Mount the vibration sensor at a point that follows the assessment objective, on a stiff part of the observed element, using a fastening method verified for the required band. Document the axes, tightening torque or adhesive, surface preparation, cable routing, environmental conditions and post-installation test. The foundation, wall and slab are not interchangeable locations: each shows a different structural response.
Most important before entering the site
- First choose the phenomenon and criterion, then the point and fastening type.
- Plaster, a loose tile, a soft pad and a thick adhesive layer create their own dynamic system.
- An orientation error mixes components. A photo of the sensor without axis labels is not enough.
- Cable motion can generate a signal. The cable must be strain-relieved, secured and kept away from interference sources.
- The platform will also show a very precise bad measurement. Acceptance takes place in the field, not on the basis of an attractive chart.
The most deceptive measurement failure does not give you a blank screen. It gives you a beautiful chart. The dominant peak is sharp, alarms arrive, and the data carry complete timestamps. Yet the sensor was glued to a finish layer, the bracket loosened after a few days, or the lead strikes the housing at every gust of wind.
The software does not know whether the transducer moves together with the structure. It only sees numbers. That is why the installation protocol is part of the result, not an attachment to be handled later. ISO 5348 was created precisely to describe the technical properties of different vibration transducer mounting methods and their influence on bandwidth and measurement fidelity.
The measurement point answers a specific question
The question "foundation, wall or slab?" does not have one correct answer. The choice depends on the propagation path, the required metric and the assessment document. For works near buildings, the relationship between criterion, direction and point is explained in the guide on building vibration monitoring according to PN-B-02170.
Foundation or the part of the building connected to the ground helps describe vibrations introduced from the substrate. It may be the right location for assessing the impact of earthworks, piling, demolition or heavy traffic, if the adopted method says so. It does not automatically tell you what happens on the top slab.
A load-bearing wall or column element shows the response of a specific part of the structure. The point should have a clear relationship to the load-bearing system. Mounting on detached plaster measures the layer, which may resonate and fall away independently of the wall.
A slab is important when assessing service response and human comfort, because it can amplify vibrations near natural frequencies. A location near a support and at midspan gives a different result. If the criterion concerns the place where people are present, the point should represent that space in line with the relevant procedure.
A machine and its foundation are not the same point either. A sensor on the housing diagnoses the source and rotating components. A sensor on the foundation shows part of the energy transferred into the structure. One does not replace the other if the decision concerns the transmission path.
| Purpose | Starting point for the design | What can distort the conclusion |
|---|---|---|
| impact of works on a neighbouring building | a point required by the standard on a stiff element connected to the structure | mounting on cladding or on an arbitrary part of the facade |
| slab response and comfort | a representative slab point in the assessed room | measurement only on the foundation |
| transmission from a machine | source, foundation and selected receiving point | only one point, with no possibility of separating the path |
| diagnosis of local resonance | points on both sides of the connection or element | transferring the result to the whole structure |
| documentation of a construction event | fixed points with known geometry and directions | moving the sensor without a new baseline protocol |
ISO 4866 covers transducer location and mounting as part of structural measurement planning. The standard itself does not give one universal location for every source and building. The measurement designer must combine the assessment document requirements with a propagation-path model and on-site access.
Fastening changes the characteristics of the entire measurement chain
An accelerometer does not measure "through air". Motion must pass from the tested element through the surface, bracket and base to the sensing element. Every connection has mass, stiffness and damping. If it is too soft or loose, an additional filter and resonance appear.
ISO 5348:2021 lists not only the mounting method, but also issues such as misalignment, base bending, cable motion, temperature, electric and magnetic fields, and tightening torque. The standard makes it possible to estimate the limits of a given installation method. This is an important distinction: "the sensor has a bandwidth up to X" applies under specific mounting conditions, not to any adhesive layer on rough concrete.
