Summary: A vibrating wire sensor carries its measurement as a frequency rather than a voltage, and that one property is why the reading survives kilometres of cable, a bad splice and a wet joint, because resistance changes the amplitude of the returning burst and cannot change its frequency. The price is paid at the specification stage. Because the published conversion is a gauge factor multiplied by the difference of the squares of two frequencies, the strain value of one hertz is proportional to the reading, so the same 0.1 hertz resolution published by ZC Sensor’s ZC-AC104-4B-CAT1 is worth 0.12 microstrain at 200 hertz and 3.63 microstrain at 6,000 hertz, while the relative frequency resolution over the same window improves by the same factor of thirty in the opposite direction. A published gauge table giving 0.15, 0.51 and 0.98 microstrain across one band confirms that resolution is not a constant.
1. Why the Reading Is Not the Sensor
A vibrating wire reading looks like a single number, but three things produce it: the sensor, the acquisition unit, and the cable between them. Almost every procurement conversation is about the first of the three, and a great many field problems are in the second or the third. This guide takes the sensor as given and works through everything downstream of it.
There is a reason to start there. The vibrating wire family is unusual in one specific way, and the whole rest of this article follows from it: its output is a frequency rather than a voltage. A voltage is an amplitude, and amplitude is precisely what a long cable, a poor splice, a corroded terminal and a wet joint all degrade. A frequency is a rate, and none of those faults can change a rate.
That is the property that put this sensor family on dams, tunnels, mines and slopes, where the cable run is measured in hundreds of metres of wet ground and where nobody is going to re-terminate a junction box every season. A frequency signal tolerates conditions that would end the useful life of an analogue channel.
Read the same sentence from the other side and it becomes a warning. The immunity is a property of the signal, not of the system. A cable fault that leaves the frequency untouched can still destroy the burst that carries it, and a unit that reports a number without reporting anything about how that number was obtained leaves the user unable to tell a healthy reading from a marginal one. The honest version of the claim is narrower than the confident version, and separating the two is the work of the sections that follow.
One boundary belongs at the start. The gauges themselves come from many manufacturers, each with its own gauge factor, its own frequency band and its own housing, and the conversion from a frequency in hertz to a strain in microstrain is a property of the gauge rather than of the instrument that reads it. What an acquisition unit owes a user is four things: an accurate frequency, a temperature, a timestamp shared across its channels, and a channel that can be trusted. The figures used in this guide come from ZC Sensor’s own wireless vibrating wire datalogger ZC-AC104-4B-CAT1 and from datasheets published by gauge manufacturers, and each is attributed where it appears.
2. What Vibrates, and Why the Output Is a Frequency
The sensing element is a tensioned steel wire, hermetically sealed, with an electromagnetic coil beside it. The physics is the same equation that tunes a guitar string. The resonant frequency of a stretched string rises with its tension, and the relationship is f equal to one over two L, multiplied by the square root of T over mu, where L is the length of the wire, T is its tension and mu is its mass per unit length.
That single equation covers the whole product family, because the only thing the wire actually measures is its own tension. Attach one end to a diaphragm and the wire reads pressure. Attach it through a spring and it reads the opening of a crack or a joint. Build it into a load cell body and it reads force. Clamp it between two points on a structure and it reads the strain of that structure, because the structure stretches or shortens and the wire stretches or shortens with it. A piezometer, a crackmeter, a sister bar, a surface gauge and an embedment gauge are the same wire with different machinery at its ends.
The read cycle is a pluck and a listen. The unit pulses the coil, which pulls a small armature on the wire and then releases it. One instrument maker describes what follows with unusual precision: every frequency except the resonant one dies out within about 20 milliseconds, and the wire then continues to vibrate at its resonant frequency for some further time. As the wire moves through the magnetic field it induces a current in the same coil, so the burst that returns is a copy of the wire’s own motion, at the wire’s own frequency.
