Cleverly Using Oscilloscope Frequency Domain Analysis Method to Analyze Power Supply Noise

power supplyNoise is a type of electromagnetic interference, with a frequency spectrum of approximately 10kHz~30MHz and up to 150MHz for conducted noise. Power supply noiseSound, especially transient noise interference, has a fast rising speed, short duration, high voltage amplitude, and strong randomness, which can cause serious interference to microcomputers and digital circuits.

Application of Oscilloscope Frequency Domain Analysis in Power Debugging

This article discusses how tomany yearscomeElectricity that attracts attentionSource noise measurementaskThe question is based on practical experience summary, supported by actual test cases, and combined with simulation analysis.
In the analysis of power supply noise, a classic method is to use an oscilloscope to observe the waveform of power supply noise and measure its amplitude, based on which the source of power supply noise can be determined. However, as the voltage of digital devices gradually decreases and the current gradually increases, the difficulty of power supply design increases, and more effective testing methods are needed to evaluate power supply noise. This article usesA case study of analyzing power supply noise using frequency domain method. When the fault cannot be located by observing the time-domain waveform, FFT (Fast Fourier Transform) method is used for time-frequency conversion to convert the time-domain power supply noise waveform to the frequency domain for analysis. When debugging circuits, examining signal characteristics from both time and frequency domains can effectively accelerate the debugging process.

During the single board debugging process, it was found that the power noise of a network reached 80mv, which exceeded the device requirements. In order to ensure stable operation of the device, it is necessary to reduce the power noise.
Before debugging the fault, review the principle of power supply noise suppression. As shown in the figure below, different frequency bands in the power distribution network are suppressed by different components to suppress noise. Decoupling components include power adjustment modules (VRMs), decoupling capacitors, PCB power ground planes, device packaging, and chips. VRM includes a power chip and peripheral output capacitors, which act approximately in the DC to low frequency range (around 100K). Its equivalent model is a two element model consisting of a resistor and an inductor. Decoupling capacitors should be used in conjunction with capacitors of multiple orders of magnitude to fully cover the mid frequency range (around 10K to 100M). Due to the presence of wiring inductors and packaging inductors, even if a large number of decoupling capacitors are stacked, they are difficult to function at higher frequencies. The PCB power supply ground plane forms a flat capacitor, which also has a decoupling effect, acting approximately at tens of megabytes. Chip packaging and chips are responsible for high frequency bands (above 100M). Currently, devices generally add decoupling capacitors to the packaging, and the decoupling range on the PCB can be reduced to tens or even a few megabytes. Therefore, under the condition of constant current load, we only need to determine which frequency band the voltage noise appears in, and then optimize the decoupling components corresponding to this frequency band. When two decoupling elements are in adjacent frequency bands, they will work together, so when analyzing the critical point of decoupling elements, the decoupling elements in adjacent frequency bands should also be considered simultaneously.

Based on traditional power debugging experience, some decoupling capacitors were first added to the network to increase the impedance margin of the power network, ensuring that the impedance of the power network in the mid frequency band can meet the requirements of this application scenario. The ripple was only reduced by a few mV, and the improvement was minimal. There are several possibilities for producing this result: 1. The noise is at a low frequency and not within the range where these decoupling capacitors work; 2. Increasing capacitance affects the loop characteristics of the power regulator VRM, and the impedance reduction caused by capacitance offsets the deterioration of VRM. With this question in mind, we are considering using the frequency domain analysis function of an oscilloscope to examine the spectral characteristics of power supply noise and locate the root cause of the problem.

The frequency domain analysis function of an oscilloscope is achieved through Fourier transform, which essentially represents any time-domain sequence as an infinite superposition of sine wave signals of different frequencies. We analyze the frequency, amplitude, and phase information of these sine waves by switching the time-domain signal to the frequency-domain analysis method. The sequence sampled by a digital oscilloscope is a discrete sequence, so the Fast Fourier Transform (FFT) is commonly used in analysis. The FFT algorithm is an optimization of the Discrete Fourier Transform (DFT) algorithm, which reduces the computational complexity by several orders of magnitude, and the more points need to be computed, the greater the savings in computational complexity.
There are several key points to note when performing FFT transformation on the noise waveform captured by the oscilloscope.
1. According to the Nyquist sampling law, the spectral spread (Span) after transformation corresponds to half of the sampling rate of the original signal. If the sampling rate of the original signal is 1GS/s, the spectral spread after FFT is mostly 500MHz;
2. The frequency resolution bandwidth after transformation corresponds to the reciprocal of the sampling time. If the sampling time is 10mS, the corresponding frequency resolution is 100Hz;
3. Spectrum leakage refers to the mutual interference between various spectral lines in the signal spectrum, where lower energy spectral lines are easily overwhelmed by the leakage of adjacent high-energy spectral lines. To avoid spectrum leakage, it is advisable to synchronize the acquisition rate with the signal frequency as much as possible, extend the signal acquisition time, and use appropriate window functions.
power supplyWhen measuring noise, a high sampling rate is not required, so a long time base can be set. This also means that the collected signal time can be long enough to cover the entire effective signal time span, and there is no need to add a window function. Adjusting the above settings can obtain a more accurate FFT transformation curve, and then use the zoom function to view the frequency points of interest. The main energy of the power supply noise in the figure is concentrated around 11.3KHz and resonates at this frequency as the fundamental frequency. Based on this, it can be inferred that the impedance of this PDN network at 11.3KHz does not meet the requirements, and the impedance of the capacitor at this frequency point is also relatively high, which cannot reduce the impedance. Therefore, adding capacitors earlier cannot reduce power noise.
Generally speaking, 11.3KHz should be within the jurisdiction of VRM, and the presence of significant noise here indicates that the VRM circuit design cannot meet the requirements. Here, the performance of VRM is analyzed, and there are many methods for VRM analysis. The main method used here is to simulate its feedback loop Bode plot. The baud chart mainly observes several key information: 1. Crossing frequency, the gain curve crosses the frequency point of the 0dB line; 2. Phase margin, the phase value corresponding to the crossing frequency of the phase curve; 3. Gain margin, the gain value corresponding to a phase of -360 °. Here we mainly focus on two indicators: crossing frequency and phase margin. From the Bode plot of the VRM loop (as shown in Figure a), it can be seen that the passband frequency of the VRM is around 8KHz, with a phase margin of 37 degrees. There are two issues here: firstly, the phase margin of VRM generally needs to be greater than 45 degrees to ensure stable operation of the loop. Here, the phase margin is slightly smaller and needs to be increased; Secondly, the crossing frequency is too low, and the adjustment effect of VRM near the crossing frequency gradually decreases. At this frequency point, the bulk capacitor is not yet effective, so there will be a high impedance near 8KHz, and the noise suppression effect at this frequency point is poor. The figure (b) shows the Bode plot after optimizing the VRM loop, adjusting the phase margin to 50 degrees and pushing the crossover frequency to around 46KHz.

After verifying the ripple of the optimized VRM, it can be seen that the ripple is significantly reduced to 33mv, which can meet the device requirements.

The above case is the process of using the FFT function of an oscilloscope to quickly locate power issues. From this example, it can be seen that the frequency domain analysis function of an oscilloscope can play a significant role in circuit debugging. The FFT function of the oscilloscope, combined with long storage depth, can easily analyze low-frequency long-period signals, which is a prominent advantage in digital circuit debugging.