Size Matters!

Designing a sensor for wireless condition monitoring is a complex balancing act between meeting performance requirements whilst adding the most value for the end user. The choice of technology becomes very critical when trying to optimise certain aspects of the design. Form over function, or vice versa is just one of the design directions that needs to be established very early in development. Thus, choosing an appropriate power source is one of the earliest decisions that needs to be made.  

Here, we will focus on the implementation of the battery. Batteries are available in all shapes and sizes, different chemistries, non-chargeable, re-chargeable, etc. However, the form factor of the power source will largely dictate the overall dimensions of the sensor. Many wireless sensors coming to market have very large batteries in order to meet the longevity requirements that are needed by customers to trend their equipment performance over many maintenance cycles. If a more power-hungry technology is used to perform measurements and serve the wireless communications, then the battery consumption will be substantial. However, a larger battery can have many knock-on effects. A larger battery adds additional weight and influences the centre of gravity, which most likely increases the size of the magnet or mounting bracket needed to fasten or secure the sensor. It is difficult to describe how important and impactful this component selection is on the outcome of the product design.

Vibration analysts are calling out for smaller, compact sensors which means a tight balancing act between battery size and battery life. Most customers and product users are happy to concede a replaceable battery design, as a way of lengthening product service. However, this can often make ingress protection more challenging, as the design must allow easy access to change the battery but still maintain a high level of robustness. Having a small, lightweight, low centre of gravity sensor with a compressed mounting footprint, will allow the sensor to fit into tighter and more “ideal” surface mounting locations. This is very important when measuring vibration and temperature close to the source of the vibration.

Kappa X uses a ½ AA battery, which is one the smallest power cells on the market, yet it boasts a 4 year battery life (standard configuration). The battery cell is also replaceable, which often makes ingress protection more challenging, however, Kappa X is fully waterproof (validated IP69K). The in-house knowledge and expertise of true low-power electronics at Sensoteq enables Kappa X to have a small 25mm diameter mounting footprint and overall compact size. Having this small, lightweight, low centre of gravity sensor with a compressed mounting footprint, allows for Kappa X to fit into tighter and more ideal surface mounting locations. This is very important when measuring vibration and temperature in close proximity to the source of the vibration. To summarise, a high performing sensor, that you cannot mount correctly, has diminishing benefits in terms of machine health analysis. Ultimately, measurement performance and determination of early failure are very closely tied when it comes to adding the most value for the end user.

Kappa X makes no compromises when it comes to the designed, developed and implemented technologies. It uses a unique communications protocol capable of providing a reliable transmission over long-range in complex and challenging industrial environments. The sensing technology within Kappa X has been researched in collaboration with industry experts to deliver a market leading product.

By David Procter

Frequency is Everything

In the world of vibration analysis frequency is everything.  Frequencies are generated by many sources within a machine. The running speed of the machine is one example frequency; this is the velocity of the rotating element which in many cases is a drive shaft connected to an application (like a fan or a pump). Each frequency will have a magnitude associated with it; this is the total amount of energy of that specific frequency.  The amount of energy of any given frequency can tell us a lot about what is going on within a machine.  When we use a vibration sensor to measure a machine, this is what we are detecting, frequencies and magnitudes of those frequencies.

The maximum frequency that a sensor can measure is called the Fmax.  This value lets the user know what type of faults they can detect on a machine. It is typically listed in hertz (Hz).  The higher the Fmax value, the greater number of faults that a sensor will detect, but it will also allow for earlier indication of potential faults, like a bearing failure. Some ISO standards will reference an Fmax of 1kHz – whilst taking a reading up to 1kHz is suitable for most acceptance testing, it will not highlight even the most basic of bearing faults.  A minimum of 2.5kHz Fmax is recommended for good coverage.

Recently, a new sensor has entered the market that boasts an Fmax up to 10kHz. This Fmax gives great coverage for a wide variety of faults, and will accurately inform users of issues on complex equipment like gearboxes that have many higher fault frequencies. In addition, a higher frequency will give earlier indication of bearing faults. Bearing failures will typically start within the subsurface of the metal with a very small amount of energy being emitted to the sensing element, thus having a sensitive measurement device with the right Fmax is critical to understanding the health of your machine early, prior to failure impacting any processes or causing downtime.

By David Procter

Sub GigaHertz Advantages for the Industrial Environment

Let’s discuss for a moment; how can we design a decent, standalone sensor which can be put on the vibrating equipment, stay there and report reliably the measurements.