The readout counts the duration of a known number of cycles against a quartz reference. That step matters more than it appears to. Counting cycles against a crystal converts the measurement into a well conditioned digital quantity with no analogue-to-digital conversion in the signal path, which is a different situation from a millivolt or a 4 to 20 milliampere channel in which an amplifier sits in front of the converter. Published instrument literature makes the same point from the reliability side: the coil circuit contains no semiconductor devices, and the family carries integral ionised gas discharge protection against transients. A coil and a wire have very little in them to fail.
There is a second consequence, and it is where the long term reputation of the family comes from. A vibrating wire sensor is a mechanical sensor. The measurand acts on a piece of steel and the output is that steel’s own vibration, so published figures put zero stability at around 0.025 percent of full scale per year, which is a statement about how little the wire’s own properties wander rather than about how carefully the electronics are trimmed. Nothing in the readout chain can improve that number, and very little in it can degrade it either.
And a limit, stated at the outset rather than discovered later. The same instrument maker notes that vibrating wire sensors were historically not suited to dynamic monitoring, and that later developments made it possible. That is an argument for treating this as a static and quasi-static channel by default. It is also why the sections that follow are about resolution, temperature and cable integrity rather than about bandwidth and phase: a logger that wakes once a minute, or once an hour, is not a dynamic acquisition system whatever the sensor can now do on a test bench.
3. The Cable Test: What a Frequency Survives and a Voltage Does Not
Four things go wrong along a cable. Series resistance arrives from length, from a poor splice or from a corroded terminal. Shunt leakage arrives from moisture in a joint or a damaged jacket. Induced noise arrives from adjacent power cables or from a nearby strike. Intermittency arrives from a connection that was never properly crimped on stranded wire. Each of the four changes the amplitude of the returning burst. Not one of them can change the frequency of the wire.
The clearest published statement of this comes from an instrument maker comparing output types. It is not a problem, that source writes, if the cable is cut and then reconnected with a poor connection that increases the cable resistance, because cable resistance does not affect the frequency of the signal, only the amplitude of it. A second maker, describing piezometers in permanent installations, lists the faults the frequency signal is immune to in a single sentence: length to several kilometres, splicing, resistance, noise pickup and moisture. A third, writing on dam practice, puts the number at 5 kilometres of cable without loss or degradation due to resistance changes in the cable and connections.
Read those three claims as claims about the signal and they are all reasonable. Read them as claims about the system and they overreach, because a frequency is only the content of the burst and the burst also has to be detected. A wet joint lowers the amplitude. A long cable adds capacitance and rounds the edge. A shunt across the coil damps the wire so that the burst decays faster and carries fewer clean cycles. In every one of those cases the frequency that does come back is still correct, and the failure is a read that never completes, or a number built from too few cycles, or a value that repeats only sometimes.
The symptom is intermittency rather than drift, and that difference is the whole diagnostic. A drifting reading is a measurement problem. An intermittent reading is a wiring problem, and a wiring problem on this family is usually a two minute job with a multimeter rather than a site visit with a laptop.
This is also why a frequency reading needs a quality indicator rather than only a value. Practically, the three things worth recording on a commissioning sheet for every channel are the reading itself, the temperature, and how stable the reading is across several consecutive reads taken within a minute of each other. A channel that repeats to within one resolution step is healthy. A channel whose reading wanders while the structure is doing nothing is telling you about the coil circuit, not about the structure.
The one wiring rule that keeps the immunity intact is a screen rule. In a published investigation of vibrating wire strain gauges in a nuclear power plant structure, the screens had been tied down at both ends, and the conclusion was that the gauge signals may have been contaminated as a result, with the recommendation that system screens be tied down at one point as close to the instrumentation as possible. A screen tied at both ends is a loop, and a loop collects exactly the interference that a frequency signal was supposed to ignore. The same investigation found many terminations in junction boxes that had not been crimped on stranded wire, which is the other half of the same lesson: the frequency survives a bad connection and the burst does not.
