Preventive maintenance begins when recording the data

Cracks and breaks do not come as a surprise: preventive maintenance can avoid failure and stoppages. In order to achieve this, the most important functional elements of machines must be monitored permanently. Roller bearings are central components in rotary machines and therefore important elements for recording the status of the machine. Sound emission testing based on high-speed data recording serves to recognise damages to roller bearings early.

Excellent processing is aimed at meeting the market requirements for high-quality and cost-efficient products. Against the backdrop of just-in-time production, modern-day manufacturers are doubtless subject to high pressure. The key to the manufacturers' success is to ensure the best performance of their machines at all times because heavy damages often result from defects.

The aim of the preventive maintenance of roller bearings is to guarantee their stable and reliable function in the long-term. Surface defects are the most frequent roller bearing defects. As a result of fretting fatigue, breaks occur in the surface material of the tracks or roller elements. The vibration signals from the roller bearings are very weak in the early phase of the formation of fatigue damages. They are mixed with the vibration signals from other machine components and subject to background noise, making the identification of damages difficult. Luckily, the measurement of acoustic emissions can identify the growth of cracks caused by fatigue damages. This provides a significant advantage for the early prediction and diagnosis of roller bearing damages.

Complete capture of waves for acoustic emission signals

The analysis of acoustic emission is an important method for dynamic monitoring of material or structure statuses. This method has developed from its initial test size recording step-by-step to today's complete wave shape analysis. A system for the analysis of acoustic emission consists of an acoustic sensor, a pre-amplifier and high-speed data capture. Acoustic emission signals have wide spectrums and demonstrate frequencies from a few kHz up to the 10 MHz range. Therefore, a digitizer with a sampling rate of 20 MS/s or higher is required for data capture. Moreover, acoustic emission signals are very weak. Original acoustic emission signals are mostly at µvolt level and also only achieve a few mvolts after pre-amplification. Additionally, they have a large wave shape amplitude range. This explains the high demands pertaining to scanning precision and the dynamic range of the measurement instruments.

Realising the recording of acoustic emission signals

Adlink equips the high-speed digitizer PCI-9846 with a driver for LabVIEW. Following installation of the DAQPilot driver, the tool collection DAQPilot is available in the LabVIEW database. This tool provides control functions for data recording. The control functions and access methods are in accordance with the DAQmax driver software from National Instruments and can therefore be used comfortably. For example, the four elements in the following illustration (PLT Create Virtual Channel, PLT Timing, PLT Read and PLT Clear Task) form a full function for continuous data recording. On top of this, the DAQPilot tool collection also provides LabVIEW sample programs with functions such as inputs and outputs of analogue and digital signals. Based on these examples, developers can acquire the required competences fast.

Continuous data recording with the high-speed digitizer PCI-9846Continuous data recording with the high-speed digitizer PCI-9846

The signal recording module which is the basis for the test system for testing acoustic emission described in this document is based on the scanning procedure as illustrated in Figure 1. After the input channel, the amplification of the pre-amplifier, the filter and further parameters have been determined, the user can start continuous data recording. Following reduction of the original signal in accordance with the magnification factor and it running through the digital filter, signal noise and superfluous frequency bands are removed. The filtered continuous signal can be saved in binary format or as a text. Figure 2 shows the acoustic emission signal for a roller bearing recorded during operation. The sampling rate is 2 MS/s, the filter is a fourth order Butterworth low-pass filter and the limit frequency is 50 kHz.

Continuously recorded acoustic emission signalContinuously recorded acoustic emission signal

Parameter analysis of an acoustic emission signal

After filtering, the continuously recorded signal is transferred to am amplitude detector. Dependent on the defined amplitude threshold and the period determined, the wave shape data for the suddenly beginning acoustic emission is intercepted from the continuously recorded signal. Following this, characteristic parameters such as ring down count, amplitude, frequency, energy, rise time and duration are calculated and saved in an acoustic emission parameter table so that they can be observed directly and are available for further-reaching system analysis. The recorded data for the wave-shape acoustic emission can be saved as text. Here, the storage of this wave shape data is independent of the continuous data storage.

The PCI-9846 card offers abundant trigger scanning modes for rising or falling edges, window, reference and other analogue trigger modes. Wave-shape acoustic emission signals can be scanned directly by the hardware. This reduces the processor load and saves resources which are therefore available for the software extraction algorithms. Using test assembly it has been found out that acoustic emission can easily experience interference via factors such as electromagnetic fields. Therefore, the signals must be filtered before the wave shape is recorded. However, the digitizer PCI-9846 does not have programmable filters. Suchlike could be integrated into the front end of the PCI-9846 in future system extensions, enabling full exploitation of the trigger options provided by the hardware.

Wave shape of the suddenly beginning acoustic emission signal of a roller bearingWave shape of the suddenly beginning acoustic emission signal of a roller bearing

Checking the performance of the test system

In order to determine the performance of the test system for checking acoustic emission, a comparison test was conducted. For this purpose, a rotation machine test system QPZZ-II was used to simulate the acoustic emission and vibration of a roller bearing with faulty roller elements. The test object was a type NU205 roller bearing with a pitch diameter of 39 mm, 12 roller bearings with a roller bearing diameter of 7.5 mm and a 0° contact angle. Small cracks were made on the roller element surfaces with straight line cuts in order to simulate the damages which are incurred by "peeling off" the surface as a result of fatigue. The rotation speed was 570 rpm with a specific rotation frequency of 9.5 Hz. According to the calculation, the roller elements' pass-through frequency was 18.3 Hz.

Comparison of the acoustic emission and vibration signals of a damaged roller bearingComparison of the acoustic emission and vibration signals of a damaged roller bearing

An acceleration sensor was accordingly attached to the roller bearing holder in the vertical axis position. In addition to this sensor, the vibration text system consisted of an oscillation measurement device for recording dynamic signals. The vibration data was imported to a Matlab environment for analysis. In case of unchanged rotation speed, the acceleration sensor was exchanged with an acoustic sensor. The test system for acoustic emission described here was used for data recording. Data analysis was made via the Matlab environment. Figure 4 shows the results of the comparison between the acoustic emission signal and the oscillation acceleration signal. As shown by the illustration, the signal-noise ratio of the acoustic emission caused by the roller element's surface damages is considerably higher than the signal-noise ratio for the oscillation acceleration. Figure 5 shows the course of an envelope spectrum of both signals following narrow band filtering. As is apparent from the illustration, the acoustic emission signal's envelope spectrum has a simpler line structure, enabling easier determination of the characteristic frequency of the damage.

Comparison of the acoustic emission signal's envelope spectrums and the oscillation acceleration signal from a damaged roller bearingComparison of the acoustic emission signal's envelope spectrums and the oscillation acceleration signal from a damaged roller bearing

The test system for checking acoustic emission combines the outstanding price-performance ratio of the high-speed digitizer PCI-9846 and its easy handling using the software. The result is low costs, a comfortable development environment and great expansion potential. The test system was used to evaluate the acoustic emission signal generated by a roller bearing on a roller element in operation. The results of this test were compared with the results of a conventional vibration test. The comparison shows that the acoustic emission has a higher signal-noise ratio. This makes the huge potential of acoustic emission testing for the early recognition and diagnosis of roller bearing damages clear.