Let’s look at very frequent set of requirements:

• It is going to withstand (and measure) potentially very high g forces in excess of 50g
• It is going to withstand (and report) temperature over 60°C (140°F)
• The place is hardly accessible (not good for frequent human inspection)

Additionally, we would like the sensor to measure very precisely the equipment RPM (may be very slow rotating equipment) but at the same time detect very high frequency events around 10kHz (hitting the ‘hard-spots’). This is all becoming possible nowadays in the battery powered equipment but there are some limitations in the way. We can understand that there is practical limit for handling the sensor which does not have to be neither very big nor too small. There is obvious realisation that a lightweight sensor is easy to attach, and it is more likely that a smaller object can fit easier within the space provided.

Another point is not so obvious:

The smaller is the sensor and any of its electrical components, the more likely it can withstand very high g forces.

The tiny battery of small sensor dictates focus on low power and efficient RF link.

The lower is the frequency of operation, the greater is the RF link over the distance. Sub GHz ISM (Industrial, Scientific and Medical) bands are available to use for such applications E.g. 868MHz, 433MHz, 315MHz. These frequencies have wavelengths of 34.5cm, 70cm & 95cm and are close to the frequency range already used for some of the M2M communications and Smart Metering (‘450 MHz band’). Availability of any lower frequency would be nice but not offering advantage for the tiny sensors, as there is practical limit of the electrically shortened antenna efficiency.

As for vibration sensors, any use of the modem in licensed ‘450 MHz band’ is currently prohibitive, not just because of the equipment cost but in large part due to higher energy usage in data exchange. It is obvious that reporting of the vibration sensor every several minutes is much more frequent than Smart Meter (every several hours) and vibration sensor needs to remain very energy efficient. More importantly, the modems require high current pulses, which in case of small sensor battery must be buffered with massive capacitors, and these are not designed for high g forces experienced by the vibration sensors.

These details are subject to change and most likely in the future we would be able to use tailored 5G/6G solutions more suitable for the sensor needs. As for now, with current mobile networks deployments, there is much focus on bandwidth and GHz frequency bands suitable for the mobile entertainment. Human centred wireless does look for the highest possible data capacity and lowest bi-directional link latency in human-to-human or human-to-equipment interactions. Ideally vibration sensors are required to report the readings reliably but with the lowest energy per data used.

This indicates the main reason for the bespoke solutions like using the unlicensed ISM bands.

In ‘450MHz band’, the achievable distance is doubling that of 900MHz (frequency reduced by factor of two offers significant 6dB advantage). This allows maintaining a more reliable connection or reducing number of gateways by factor of 2 – 4 when aiming to achieve similar area coverage. The advantage of lower frequency band is also clearly visible when comparing very strong attenuation of higher frequency signals passing through thick structures like the concrete walls.

Graph showing the Decibel attenuation level across Frequency (from Phillips Laboratories “Measurement of RF Propagation into Concrete Structures…”)

On the small sensor’s side, one thing is the most likely to remain the same in the future: staying with low frequency bands for the sensor’s communications, in order to save the battery life and being able to maximise link budget with low transmit power of the sensor (or to minimise number of dedicated gateways to receive the sensors data in the area covered), whilst offering a robust communications link in an industrial environment.

By Thomasz Kawala

Self-Resonance in Sensor Design

There are many things to consider when designing any kind of sensing product, but the main considerations are generally to do with measuring a clean, precise, and isolated signal. Many techniques are employed to minimise noise, filtering, amplification, etc, to allow the different kinds of sensing element signals to be processed. However, if you are trying to measure vibration specifically, and you design an assembled structure that contains various levels of self-resonance, then some of your electronic efforts to measure a “clean” signal may be compromised.

Any object with a mass will have a primary excitation frequency that will cause a resonating response. This will occur close to the natural component frequencies with potentially multiple orders and with additional harmonics at higher frequencies. When you assemble various components together, each component will behave both individually, and also, as part of a system. The most dramatic example of self- resonance causing catastrophic failure is arguably the well documented Tacoma Narrows Bridge collapse, where excessive oscillating deflections were introduced by 40-mile-per-hour winds exciting the suspended structure. When designing a vibration sensor, the effect of self-resonance may be considered less dramatic. However, it can mean that misleading peaks at certain frequencies in the measured vibration data plots, will be visible to the end user.

Ideally, if you can perform a frequency sweep, and then measure the frequency responses where high amplitude measurements occur, then it can be determined if there are any self-resonances within the actual measurement range of the sensor. Simulations can be carried out using modal frequency analysis software. This software is becoming more readily accessible and is provided by many of the main 3D CAD design packages as either a free or upgrade add-on. Be careful with simplifications that you make to your CAD geometry and the material properties provided or other inputs for the simulation, including dynamic parameters that may influence the results.