NI documentation indicates that threaded mounting usually provides the widest usable bandwidth, while the added mass of a magnetic or bonded base lowers the resonant frequency of the system. PCB shows the same relationship in installation instructions and recommends a smooth, flat surface and a torque that matches the documentation for the specific sensor.
This does not mean that every building sensor must be screwed directly into the structure. In the 1-100 Hz band, a properly designed rigid bonded pad may work well enough, if the manufacturer allows that variant, the substrate is load-bearing, the adhesive is suitable and the whole assembly passes testing. However, "well enough" must mean a tolerance defined before installation and referenced to the selected metric: PPV, RMS, FFT or third-octave bands.
Comparison of mounting methods
| Method | Advantage | Limitation | What to record in the protocol |
|---|---|---|---|
| thread or anchor | high stiffness and repeatability | intrusion into the element, preparation required | connector type, depth, torque, surface |
| rigid bonded base | no thread in the sensor and easy removal | dependence on adhesive, substrate and bond thickness | product, batch, preparation, temperature, curing time |
| magnet | fast installation on steel | only suitable substrate, added mass, risk of shifting | surface condition, geometry, holding force |
| wax or temporary adhesive | quick laboratory test | temperature and limited bandwidth, low durability | use only in a qualified test |
| handheld probe | quick walk-through | low repeatability and limited bandwidth | operator, pressure, angle and reference point |
In long-term building monitoring, removal matters too. The base should allow the sensor to be replaced without losing the definition of the point. After replacement, you need to restore the orientation and perform a comparative test, not just copy the device name into the system.
Twelve mistakes that look like real vibration
1. A point chosen for convenience of access
The installer chooses a place near power supply or a ladder, although the criterion concerns a different element. The result may be correct for that point and useless for the decision. Approve the location on the drawing before drilling.
2. Mounting on a finish layer
Plaster, insulation, tile, paint with weak adhesion and loose sheet metal do not guarantee coupling to the structure. Remove the weak layer or use a designed connector to the load-bearing element. Conservation or owner approval may limit intervention, in which case the limitation must be written into the report.
3. A soft or too thick bond line
An elastic adhesive behaves like a spring and a damper. Excess material increases the distance from the substrate and may lower the resonant frequency of the assembly. PCB recommends rigid adhesives and a thin layer on a smooth surface when bonded bases are used.
4. Dirt, dust, rust and unevenness
A small bump can make the base rest on an edge. The connection then works under every impulse. The protocol should describe cleaning, degreasing, levelling and surface condition, and a photo with a scale bar later helps find the same point again.
5. Wrong torque or a connector bottoming out
Too little torque gives weak coupling. Too much can damage the thread or the sensor. If the stud bottoms out before the base is clamped, the connection looks tight, yet the surfaces do not touch properly. Use the documentation for the specific assembly and a torque wrench with control.
6. A bracket with its own compliance
A long angle bracket, thin plate or printed adapter may have a resonance in the band of interest. Then you are analysing the bracket. Qualify the bracket together with the sensor, not as a neutral mounting element.
7. Axes described "by eye"
An ideal sensor rotated by an angle θ splits one component into cos θ in the main axis and sin θ in the transverse axis. At a 5° error, the transverse component is about 8.7% of the signal, and at 10° about 17.4%. This is a simple geometric effect, independent of the transducer's own cross-axis sensitivity.
On the protocol, draw the positive direction of each axis relative to the building, not relative to the lettering on the housing. The note "vertical" without information on which direction is positive makes comparison after replacement difficult.
8. A cable that works like a sensor
Cable motion can generate triboelectric noise, load the connector or transfer force to the housing. HBK recommends securing the cable close to the transducer and routing it away from strong electromagnetic fields. Leave controlled slack for thermal movement, but not a free loop that strikes the structure.
9. A ground loop or a shared route with power cables
A shield and ground connected at multiple points can introduce interference. A cable running alongside a drive, inverter or power line may pick up electrical components. Do not repair such a signal with filtering before checking the installation and grounding.