4. What the Datasheet Window and the Resolution Figure Actually Buy You
ZC Sensor’s wireless vibrating wire datalogger ZC-AC104-4B-CAT1 publishes two numbers that sit next to each other on the datasheet: a frequency range of data collected from 200 to 6,000 hertz, and a resolution of 0.1 hertz. Read as a pair, the arithmetic looks reassuring. One tenth of a hertz out of 6,000 is one part in 60,000, and out of 200 it is one part in 2,000, so on either reading the count is far finer than any strain gauge needs.
That reading is correct and it is also useless, because frequency is not what the user buys. The user buys microstrain, and the conversion between the two is not proportional. The published convention for a vibrating wire strain gauge writes the change in strain as the gauge factor multiplied by the difference between the squares of the datum frequency and the measured frequency. Because the frequency enters as a square, the strain value of one hertz is not a constant. It is twice the gauge factor multiplied by the reading frequency, so it grows as the reading grows.
Put a number on that. On a published gauge factor of 3.025 times ten to the minus three microstrain per hertz squared, one hertz is worth 1.21 microstrain at 200 hertz and 36.3 microstrain at 6,000 hertz. Scale both to the datalogger’s published 0.1 hertz and the same figure is worth 0.12 microstrain at the bottom of the window and 3.63 microstrain at the top. The measurement resolution moves by a factor of thirty across the window, and the number printed on the datasheet has not changed at all.
The inversion is worth stating plainly, because it is where a careful reader and a careless one part company. Over that same window the relative frequency resolution improves by the same factor of thirty, from one part in 2,000 to one part in 60,000. So the two honest readings of one specification point in opposite directions by exactly the same factor. The number that gets better is the one being quoted; the number that gets worse is the one being measured.
| Reading frequency | Strain value of 0.1 hertz | What that means for the readout | Relative frequency resolution |
| 200 hertz | 0.12 microstrain | Finer than any gauge needs, so the count is not the limiting term here | 1 part in 2,000 |
| 1,000 hertz | 0.61 microstrain | Of the same order as a good gauge’s own resolution in the middle of its band | 1 part in 10,000 |
| 2,000 hertz | 1.21 microstrain | Coarser than a good gauge’s own resolution, so the count has become the limiting term | 1 part in 20,000 |
| 3,500 hertz | 2.12 microstrain | Coarser again, and this is where a common load cell band ends | 1 part in 35,000 |
| 6,000 hertz | 3.63 microstrain | Thirty times coarser than at 200 hertz, from the same unit and the same figure | 1 part in 60,000 |
The practical consequence is a change to the question asked at procurement. Ask for the resolution in microstrain at the frequency your gauge actually sits at, and ask for the gauge factor so that the arithmetic can be checked. A specification of 0.1 hertz is a complete answer to a question about frequency counting and an incomplete answer to a question about strain.
One further note about the window itself. Because the unit is designed to read whatever gauge is attached, its window has to be wider than any single gauge’s band, and published bands are indeed narrower than the window. An embedment soil strain meter and a load cell both publish 1,400 to 3,500 hertz, and a 125 millimetre gauge in a published nuclear plant investigation is described as usable from 500 to 1,400 hertz, a strain span of 4,100 microstrain. The 200 to 6,000 hertz window is therefore a compatibility statement. What decides the quality of your data is where your own gauge sits inside its own band, which is the subject of the next section.
5. Resolution Is Not a Constant, and One Published Table Proves It
A gauge datasheet that publishes its own resolution at three points in its band is worth more than a page of theory, and one does. Gage Technique International’s TES/5.5/T embedment strain gauge states a gauge length of 5.5 inches, a gauge factor of 3.025 times ten to the minus three microstrain per hertz squared, a measurement range greater than 3,000 microstrain, a coil resistance of about 100 ohms and a thermal coefficient of the vibrating wire of 11 parts per million per degree Celsius. It then gives a table of typical upper limit, mid range and lower limit readings with the resolution at each.