There are many 1.0 to 2.5kHz sensors on the market that may be manipulated to push those modal frequencies above the measurement range of the sensor. Just like mounting an engine on spring mounts with addition or reduction of mass, make sure you utilise techniques to change the frequency response of your system. Increasing stiffness, adding relieving cut-outs, changing the bracing/fastening, adding isolating features, encapsulation, understanding the individual and overall centre of gravities, changing or adding materials, etc, are just some of the things that you can try to learn more about the system responses.

The market now has more affordable 5.0 to 10.0kHz plus, sensors being developed. As a result, it might be almost impossible to try to shift all individual modal frequencies above the now wider measurement range, through design. With that in mind, the more important considerations will need to be more closely tied to the application, and less about eliminating them completely. Understanding the frequencies generated by the different types of equipment you are planning to monitor, will help you to decide which design manipulations you need to adopt to avoid self-induced noise.

By Gordon Maguire

‘What’s for dinner?’ A Peek into Process Diagnostics

A clean velocity trend with all points below alert and alarm limits.

How do you read this? Well, that depends on who you are and what your position is in your organisation. If you are an area owner and this is one of your critical machines, then you’ll probably be really happy. Readings are being taken and everything looks good. The actual alert and alarm limits, enveloped spectrum alarms, high-frequency detection-that’s all been taken care of by the reliability department.

On the other hand, if you’re a technician and it’s your job to take these readings with a hand-held data collector, this trend can be described with one word – boring. While there is always value in taking a reading on a rotating asset and using your 5 senses to look for issues that might not be picked up with vibration, this only extends so far for machines that rarely have issues. For AHUs with sensors already mounted and cabled to the outside with BNC connections, you’re not even using the 5 senses. The most difficult part of the job is keeping your concentration and matching the asset in the route to the asset you are at. This has to compete with your attention span which might drift to what you’re going to have for dinner tonight or what you might watch on Netflix after. The technicians of today are hungry to learn and want to be as much a part of Industry 4.0 as anybody else. They will naturally gravitate to work that is interesting and challenges their grey matter. Taking readings on AHUs does not do this, especially if they are not involved in the analysis.

So, if you already have sensors on your AHUs, why not go the extra mile and stick a transmitter on the outside and send all that data up to the cloud. The reliability team are then not limited to one reading a month to figure out what’s actually going on. If you have your sensors on process equipment, then you can not only look at data related to machinery diagnostics, you can use export your overall trend into a data analytics package and overlay vibration against pressure, tank levels, valve states, flow rates. Maybe your pump is dead-heading against a manual valve that has no feed-back. Maybe it’s pumping against a partially closed control valve that is choking flow so the system can come up to temperature. This would then result in increased axial vibration, possibly over the alarm limits, that would then prompt an analyst to recommend a pump check, focusing on the coupling. In reality, this will be a temporary rise in vibration and by the time the technician gets there, will probably be passed. Maybe it’s being fed from a tank and the low-level interlocks aren’t working so it’s cavitating. Take it one step further and you can look at flow-induced vibration from where the system curve intersects the flow curve for different flow rates. A whole new world opens up.

To do this, you’ll need to think of systems as opposed to just individual assets. P&IDs, dampers, control valves, tank levels, flow rates-all the information from this instrumentation will form part of your assessment as opposed to just drive end and non-drive end vibration. This scenario is coming to an organisation near you, whether you like it or not. You just have to decide if you want to be a part of it.

Condition Monitoring and Security in the Cloud

‘The Cloud’ refers to software, services and data that are stored on the internet.  However, this data is not just stored in a random location; it will be located on a group of computers operated by a service provider. Typically it will be Microsoft, Amazon, or Google, but other providers are available.  These companies have vast arrays of computer servers located in Data Centres around the world, and your data will be kept in one or more of these data centres – and generally restricted to one region (for example, the US or EU) upon request.

Cloud technology can provide a huge amount of benefit when used and implemented correctly. It can greatly reduce the risk of data loss because of how cloud providers offer managed services to handle backups and failovers systems. 

For a condition monitoring platform using the cloud, it is essential that sensor data reaches the cloud using a secure method. Gateways, or base stations, that send vital condition monitoring data to the cloud do so using secure network protocols and encryption. Data sent using SSH, IPsec, PGP or TLS/SSL with encryptions such as AES, 3DES or RSA are industry standard. This makes the connection very robust helping to keep your data safe on the internet. Once the data reaches the cloud, it is good practice to further encrypt the data whilst being held in the database (encrypted at rest) using the same or similar methods highlighted above.

Using an internet connection and browser, an authorised user will have access to the data from anywhere in the world. User login credentials require a token (access to the encrypted data at rest) to safely and securely view valuable condition monitoring data.

