and safe connection rules.
- RST Instruments, VW2100 Vibrating Wire Piezometer Instruction Manual, no reading, instability, shield, resistance, cable, and thermistor.
- GEOKON, Model 4420 Series Troubleshooting, swept-frequency excitation, mechanical range, interference, and cable continuity.
- Geosense, How to Troubleshoot VW Instruments, field resistance measurements and localization of common problems.
- RST Instruments, Vibrating Wire Soil Extensometer Manual, unstable readings, electrical noise, conductor resistance, and connection checks.
If a point disappears, returns, and every visit ends with a different hypothesis, send an anonymous excerpt of the history and a description of the chain. We will build a test card so that the next intervention ends with evidence, not another restart.