That table gives three resolutions: 0.98 microstrain at the upper limit, 0.51 at the middle and 0.15 at the lower limit. Those are not three measurements of one instrument in three moods. They are the arithmetic of a period counting readout, and the arithmetic can be reconstructed from the datasheet alone.
The readout in that datasheet displays the period of the wire in units of ten to the minus seven seconds, and the frequency is ten to the seventh power divided by that period. One least significant digit of the period therefore corresponds to a frequency step of the frequency squared divided by ten to the seventh power, which grows as the reading grows. Multiply that step by the strain value of one hertz, which also grows as the reading grows, and the strain resolution grows as the cube of the reading.
Check the model against the published table. At 625.00 hertz the period step converts to 0.0391 hertz, and 0.0391 hertz at that frequency is 0.148 microstrain, against 0.15 published. At 944.29 hertz the step is 0.0892 hertz and the resolution is 0.509 microstrain, against 0.51 published. At 1,176.50 hertz the step is 0.1384 hertz and the resolution is 0.985 microstrain, against 0.98 published. Three points, three matches, computed from the gauge factor and the period step and nothing else.
| Reading frequency in hertz | Frequency step of a period counting readout | Strain resolution, computed and as published |
| 625.00 | 0.0391 hertz | 0.148 microstrain computed, 0.15 microstrain published |
| 944.29 | 0.0892 hertz | 0.509 microstrain computed, 0.51 microstrain published |
| 1,176.50 | 0.1384 hertz | 0.985 microstrain computed, 0.98 microstrain published |
The cube law then predicts the shape of the whole table, and the table agrees with it. From the lower limit to the upper limit the reading rises by a factor of 1.88, and the cube of that is 6.67, while the published resolutions stand in the ratio 6.53. From the lower limit to the middle the reading rises by a factor of 1.51, whose cube is 3.45, against a published ratio of 3.40. The agreement is close enough that the model is not in doubt: across one gauge’s working band the resolution moves by a factor of nearly seven.
Two consequences follow for anyone specifying a system, and they are uncomfortable in the same way. The first is that a claim of a resolution better than some figure is an envelope statement about the whole band rather than a promise at every point in it, and the point that matters is the one your gauge actually occupies. The second is that two channels in the same project, read by the same unit, can have resolution floors differing by a factor of six or seven, because the two gauges sit at different points in their own bands. Nothing is broken. It is the shape of the conversion.
A qualification belongs here, because the cube law belongs to the readout and not to the sensor. The three published resolutions come from a unit that counts period. A unit that counts frequency to a fixed 0.1 hertz behaves differently across the same band, and the direction of its change is the one worked out in the previous section: on the same gauge factor, 0.1 hertz is 0.38 microstrain at 625 hertz and 0.71 microstrain at 1,176.5 hertz, a factor of 1.88 rather than a factor of 6.5. Same sensor, different readout design, different resolution curve. That is the reason to compare instruments on the quantity being measured rather than on the units they happen to count.
6. The Temperature Channel Is a Second Measurement
Every vibrating wire sensor carries a temperature sensor, and the reason is not that the manufacturer had space left in the housing. The resonant frequency of the wire responds to temperature as well as to load, because the length and the tension of the wire both move with it. A reading taken without a temperature is a reading of two effects added together, and the method for separating them is part of the standard data reduction rather than an optional refinement.
The convention across the family is a negative temperature coefficient thermistor of 3,000 ohms at 25 degrees Celsius, read on the same cable as the coil. Geokon publishes 3 kilo-ohms at 25 degrees Celsius with an accuracy of plus or minus 0.5 degrees Celsius for both its soil strain meter and its load cell, and an instrument maker’s field guide gives the resistance at nearby temperatures: about 3,478 ohms at 20 degrees Celsius, 4,714 at 15, and 5,971 at 10. ZC Sensor’s ZC-AC104-4B-CAT1 publishes the same interface from the reading side, with a 3 kilo-ohm applicable temperature resistance, a temperature range of minus 20 to 70 degrees Celsius and a temperature accuracy of plus or minus 0.5 degrees Celsius.
