Since many automobile NVH issues involve the analysis of order spectra it is highly important that the methods used should be accurate and consistent. A review of the literature shows little discussion of several key issues which could cause problems and possibly invalidate test results. Also, experience indicates that different methods may yield different results unless these key issues are dealt with.
This paper compares four different approaches to order tracking as applied to typical engine sweep data: order analysis of conventional fixed frequency waterfall plots, synchronous sampling order tracking, computed digital order tracking, and Kalman filtering. It is shown that unless the analysis parameters are carefully chosen the results of the analysis may be in error. The major issues for the conventional fixed sampling rate waterfall method arise because sweeping the orders causes a frequency smearing effect. This paper compares some methods of dealing with the smearing problem and suggests an optimum strategy. The optimized waterfall procedure is then compared to the other three order tracking methods. Finally it is shown that synchronous clock order tracking methods can exhibit an order smearing problem unless precautions are taken.
