Wireless vibration sensor: how to choose and what drives the cost

Anyone shopping for a wireless vibration sensor runs into two problems. The technical one: datasheets carry numbers that cannot be compared with each other. And the commercial one: almost nobody publishes prices, and quotes come in different formats. This article deals with both: the criteria that change the result, what makes up the cost and what to ask.

Criterion 1: the useful frequency range

This is the criterion that separates sensors the most, and the one most dressed up in datasheets.

• From 10 Hz to 1,000 Hz is the band in which ISO 20816-3 evaluates vibration severity. That is where unbalance, misalignment, looseness and electrical problems live. Any serious sensor covers this range.
• Above 1 kHz is where a bearing warns first, where a gear shows its gear mesh frequency and where cavitation appears. These three defects usually show up first above 5 kHz, which is why a narrow-band sensor misses the start of the problem.

The detail almost nobody checks: the useful range depends on the mounting. A sensor that responds up to 20 kHz on a threaded stud drops to 3 to 4 kHz when held by a magnet on an unprepared surface. Ask for the useful range with the mounting that will actually be used.

Criterion 2: the sensing element

Piezoelectric sensors still have the lowest noise and the widest bandwidth, and they are the choice when the target is early-stage bearing damage, gear mesh and cavitation. MEMS sensors are smaller, cheaper and able to run for years on a battery; current ones have come close to piezoelectric performance in much of condition monitoring, which is why they dominate the wireless market. Their historical limitations were noise, bandwidth and acceleration range.

There is no single answer. There is the right question: which failure mode do I need to catch, and at what frequency does it appear?

Criterion 3: sampling and what gets sent

To see up to a given frequency, you must sample at least twice that frequency; in practice 2.56 times is used, because of the anti-aliasing filter. Covering the standard’s band takes about 2.56 kHz; seeing an early bearing fault requires sampling above 10 kHz.

What comes out of the sensor also matters: overall values (RMS per axis, peak, temperature, dominant frequency), sent frequently to feed trends and alerts, and the FFT spectrum, which is heavy, on demand, for diagnosis. A sensor that only sends one number tells you something changed but not what. The combination of the two layers is what works, and it is what our platform uses.

Criterion 4: axes, temperature and mounting

Triaxial sensors cover all three directions at one point, as long as the orientation is recorded; uniaxial sensors cost less and require getting the direction right. Built-in temperature comes with almost all of them and is very valuable: friction and lack of lubrication show up as heat along with, or before, vibration. Mounting determines the bandwidth and the life of the data: a threaded stud is best, structural adhesive is close behind, and a magnet is for comparison, not for permanent monitoring. Details in where to mount the sensor on an electric motor.

Criterion 5: network and battery

Two architectures dominate: short-range radio to a gateway, which connects out over the cellular network or the plant network, and cellular in the sensor itself, which does away with the gateway at the cost of higher power consumption and one data plan per device.

Battery life has no single number: it is a consequence of the transmission rate. Ask for the stated battery life for the configuration you will use, and ask whether the battery can be replaced in the field. Check the ingress protection rating against dust and water, the temperature range and, in hazardous areas, the certification for explosive atmospheres.

Criterion 6: what the platform delivers

Half the value is here, and it is the part least compared. How long does it keep history? Does it compare against a standard threshold, against yours, or both? Does the alert go out only when the level changes, or at every reading above the limit (the second option kills the program)? Does it have an API and export, in other words, who owns the data?

What makes up the cost

Five components, and honest quotes show all five:

1. Sensor, per monitored point. A well-instrumented motor usually has two points, one per bearing.
2. Gateway or connectivity, per area or per device, plus the data plan.
3. Installation and commissioning: surface preparation, mounting, asset registration, setting thresholds and the baseline. This is the step that most affects data quality, and the one that most often disappears from quotes.
4. Platform, almost always per point per year.
5. Analysis and support: from nothing (you read the data) to a periodic signed report.

Add what comes later: battery replacement, replacing lost sensors and the team’s time spent looking at alerts.

Compare per monitored point, per year

A quote with a cheap sensor and an expensive platform can cost more than the reverse. The only comparison that adds up is total cost per monitored point per year, over three years, installation included. Ask every vendor for it in that format, and also ask for the cost of scaling up.

The market includes companies that sell sensors with built-in cellular connectivity and their own platform, as well as product lines from bearing and instrumentation manufacturers with decades in the business. These are different architectures, all legitimate. We do not quote competitors’ prices because there is no comparable public figure, nor do we claim to be cheaper without your plant on the table. Our difference is of another kind: we design and manufacture the sensor, the gateway and the platform in-house, in Belo Horizonte, Brazil, and the people who install it are the people who built it.

Questions to ask when requesting a quote

1. What is the useful frequency range with the proposed mounting?
2. What sampling rate, and is the spectrum available on demand?
3. What is the battery life with the recommended transmission settings?
4. How many gateways cover the plant, and what happens when the signal drops?
5. What is the cost per point per year, over three years, with installation and thresholds included?

If you want to know what your plant needs, talk to one of our engineers.

---

Sources: ISO 20816-3:2022 (evaluation band of 10 Hz to 1,000 Hz); ISO 5348:2021, Mechanical mounting of accelerometers (mounting methods and their effect on frequency response); ISO 17359:2018 (selection of parameter and measurement interval); Analog Devices, How to Choose the Best MEMS Sensor for a Wireless Condition-Based Monitoring System (MEMS versus piezoelectric, bandwidth and noise); manufacturers’ public product pages, accessed in September 2026.