The published correction equation is a sum of two terms. The nuclear plant investigation quoted earlier gives the change in strain as the gauge factor multiplied by the difference of the squares of the datum and measured frequencies, plus a second term in which a coefficient difference multiplies a temperature difference. The coefficient difference is between the thermal expansion coefficient of the wire and that of the host material, so the correction is driven by a mismatch rather than by an absolute value.
That structure produces a fact worth carrying into any measurement plan. The same gauge reads differently in concrete and in steel, because the second term is a difference between the wire’s coefficient and the host’s. The published thermal coefficient of the vibrating wire in that gauge is 11 parts per million per degree Celsius. Handbook coefficients are of the order of 8 to 12 parts per million per degree Celsius for concrete and about 12 for steel, so the difference term is of the order of 1 to 3 microstrain per degree Celsius, and in a steel host it can approach zero. Two physically identical gauges in two different host materials, with the same gauge factor and the same resolution, do not have the same temperature sensitivity.
Now put the two channels side by side, and the result is not the one most users expect. A temperature accuracy of plus or minus 0.5 degrees Celsius, multiplied by a coefficient difference of 1 to 3 microstrain per degree Celsius, leaves a residual of 0.5 to 1.5 microstrain after the correction has been applied. The frequency channel of a gauge whose published resolution floor runs from 0.15 to 0.98 microstrain is in the same range. The residual left by the thermal correction is therefore the same order as the frequency resolution, and on the least favourable combination it is several times larger.
So neither channel dominates, and quoting only the frequency resolution describes about half of the measurement. This is not an argument against the 0.1 hertz figure. It is an argument against treating that figure as the whole error budget.
There is a practical response that costs nothing and is worth more than a better thermistor. The correction removes the bulk of the thermal term and leaves a residual; a comparison between two readings taken at the same temperature removes the residual as well, because the same temperature carries the same correction error on both readings. In a record that spans seasons that means comparing this month’s reading with the same month last year, or comparing readings taken at the same time of day, rather than comparing consecutive readings. It is the same logic that applies to any sensor whose zero drifts with temperature, and on a vibrating wire channel it is the difference between seeing a trend of a few microstrain and seeing the weather.
One interface note catches people out, and it is a matter of which range belongs to whom. The minus 20 to 70 degrees Celsius range next to the 3 kilo-ohm thermistor is the sensor side interface, because it is the range over which the gauge’s thermistor is specified. The datalogger’s own working environment is a separate published figure, outdoor at minus 10 to plus 60 degrees Celsius and 95 percent relative humidity at 40 degrees Celsius without condensation. A logger inside an enclosure on a tropical site can sit outside the second range while every reading it reports is inside the first, and both statements can be true at once.
7. Where a Vibrating Wire Channel Fails in the Field
The most useful thing about this sensor family is that almost all of its faults can be found with a multimeter, before anyone opens a software package. The principle is that the coil is a length of copper wire and the thermistor is a resistor, so both have a resistance that can be measured and compared with what the manufacturer says it should be.
An instrument maker’s troubleshooting guide publishes the expected coil resistances: about 160 ohms for a pressure transducer, about 180 ohms for a strain gauge, and about 50 ohms for a load cell, a sister bar or a miniature strain gauge, all within plus or minus 10 percent. Measured at the logger end of the cable, that number is the sum of the coil and the cable, so it also says whether the run is intact. A reading well above the expected value is a broken or high resistance conductor, a reading far below it is a short, and a value that is right at the junction box and wrong at the logger locates the fault between the two.
| Instrument | Expected coil resistance, within plus or minus 10 percent |
| Pressure transducer | About 160 ohms |
| Strain gauge | About 180 ohms |
| Load cell, sister bar or miniature strain gauge | About 50 ohms |
| Thermistor | 3,000 ohms at 25 degrees Celsius, about 3,478 at 20, about 4,714 at 15, about 5,971 at 10 |
The thermistor gets the same treatment, and its value moves with temperature, which is exactly what makes it useful rather than awkward. A reading of about 3,000 ohms says the sensor is near 25 degrees Celsius, about 4,714 says 15, and about 5,971 says 10. If the resistance is right at the sensor and wrong at the logger, the temperature conductors are faulty, and a strain channel whose temperature is wrong cannot be corrected however good its frequency is. The two channels are read from the same cable, and they fail separately.
