March 05, 2021 · 5 min read
In electronics, noise can be any unwanted form of energy that disturbs a process or communication. In Vibration-Based Smart Maintenance, noise can impact the quality of the analysis. That's why a good electronic system is designed when also the power of this unwanted element is taken into account.
The Problem of Noise
Suppose you need to attend an important business lunch in a restaurant. The outcome is to close an important deal with foreign customers. Clear conversation is fundamental to be successful. However, in the restaurant room, close to your table, a group of friends is celebrating. During the evening, they increase the volume of their voices and laugh in a way that degrades your understanding of the subject of your meeting: price values. 9.8 million or 5.8 million? What is the actual value? If you assume the wrong value, you can be in dire straits.
The metaphor above illustrates a very common problem in electronics data acquisition and measurements: the noise. Noise is everywhere, and we need to deal with it. A good electronic system is designed when also the power of this unwanted element is taken into account. When we think about the industrial environment, the noise becomes more critical. Factories, with their machines, some of them working with high power, create a very noisy context that affects the proper operation of monitoring sensors very negatively.
What Is Electrical Noise, What Causes It, and How Can It Be Categorized?
In electronics, noise can be any unwanted form of energy that disturbs a process or communication. For you in the restaurant, the friend's voice from a close table represents a noise. Noise can present itself in different forms, such as an unwanted electric signal conducted by wires or an electromagnetic disturbance propagating through the air. When the noise is presented as an electric signal in a circuit, we call it Conducted Noise, and when it spreads in the air, it is called Radiated Noise. Both can be a source of the problem, and the origin can be natural (a lightning strike, for instance) or artificial. When it is artificial, it can be intentional or not. When we make phone calls, our voice travels from our phone antenna through the air as electromagnetic waves. In the past, scientists discovered we could transmit information like voice by causing periodic variations in electric and magnetic fields. So, a phone call can be a noise to a circuit sensitive to it. In a big factory, with several motors, robots, and computers, the operation of these machines transforms the electricity in unwanted conducted and radiated noise. Whether the noise is a problem or not to a system depends on the frequency sensitivity of the system itself. This is similar to human hearing. A human can hear sounds in frequencies from 20Hz to 20000Hz. So if the noise is in this frequency range, it can annoy a person. Dogs and cats have more frequency sensitivity. Bats can listen to an even larger range. What dogs, cats, and bats perceive as noise, for us, is not a problem. In electrical noise too, frequency matters.
Noise can come not only from the outside but also from the inside of a system. In that case, it’s called internal noise.
In conclusion, noise is an electric or magnetic disturbance arising from nature or human devices, intentional or not. The degree to which noise affects a system may vary, but to operate correctly, we need to isolate our circuit from it.
Noise and Sensors for Smart Maintenance
Why is the problem of noise so relevant to the creation of our sensor nodes? For many different reasons:
- Our sensor kits have to deal with all the different types of noises we discussed: the external ones, coming from the harsh environments of factories and the internal ones, a strong, friendly fire. In fact, the sensor node power supply operation, the processor while it computes, the converters, and even single resistors can all cause noise.
- The type of analysis we perform. Our analysis mainly relies on vibration. By interpreting the behavior of machines’ performance, we can tell if something is not working, why, and when it is going to break down. To perform this analysis, our sensor nodes have a very important component: the accelerometer. It converts the sensed acceleration to an electric signal to be processed in the next stages. In the following paragraph, we will explain the importance of the accelerometer and how noise can impact it badly.
The Importance of the Accelerometer in Vibration-Based Monitoring
The accelerometer is a fundamental component in vibration analysis since it allows the measurement of vibration. In the electronics world, there are two main languages: analog and digital. The analog is the language of nature because analog signals exist everywhere and can assume infinite values. For instance, from 0 volts to 1 volt, we can have an infinite amount of voltage values. In the digital world, there are only two possible values, 0 and 1. In spite of this simplicity of digital, there are very powerful characteristics that allow us to create amazing existing computers, cell phones, satellites, and a giant list of excellent devices.
When our sensor nodes monitor your machine, the accelerometers convert the acceleration of vibration into analog electric signals. To become actionable insights, the signals have to be analyzed by a processor, that only “speaks” the digital language. So, we need to include a translator in this conversation between the accelerometer and processor that converts the analog signal to digital signal so that the processor can do the calculations. This translator is a very important component, and it is called ADC, Analog to Digital Converter. An important requirement of the ADC is its resolution, expressed in the number of bits. Basically, the higher the number of bits, the better the resolution of the measurement system. But we know there are no free meals. An ADC with higher resolution implies, of course, that the vibrations produced by the machines can be caught better, but also that the noise plays a more important role. And poor accuracy can lead to false-positive and/or false-negative alerts.
How To Deal With Electrical Noise When Building a Sensor Node
In hardware development, the noise should be taken seriously, and specific measures to minimize its impact should be applied in the designs. The AiSight team uses and suggests different techniques to attenuate and block the noise, such as:
● Choosing good quality components with low intrinsic noise figures.
● Designing proper filters according to the context and circuit’s needs.
● Applying good practice of placement and routing components in the PCB, Printed Circuit Board where the circuit components are soldered and connected.
● Making computational simulations and mathematical analysis to achieve the required performance to reduce the loop designs.
● Considering and evaluating metallic shielding options.
● Testing and testing, from bench level at our lab to EMC/EMI compliance test in accredited labs.
We keep in constant battle to use the best tools, techniques, and processes to deliver high accuracy measurements, reliable and robust hardware that can contribute to an unlimited uptime. If you want to get started with Aion, schedule a demo now.
© 2021 AiSight GmbH. All rights reserved
Phone: +49 (0) 30 40363399