10. Water, temperature and no strain relief at the connector
A sealed housing does not mean a resistant connector after poor installation. Condensation, solar radiation, flooding and temperature cycles can change sensitivity or cause corrosion. Record gland orientation, drip loop, sealing and temperature range.
11. Sensor mass changes a thin element
NI gives a practical rule that the accelerometer mass should be much smaller than the mass of the tested element, and roughly no more than 10%. For a reinforced concrete wall the problem is usually small, but for thin sheet metal, a light cover or a laboratory model, the sensor with its base may shift the natural frequency.
12. No post-installation test
A zero reading in silence does not verify the full bandwidth. You need a controlled excitation, axis comparison, noise, polarity and repeatability. If the point is fixed, keep a characteristic acceptance record for comparison after service.
Numerical example: sensor rotation changes interpretation
Assume the actual vertical acceleration during a controlled test has an amplitude of 0.10 m/s², and the horizontal component is negligible. The sensor was rotated by 10° relative to vertical.
The ideal axis described as vertical will show:
0.10 × cos(10°) ≈ 0.0985 m/s²
The transverse axis will show:
0.10 × sin(10°) ≈ 0.0174 m/s²
If the analyst does not know about the rotation, they may treat 17.4% of the component as real horizontal motion. In a real measurement, cross-axis sensitivity, mounting inaccuracy and structural response also appear, so a simple calculation must not be used as an automatic correction. The example shows why a photograph with a compass or an axis sketch has measurement value.
Acceptance test on site
The test should distinguish an error of point, fastening, axis and electrical path. First record silence with the known source switched off. Then apply a repeatable, safe excitation at a defined location. This may be a shaker, a field calibrator or a procedure agreed by a specialist. Hitting with a hammer without force control is suitable at most for checking whether the system responds, not for amplitude calibration.
Compare three repeats. Assess time, sign, amplitude, spectrum and axis relationship. Check the behaviour after moving the cable without exciting the structure. If the cable itself creates a peak, the cause is already known.
Then perform a digital data-path test: UTC timestamp, point ID, unit, full waveform, completeness, test alarm and export. Check record parameters according to the explicit specification for sampling, trigger and pre-trigger, not only the stated frequency. Installation acceptance ends only after the field evidence and the platform evidence are linked.
If the record may later support a claim, the installation protocol, control record and configuration must be preserved together. This evidential continuity is described in the guide on monitoring data as evidence in a dispute.
Installation protocol checklist
- [ ] The measurement objective and assessment document were identified before the point was selected.
- [ ] The point was marked on the plan, section and photograph.
- [ ] The load-bearing element and all intermediate layers were described.
- [ ] The fastening method is permitted for the sensor and the bandwidth.
- [ ] The surface is clean, load-bearing, flat and dry.
- [ ] The connector, base or adhesive type was recorded.
- [ ] Torque or conditions and curing time were recorded.
- [ ] The positive directions of the axes were referenced to the structure.
- [ ] The cable has strain relief and does not strike the structure.
- [ ] The route was checked relative to power supply and electromagnetic fields.
- [ ] Connectors and pass-throughs were protected against water.
- [ ] The mass of the assembly does not change the response of a light element.
- [ ] The result of the silence test and controlled excitation was retained.
- [ ] The test was repeated after enclosure closure and construction work.
- [ ] The protocol identifies the person, date and reference equipment.
How Inclify supports this
Inclify stores vibration events as complete records with time and analysis results. For each event, you can view the waveform, FFT spectrum and 21 third-octave bands from 1 to 100 Hz. Dynamic signal calibration is set separately for each axis. This makes it possible to retain different coefficients instead of assuming one for the entire assembly.
The 4th-order Butterworth filter profile calculates RMS and MAX in the bands. The dynamic alarm compares active bands with the reference envelope and operates with hysteresis. The full record makes it easier to verify whether the alarm came from a real event, short contact with the housing or noise. You still need the installation protocol, because the waveform shape alone rarely proves the cause.