The second class of fault is the one a frequency signal hides. In the published nuclear plant investigation, three of seven gauges behaved erratically and the typical returned frequency from those gauges was in excess of 1,700 hertz. The context makes that number diagnostic rather than merely odd: the same document describes a 125 millimetre gauge’s usable frequency range as 500 to 1,400 hertz, so a reading above 1,700 hertz is outside the band the gauge can physically occupy. The instrument was not measuring a strange strain. It was counting the wrong cycles. The follow-up inspection found screens tied down at both ends and terminations that had never been crimped on stranded wire, and concluded that the screening arrangement may have allowed noise from transients and adjacent cabling into the signal.
That produces the single most useful field rule in this guide. Record the as-built frequency band of every channel at commissioning, from the datum reading and the gauge’s own published range, and treat any later reading outside that band as a channel fault rather than as a measurement. A gauge cannot leave its band under load, so a reading that does is a statement about the wiring rather than about the structure. Put an alarm on that condition, not only on the magnitude of the strain.
The third class is surge. A vibrating wire coil has no semiconductor in it and, as published instrument literature describes, carries integral ionised gas discharge protection, so the family tolerates the transients that arrive on a long cable in an exposed location far better than a sensor with an amplifier at its front end. That is a real advantage and it is also a reason not to assume the acquisition unit is equally protected, because the transient that a coil survives still arrives at the terminals of the box. ZC-AC104-4B-CAT1 publishes compliance with GB/T 17626.2, the electrostatic discharge immunity test of the GB electromagnetic compatibility series, and with GB/T 2423.1, the cold test of the GB environmental test series, which is the acquisition side of the same problem.
8. What the Datalogger’s Own Numbers Decide
A datalogger is bought for two reasons, and a datasheet answers only one of them well. The first is the ability to read the sensor, which is settled by the frequency window and the excitation scheme. The second is whether the unit will still be there in three years, which is settled by the power budget, and this is where a table of currents can be read carefully or carelessly.
ZC-AC104-4B-CAT1 publishes five current figures: 20 microamperes asleep, below 45 milliamperes while obtaining a position fix, below 100 milliamperes while acquiring vibrating wire data, below 700 milliamperes while accessing the internet, and below 300 milliamperes while uploading data. It also publishes a 4.2 volt, 9,000 milliampere hour rechargeable pack built from two INR21700 cells, a collection interval from 0 to 65,535 minutes, a position update once every 24 hours, an SD card holding more than 8,000 readings with loop refresh, support for a primary and a secondary target server, a UART and TTL configuration port, IP67 protection, and a working environment of minus 10 to plus 60 degrees Celsius at 95 percent relative humidity without condensation.
What the datasheet does not publish is the duration of each state, and without durations the currents cannot be turned into a life figure. That is not a defect. It is the honest situation, because the duration of a scan depends on how many channels are connected and how the unit is configured. The way to close the gap is to measure it once: put a current logger on the supply, or read the unit’s own diagnostic output, and time one scan and one upload on the bench with the actual sensors attached. Two numbers, and the rest of the arithmetic is multiplication.
