Abstract--- This paper illustrates the use of the Modulo Time Plot to facilitate diagnosis of data acquisition systems and components. While conventional techniques, involving spectral analysis and histograms, provide certain useful and necessary measures
sample number, k, for a set of NS samples, x(kTS )
as points and are not connected by lines. The reordered set, xR, can have each value connected with tk= k=FS for k= 0; 1;:::; NS? 1: (1) lines; however, the x and~k points should not be t connected when they are plotted. This gure clearly Equation 1 was used to plot the time axis for the set shows that the sampled signal is a sinewave with the of sample values shown in Fig. 1. Now consider that correct period and its peak-to-peak operation. when the test frequency is set at some fraction of the While we have just derived the relationship for the sample frequency, the samples are taken at di erent reordered sample set, it is instructive to note that the points in successive periods of the test signal. Let the transform of the reordered set has also reordered the signal test frequency, FT, be set on the mth bin of harmonics of the test signal in the following fashion. the FFT basis frequency set for the samples as given Equation 7 gives the DFT for the reordered sample in Eqn. 2. The number of samples, NS, is normally set NS?1 a power of 2 for e cient FFT implementation. The m is generally selected to be an odd number for this (7) xR (`)e?j 2`n=NS: XR (n)= case`=0 F (2) When NS and m are relatively prime, Eqn. 7 may be FT= m NS for m odd: S rewritten in terms of a summation index, k, where k The period of the test signal is then given by and` are related by NS`= mk mod NS: (8) T= F1= mF: (3) T S into Eqn. yields When the sample time, tk, is represented modulo the Substitution of Eqn. 8transforms7for the the following identity between the sample sets signal period, T, then the samples may be reordered to display their position within a single period of the NS?1 test signal.
The reordering for plot purposes may be XR (n)= xR (mk mod NS )e?j 2 (mk modNS )n=NS~k as the modulo T accomplished as follows. De ne t k=0 values of tk NS?1~k= tk mod T: t (4) x(k)e?j 2 mkn=NS k=0 A plot of x(kTS ) versus~k will be referred to as X (n) t X(nm): (9) R the Modulo Time Plot. It is possible to show that if NS is a power of 2, and m is odd, i.e., m and NS are signal~ relatively prime, then the values of tk are uniformly The harmonics of the test for the are reordered sequentially in the transform reordered sample spaced over one full period of the test signal. Select- set. ing m and NS relatively prime may also reduce the The next presents several examples of how experimental uncertainty, as suggested in 5] for the the modulo sectionplots provide useful time-domain time diagnosis of ADC integral and di erential nonlinear- interpretations for sampled data sets. ities. In this case, however, reordering is the issue so that the modulo time plot may also be created III. Modulo Time Plot Examples by reordering the samples, x(kTS ), to relate index The Modulo Time Plot is presented as a complimennumbers of the original set to indices of a reordered tary tool to assist in the diagnosis of the behavior of set a data acquisition system. It is really only one more kR= mk mod NS mFS tk mod mFS T: (5) tool that can be used to determine whether a system is working correctly or not. Examples are given in The kR gives the index of the kth sample, x(kTS ), this section to illustrate the usefulness of this data in the reordered set. The reordered set variables are presentation method. then given by A. Residual Error Measured data for a sampled sinewave plus noise is xR (kR )= x(k) x(kTS ) shown in Fig. 1. The noise is added to dither, or k: k (6) randomize, what would otherwise be harmonically tR= N F S T mFS dependent quantization error 6]. Because a sinewave Figure 3 shows the data of Fig. 1 plotted either as x signal is used, it is possible to estimate both the funversus~k or xR versus tR . The samples are plotted damental component and the average DC component t
X
X X
Correction to “Comments on ‘The Modulo Time Plot: A Useful Data Acquisition Diagnostic Tool’ ” Zsolt P? pay a Fig. 1 Mathcad simulation: spurious ...
Modulo Numbers 2000 Gold Coins 2021 Relative ...Plot 1142 Smith Numbers 1188 Gleaming the Cubes ...1676 What time is it? 1684 Dynamic Declaration ...
useful stuff gnuradio companion technical stuff ...The QT GUI output offers multiple views: Time (...(idx == len) idx = 0; modulo len idx = (...
useful stuff gnuradio companion technical stuff ...The QT GUI output offers multiple views: Time (...(idx == len) idx = 0; modulo len idx = (...
热门文档