Other technologies, such as machine learning, can also utilise the superior computing power that comes from the cloud, giving automated intelligence to any machine that is monitored.

Some scenarios will not be suitable for cloud computing, such as sites that do not have an internet connection, or require an offline system.  However, many typical systems can benefit from adopting a cloud-based condition monitoring platform and strategy.

David Procter, Systems Development Manager, Sensoteq

Key features for a Vibration Analyst

When monitoring the vibration data of a machine, an analyst will want to look at the Time Waveform (TWF) and the Frequency Spectrum or FFT (Fast Fourier Transform) output, as shown below.

Example of a TWF and Spectrum

These two example plots are very useful for determining obvious failure modes, such as, an imbalance with a high amplitude value of Overall (OA) Velocity in the 1 X RPM band or a slippage between two components causing amplitude modulations or beat frequencies visible in the TWF.

For more complex analysis, analysts require further tools to diagnose better the issues. Playing the TWF as an audio file is a good way of listening to the vibration to potentially hear knocking, pinging or high frequency whistling, denoting a problem to further diagnose, as if you were actually beside the machine.

TWF being played as an audio file

Auto-correlation and circular plots are two other useful techniques applicable to the TWF. Auto-correlation can be useful to remove noisy signals and determining the periodicity or repeating patterns in the vibration signal. The TWF below shows this before and after auto-correlation is applied.

Before auto-correlation
After auto-correlation

Finally, an additional step to visualise this better is to perform a circular plot. In the example below the three peaks indicate this application is a pump with three vanes. In this plot view using it can easily determine such failures as a broken fan blade, an imbalance on a vane, or rotor bar issues (if on a motor).

Example Circular Plot So as a vibration analyst make sure you use all the tools available to you in your software platform for ease of your analysis.

Alan McCall, CTO at Sensoteq

What to consider when choosing a wireless condition monitoring system

As wireless sensor are starting to become mainstream in the condition monitoring world, there are now many more choices to consider when selecting a system to fit your needs.  This tip aims to look at the key features that will impact your decision.  When purchasing your next wireless system, quiz your supplier using the information below.

Data Captured

Not all wireless sensors are created equal. Some will only capture an OA, or Overall Value, which is a single number indicative of the amount of vibration measured at the sensor.  OA is useful for detecting certain types of faults, but it will not assist in diagnosing the problem, or inform the user of how severe the potential issue is.

The good news is that there are a lot of sensors on the market that capture a full time waveform, TWF.  This waveform can be transformed into a spectrum revealing the energy associated with the underlying frequencies.  Using both a TWF and Spectrum is a reliable tool for diagnosing faults. 

One step better is a sensor that will capture both an OA value regularly, and a TWF periodically – ensuring both a timely and informative strategy.


This is the maximum frequency your sensor will capture, higher values allow for detecting faults on higher speed machines, or on very low speed machines (providing the sample window is sufficient).  The general rule of thumb is Fmax = 70 x Running Speed. As the vast majority of wireless sensors are tri-axial, you should ensure you are aware of the Fmax for each axis, as you may only need a high Fmax reading in one axis.

Installation and Maintenance

Ease of installation is a huge area for wireless sensors.  Does the sensor and gateway come pre-linked? How do you connect a new sensor to a gateway? Can you easily swap a sensor should one need replaced? Ask your supplier to show you a typical installation so you understand the time involved.

Wireless Technology & Range

There are numerous wireless technologies available, but make sure you are aware of the pros and cons of each before choosing your supplier.  WiFi and Bluetooth (both 2.4GHz) are very common and easy for most to understand, but their range can be limited when used in an industrial environment, furthermore, some factories control their entire plant using 2.4GHz and will simply not allow the addition of more transmitters on that frequency. 

Other technologies such as LoRa and ISM will utilise sub-1GHz bands and will not interfere with existing equipment. They also benefit from better propagation through certain materials due to their lower frequency, meaning better range for similar battery life – with less existing equipment on this band, it will trouble IT teams much less.


There’s no point having a wireless sensor if it’s powered via cables – thus most, if not all, wireless sensors are battery powered.  You will find two key options, non-replaceable and replaceable. 

Non-replaceable batteries will benefit when it comes to ingress protection and intrinsic safety. As the sensor will not be user serviceable, it greatly increases robustness.

Replaceable batteries increase the useable life of the sensor – but be aware of misleading claims from manufacturers quoted battery life, as it can typically only be achieved when the sensor is in a low power state and limited use mode – ask your sensor supplier for the true battery life for your given application. The more times you have to replace a battery, the greater cost to you, mostly for the time it takes to replace (on average $100 per sensor), so higher battery life is better, even for replaceable battery.

David Proctor, Global Systems Data Specialist at Sensoteq