| State | Published current | What decides its share of the budget |
| Sleep | 20 microamperes | Twenty four hours a day, but five thousand times smaller than a scan, so it never dominates unless the interval is very long |
| Positioning | Below 45 milliamperes | One fix every 24 hours, so it is a fixed daily cost that does not change with the interval |
| Vibrating wire acquisition | Below 100 milliamperes | The number of scans per day multiplied by the length of one scan, which is the term the interval multiplies |
| Accessing the internet | Below 700 milliamperes | The number of uploads, so it follows the upload schedule rather than the scan schedule |
| Uploading data | Below 300 milliamperes | The number of uploads multiplied by the length of each upload |
That table already settles one question without any durations at all. Sleep is 20 microamperes and a scan is below 100 milliamperes, a ratio of 5,000, so the life of the unit is decided entirely by how often it wakes and how long it stays awake. At a one minute interval there are 1,440 scans a day and the acquisition term outweighs the sleep term by one to two orders of magnitude even if a scan lasts a fifth of a second. At an interval of a day there is one scan and sleep stops being a rounding error in the other direction. The interval, not the battery, is the lever.
The interval range deserves a second look, because it is much wider than most users expect. Zero to 65,535 minutes is zero to about 45 days at the top of the range. Practically, almost every structural monitoring application lives in the first hundred minutes of that range, and the top end exists for the case where a reading is wanted seasonally rather than daily. The point of noting the range is that a choice at the bottom and a choice at the top are not variations of the same decision. The bottom is a power problem and the top is a data problem.
The local memory is a buffer rather than an archive, and the arithmetic says so. More than 8,000 readings across four channels is 2,000 complete scans. At one scan an hour that is about 83 days of history; at one scan a minute it is under a day and a half. The card’s job is to survive a period without a network, which is what makes the dual server and the position fix part of the same story: the unit records locally and uploads when it can, and the daily position fix is what lets a recovered file be tied back to a location without a site visit.
Finally, a four channel unit is a single point of failure for four channels, and that is worth saying plainly. The nuclear plant investigation quoted earlier makes the same observation from the other direction, noting that a deficiency in the logging equipment could affect the complete system and all the gauges. The counterweight is the shared timestamp: four gauges read by one unit are read in the same scan, so their readings can be compared with each other without any time alignment, which is worth a great deal when the purpose of the array is to compare four locations on one structure. Distributed acquisition buys independence and gives that up. Where the same acquisition pattern is applied to in-place inclinometer strings instead of vibrating wire gauges, ZCT-IPI-W1 reads up to 64 probes on a single string at intervals from 1 minute to 10,080 minutes, with a local memory of 60,000 readings and an upload to the user’s FTP server over the same 4G network.
| Parameter | ZC-AC104-4B-CAT1 | ZCT-IPI-W1 |
| What it reads | Up to four vibrating wire sensors | Up to 64 in-place inclinometer probes on a single string |
| Link and destination | 4G CAT1, primary and secondary target server | 4G CAT1, upload to the user’s FTP server |
| Local memory | SD card, more than 8,000 readings, loop refreshed | 60,000 readings |
| Collection interval | 0 to 65,535 minutes, about 45 days at the top | 1 to 10,080 minutes, seven days at the top |
| Sensor side interface | Vibrating wire pluck and count | RS485 at 2,400, 9,600 or 19,200 bits per second |
| Supply and environment | 4.2 volts direct current, 9,000 milliampere hours rechargeable, minus 10 to plus 60 degrees Celsius, IP67 | 8 to 30 volts direct current, minus 20 to plus 60 degrees Celsius |
9. Vibrating Wire and a Wireless Tilt Node on the Same Asset
The two families are usually treated as alternatives in a procurement table and as complements on a site. The distinction that survives contact with practice is about geometry rather than about precision. A vibrating wire gauge measures a change in length across its own gauge length, which is a point measurement on a member. A tilt sensor measures the attitude of a segment, which is a statement about a length scale far larger than the sensor itself.
That difference decides which one answers which question. If the question is how much a tendon has been stressed, whether a joint has opened, or what the pore pressure behind a wall is doing, the answer lives in a vibrating wire channel and a tilt sensor cannot supply it. If the question is whether a pier has leaned, whether a rack has gone out of plumb, or whether a slope is rotating, the answer lives in a tilt or inclinometer channel and a strain gauge cannot supply it. On a tailings facility or a dam both appear in the same monitoring plan, and they are not substitutes for one another.