The platform also shows the acceptance progress of dynamic fragments and the explicit status of an incomplete record. An administrator can review the device communication log over the configured period. This helps separate a mounting problem from transmission loss.
Inclify does not automatically detect a loose anchor, a thick bond line, swapped axes or a sensor glued to plaster. It does not know the tightening torque if the team did not document it. It also does not replace sensor calibration or acceptance of the entire chain against ISO 5348, ISO 4866 or the building assessment standard.
Limits of interpretation
Two correctly mounted sensors in different places may show different results, because the structure really moves differently. Do not treat a discrepancy as a fault without analysing the vibration shape, propagation path and phase. On the other hand, agreement between two points does not prove correct mounting if both use the same wrong procedure.
Agreement of amplitude at one frequency also does not qualify the whole bandwidth. A soft mount may work properly at 5 Hz and amplify or damp 80 Hz. Test several frequency points important for the decision.
Finally, the location appropriate for building damage does not have to be the right one for human comfort. These objectives require different points and often different quantities. They are separated in the article on building vibrations, damage and human comfort.
FAQ
Where should a vibration sensor be mounted on a building?
At the point required by the adopted assessment method and representing the transmission path of vibrations or the response of the tested element. For ground-borne vibrations, a stiff part of the structure close to the foundation is often considered. For comfort, a slab in the assessed room may be needed. The final choice is approved by a specialist based on the source, structure and standard.
Can an accelerometer be glued to concrete?
Yes, if the manufacturer allows a bonded base, the substrate is prepared, the adhesive has suitable stiffness and durability, and the assembly meets the required bandwidth. Do not glue it to dust, paint with weak adhesion or a loose layer. Record the product, bond thickness, temperature and curing time, then perform an acceptance test.
Does the sensor have to be perfectly level?
It depends on the device design and the processing method, but the axis orientation must always be known. Even when the device compensates for gravity, rotation mixes components in the reported axes. For directional assessment, record the positive directions relative to the building and control them after every replacement.
How can I check whether the sensor mounting is loose?
Apply a repeatable excitation and compare the response with the acceptance record. Look for amplitude changes, new resonances, unstable phase and a response to gentle movement of the cable or housing. Visual inspection and torque control remain necessary. An algorithm cannot replace a physical check of the connection.
Is a magnet enough for long-term monitoring?
It may work on a suitable steel, smooth surface and in the required band, but it is sensitive to shifting, corrosion, dirt and access by unauthorised people. An additional base changes the frequency response. A long-term installation must be mechanically secured and qualified by testing, not by installation convenience.
How often should a vibration sensor installation be checked?
After installation, after nearby works, after any event that may disturb the bracket, after service and at intervals determined by environmental risk. The check may include visual inspection, connectors, cable, torque and a repeatable test. The schedule should take into account access, humidity, temperature, vibration and the consequences of a false measurement.
Sources and further reading
- ISO 5348:2021, Mechanical mounting of accelerometers - properties of mounting methods, influence on response and sources of installation error.
- ISO 4866:2010, Vibration of fixed structures - principles for planning structural vibration measurements, including selection of equipment, location and mounting.
- NI, Measuring Vibration with Accelerometers - axes, sensor mass, mounting methods, environment and signal conditioning.
- PCB Piezotronics, Guidelines for Mounting Test Accelerometers - surface preparation, threaded mounting, bonded bases and bandwidth influence.
- PCB Piezotronics, How Sensor Mounting Affects Measurements - comparative measurements of response under different mounting methods.
- HBK, Environmental Effects on Accelerometer Measurements - cable noise, ground loops, moisture, temperature and cross-axis sensitivity.
Next step on site
Choose one critical point and ask the contractor for a full protocol: element, layers, fastening method, axes, cable, photos and three repeats of a controlled test. If the installation method cannot be reproduced, the result cannot be defended. Talk to the Inclify team if you want to connect such field acceptance with a view of full events and alarms in a pilot.