Insights / IoT
Industrial IoT: when off-the-shelf hardware is not enough
Industrial IoT is often sold with sensors and cloud. In practice, however, it begins unspectacularly: with a measurement task that cannot be solved cleanly with standard products. Whether off-the-shelf hardware is enough or individual electronics are needed is decided at the measuring point – not by product strategy.
Where off-the-shelf hardware is enough
For pressure, temperature, flow and current there is a wide, well-developed range of sensors and transducers. If the measuring method is standard, the measuring range fits and the installation situation is uncritical, individual development is not a benefit – it is an additional construction site.
- Measuring method and measuring range are standard.
- The sensor signal can be digitised anyway.
- The installation situation is uncritical (temperature, vibration, contamination).
- A standard connection is sufficient (4–20 mA, IO-Link, EtherCAT).
- Comparability of the measured values over several years matters more than laboratory accuracy.
When individual electronics make sense
Three cases justify building it yourself: when the measuring method is special, when several values have to be captured at one location together, or when communication is the actual problem. An adapter with signal conditioning, galvanic separation and a bus connection saves an expensive special-purpose solution in such cases.
- Special measuring method for which no suitable sensor exists.
- Several measured values at one point that are evaluated together.
- Galvanic separation, signal quality monitoring or self-calibration.
- Connection to existing fieldbuses without an additional gateway.
- Own circuit boards as prototype and series, when the quantities justify it.
The chain matters, not the component
An excellent sensor is of no use if its values arrive nowhere or are meaningless in the visualisation. What matters is the closed chain: measuring point, signal conditioning, transmission, storage, evaluation – and ultimately a decision that operations can act on.
In many cases the biggest lever is not better sensors but the evaluation. Anyone who records the same data consistently over years sees trends that are invisible with individual measurements.
Introducing condition monitoring
The best way in is via a machine with a high failure risk and clear operating hours. It is important to capture the operating state from the beginning: only then can maintenance windows, downtime and load curves be evaluated properly later – otherwise the evaluation remains a collection of curves without meaning.
- Choose the machine: rotating, high failure costs, clear operating hours.
- Define the measured value: vibration, temperature, current or runtime.
- Capture the measuring point and operating state, not just the value.
- Derive thresholds from normal operation, not from the data sheet.
- Evaluation in SCADA or database, with notification to maintenance.
- Check the effectiveness after the first maintenance – do not just collect data.
What IoT means at MiRADA
We entered the IoT field in 2017 and have implemented several projects there. For us this means: we know not only the standard products but also the cases in which standard products do not get you further – and how to set up an electronics development that nevertheless remains manufacturable.
Is condition monitoring worthwhile for every machine?
No. It is worthwhile where downtime is expensive or where wear should become plannable. For machines with low failure costs the effort for measurement and evaluation is often higher than the benefit.
How long until condition monitoring produces data?
Measurement starts immediately, meaningfulness grows with the history. Trend comparisons need months; a reliable service life estimate rather one to two years.
Can existing sensors be reused?
Yes, provided they deliver a signal that can be evaluated. It is often the more economical variant to use existing sensor technology and modernise evaluation and communication.