There is a second distinction, about time, and it runs opposite to what most people assume. A vibrating wire channel looks like the high performance option because its resolution is quoted in fractions of a microstrain, and in practice it is also the slower channel of the two, for the reason given in section 2. That makes the interval rather than the resolution the decision that matters. A logger on a fifteen minute interval is a static system, whatever it is reading.
This is why reading a vibrating wire record correctly is mostly about time and temperature rather than about resolution. The interesting content of a dam or slope record is a trend of a few microstrain per month, and at that size the thermal term and the position of the gauge inside its own frequency band both matter more than the resolution of the count does. Three practices follow, and they are the ones worth writing into a monitoring plan: compare readings at matched temperatures, record the as-built frequency band of every channel so that a fault can be separated from a measurement, and measure one scan and one upload so that the reporting interval can be chosen from arithmetic rather than by default.
The transferable rule is this. When a measurement is carried as a frequency, or as any other carrier that is not the quantity the user wants, the resolution figure on the instrument is a resolution of the carrier and not of the measurand. The conversion between the two is the derivative of the sensor’s own transfer function, which means the resolution a channel delivers in the units that matter is a property of where the sensor sits on its curve rather than a property of the instrument. A specification written in carrier units should be converted into measurand units at the point of use before it is compared with anything, including another specification.
What remains outside this article is the boundary that applies to every sensor in this series. Nothing here decides whether a measured strain is acceptable. That question belongs to the design of the structure and to the standard governing it. What the acquisition chain decides is whether the number that reaches the record is a measurement or an artefact.
Frequently Asked Questions
- Can I read a vibrating wire sensor with a normal analogue input? No. A vibrating wire channel has to do two things an analogue input cannot: deliver a pluck pulse to the coil, and count the frequency of the burst that comes back. A 4 to 20 milliampere or a voltage input measures amplitude, and the amplitude of the returning burst is exactly the part the cable degrades. The published coil resistances are the fastest way to check the channel before blaming it: about 160 ohms for a pressure transducer, about 180 ohms for a strain gauge and about 50 ohms for a load cell, each within plus or minus 10 percent, measured at the logger end of the cable.
- Why does my vibrating wire reading change when the cable gets wet? Usually it does not change, and that is the point. Water in a joint adds a shunt resistance and lowers the amplitude of the returning burst, but resistance cannot move a frequency, so the number that comes back is normally still the right number. What a wet joint damages is the signal to noise ratio of the burst and eventually the unit’s ability to detect it at all, which is why the symptom is normally an intermittent reading or a failed read rather than a shifted one. A reading that has genuinely shifted points at the coil circuit or at the screen, not at the moisture alone.
- Is 0.1 hertz the resolution of my measurement? Only if the quantity you want is frequency. If it is microstrain, the resolution is 0.1 hertz multiplied by twice the gauge factor and the reading frequency, so it moves with the reading. On the published gauge factor used in this guide, 0.1 hertz is 0.12 microstrain at 200 hertz and 3.63 microstrain at 6,000 hertz, a factor of thirty across the window that the acquisition unit itself quotes. The same figure on the same datasheet is a good resolution at one end of the window and a modest one at the other.
- Do I need the temperature reading if I am only interested in the trend? Yes, and for a specific reason. The trend is exactly the part of the signal that the thermal term contaminates, so a temperature channel is what lets a real trend be told apart from a seasonal one. If the instrument’s temperature accuracy is plus or minus 0.5 degrees Celsius and the wire’s thermal coefficient differs from the host’s by 1 to 3 microstrain per degree Celsius, the residual after correction is 0.5 to 1.5 microstrain, which is the same order as the frequency channel’s own resolution. Reading the trend between points taken at the same temperature removes what the correction leaves behind.
TAG: