Spectrum Analysis of System Power Supply Noise and Prediction of Radiated Emission Trend

1. FFT spectrum analysis of power plane noise

The FFT transform of a signal is mainly used to obtain the spectrum of the signal in order to analyze the frequency domain characteristics of the signal. Analyzing the signal in the frequency domain is conducive to observing more and more accurate essential characteristics of the signal, such as the harmonics of the signal, the phase change of the signal, etc.

Figure 1 below shows the relationship between the time domain and the frequency domain:

Figure a) shows the three-dimensional relationship between time, frequency and amplitude.

When the radiation emission energy of a single board exceeds a certain range, it usually cannot meet some international radiation emission standards, such as CALSS A and CLASS B. The radiation source mainly comes from the high-order harmonic components of the high-speed signal on the single board, especially the high-speed clock signal. Usually, when debugging the radiation emission capability of a single board, it is necessary to use many frequency domain instruments, such as EMSCAN , spectrum analyzer, etc. If you need to get more accurate quantitative indicators of the radiation emission capability of the single board, you need to conduct test measurements in a microwave darkroom. Sometimes, before the production of a single board, in order to roughly estimate the radiation emission intensity of the single board, some frequency domain simulation software such as Ansoft HFSS, Siwave and other analysis software are also used to model the entire system or a part of the system, so as to simulate the radiation emission of the system. When using simulation software, it is usually difficult to obtain the accurate radiation emission source.

In fact, while various frequency energies on the board radiate energy to the external space, the power plane or ground plane on the board will also be affected by these signals in some way, so the spectrum of the power noise should also include or reflect the spectrum of various high-frequency signals and their intensity. So in the debugging stage of the board, can we roughly and qualitatively analyze the possible radiation emission energy of various frequency points by analyzing the spectrum of the power noise? If this method is feasible, it will greatly facilitate our debugging - we can use the long storage + extremely fast signal processing speed + powerful FFT analysis function of today's high-end digital oscilloscopes to debug.

The figure below is the spectrum of a power supply noise. In the time domain, we can only see the random oscillation noise signal, and we cannot see any information about the clock signal. However, by performing FFT transformation analysis on its time domain waveform, we can clearly see a clock signal ( the clock signal shown in Figure 2 below is about 31MHZ ) and its harmonic components, and we can see the approximate intensity of its spectrum energy (qualitative value). Generally speaking, the stronger the harmonic energy, the greater the possibility of its external radiation energy. If we conduct some comparative accumulation analysis before and after the revision of the single board and add long-term data accumulation analysis, we may be able to get a more reasonable trend prediction. This is just a guess and an idea. If we can get some more reasonable trends, it should be very convenient for us to predict the radiation emission of the single board. At the same time, these data may also be used as a radiation emission source in electromagnetic field simulation software to further simulate and predict the near-field and far-field radiation emission intensity of the single board and the system, and finally achieve the prediction before the single board is made, which can save costs and shorten the R&D cycle.

Figure 2 31 MHz clock signal spectrum and its harmonic components observed in the power supply noise spectrum

Figure 3 22 MHz clock signal spectrum and its harmonic components observed in the power supply noise spectrum [page]

If you want to get a lower frequency spectrum and observe with higher resolution, you need an oscilloscope that can analyze very long data and sample at a higher sampling rate. In addition, when the oscilloscope performs FFT analysis in a long storage and high sampling state, the speed of the oscilloscope will be greatly affected. The current Lecroy 's new WP 7 ZI and WM 8 ZI series oscilloscopes rely on the top CPU (quad-core) hardware system and 64- bit Vista operating system, plus Lecroy's unique XstreamII signal processing technology, making it very fast in long storage data processing. It only takes a few seconds to perform FFT operations on 100MS or even longer data .

2. Features of LeCroy Oscilloscope in FFT Analysis

1. The longest analyzable storage depth

In order to observe the signal spectrum in the lower frequency band and obtain a higher spectrum resolution, it is necessary to collect enough data for FFT analysis. So does it mean that the deeper the acquisition memory depth, the more data can be used for FFT analysis? In fact, this is not the case. Many oscilloscopes have certain restrictions on the number of points that can be used for FFT operations. That is to say, although a lot of data can be collected, only a part of it can be truly used for FFT operations. The maximum data storage depth that can be truly used for operations is usually called the analyzable memory depth. Of course, if you need to use third-party software to process data and only need to use an oscilloscope to capture, a larger acquisition memory can provide more acquisition data, that is, the acquisition memory depth is more meaningful at this time. The maximum acquisition memory depth of LeCroy oscilloscopes is the maximum analyzable memory depth, which can reach up to 512MS of analyzable memory depth.

2. Can provide higher sampling rate

Lecroy 's new WP 7 ZI series oscilloscopes can provide the highest sampling rate in the same category, up to 40GS/S ; the WM 8 ZI series oscilloscopes can provide a sampling rate of up to 80GS/S .

3. Extremely fast processing speed

It usually takes only a few seconds to perform FFT operations on data of 100MS or even deeper; it usually takes only about 20 seconds to perform eye diagram testing on 100MS data .

4. Very convenient advanced spectrum analyzer option

Lecroy 's latest oscilloscope WP760 ZI has a new spectrum analyzer option. The operation interface of this option is very similar to that of R&S spectrum analyzers. It can easily set the center frequency , span range, etc., and can display the frequency value of the peak point and the corresponding amplitude value in real time. The operation is very convenient and simple. The interface is shown in Figure 4 below :

Figure 4 Lecroy WP760ZI’s newly added spectrum analyzer option

3. Radiated emission test of a system and its corresponding power supply noise spectrum test

Figure 5 below is a diagram of the radiation emission results of a certain system. We can see that the energy in the low frequency band ( within the range of 30MHZ-70MHZ ) is a little bit excessive:

Figure 5 Radiation emission diagram of a system

The power supply noise spectrum measured on the corresponding power plane is:

Figure 6 Power supply noise spectrum of a system

It can be clearly seen from the FFT spectrum analysis of the power supply that there is indeed an obvious peak in the low frequency band around 30MHZ , which does correspond to the actual radiation emission test results.

IV. Conclusion