AMP STANDARD LABORATORY REPORT FORMAT
From Center for Friction Stir Processing Guide
APPENDIX J – AMP STANDARD LABORATORY REPORT FORMAT
Laboratory Report
Noise Detection of MTS ISTIR-10 FSW Machine Work Orders: FSW06010, FSW05128, FSW05088 and FSW04005
Prepared For: CFSP / IUCRC
Prepared by: Enkhasaikhan Boldsaikhan
Advisors: William J. Arbegast Casey D. Allen Edward M. Corwin Antonette M. Logar
Approved by:
William J. Arbegast Director, NSF Center for Friction Stir Processing (CFSP) Director, SDSMT Advanced Materials Processing and Joining Center (AMP)
November 4, 2006 South Dakota School of Mines and Technology
Abstract
The purpose of the experiment was to investigate the machine dependent noise in the feedback signals. In order to do that, we collected noise signals while the FSW machine was running without actual welding (empty run, the pin tool was in the air, no material was involved), under 28 variation of system parameters values. We created the DFT’s of noise signal and compared them with the DFT’s of actual welds. It appears that the machine-dependent noise is may be embedded in the feedback signals. A machine-dependent noise is a collection of frequency values that behaves to be independent of the system parameter values and the variation of material and pin tool.
Introduction
The objective was to collect the feedback signals while the machine was performing an empty run (the pin tool was in the air) and compare them with the feedback signals of actual welds. The FSW machine is MTS ISTIR-10. We are investigating the effectiveness of a various methods to removing machine-dependent noise from the distorted feedback signals. We can handle the high frequency noise by using a low-pass filter. The only problem is how to determine and remove the low frequency noise (machine-dependent), whose sources can be any of the moving mechanisms of the FSW machine. We are assuming that the feedback signals with the frequencies of the spindle or less than the spindle frequencies may have important information regarding the weld quality. Consequently, the low frequency noise (machine-dependent noise) may lead us into a wrong conclusion on analyzing those feedback signals. Thus, it is extremely important to determining and removing low frequency (machine-dependent) noise from the distorted feedback signals.
Approach / Procedure
Task1 – Noise Data Collection. Collect the noise signals while the FSW machine is performing an empty run, under various system parameter values. (The traversing speed variation was 2, 4, 6 and 10 [ipm], and the spindle speed variation was 200, 250, 300, 350, 425, 500 and 600 [rpm])
Task2 – Noise Detection (Exhaustive Search). The objective of this task is to determine the machine-dependent noise frequency. In order to do that, create the frequency spectral plots of all the feedback signals using a discrete Fourier transform. It is worth full to create the frequency spectral plots from various welds with different system parameters, material and pin tool. Compare all these plots and search for machine-dependent noise frequencies among the dominant frequency peaks in the frequency spectral plot. If a collection of frequency peaks that behaves to be independent of the system parameter values and the variation of material and pin tool design, then it means a machine-dependent noise is present in the feedback signals.
Results
In order to compare the noise signals with the actual weld signals, we used data from FSW04005, FSW05088 and FSW05128 (The materials, which were involved in these experiments, are 0.25” Al 7075-T73 and 0.25” Al2024-T3). We created several frequency spectral plots. By examining them, we discovered that a same dominant peak is located at ~14Hz in the frequency spectrum of all the feedback signals (especially, X feedback force) independent of traversing speed and material variations. The corresponding spindle speed was 600 rpm. This dominant peak represents the machine-dependent noise frequency. (Figures 1 & 2).
Discussion
According to the experiment, the machine-dependent noise starts occurring from 600 rpm, it means that the machine-dependent noise may occur at the higher spindle speed (greater than 600 rpm). We need to verify it!
For the sake of determining the machine-dependent noise frequencies, we applied an exhaustive search method (Approach\Task2) which requires a large amount of data and time. The advantage is, once the noise frequencies were determined, it will be easy to remove these noise frequencies from the feedback signals using a forward and inverse Fourier transform. Additional noise filtering techniques (including an adaptive filter) are currently being studied.
Conclusion
MTS ISTIR-10 FSW machine has some machine-dependent noise at the spindle speed of 600 rpm or higher. The exhaustive search method (Approach\Task2) may help us to determine the machine-dependent noise frequencies of any FSW machine.
