Sources of Noise in Oscilloscopes How to Use Oscilloscopes to Reduce Waveform Noise

The communication bandwidth of modern satellite communications, 5G mobile communications, and wireless LAN is very wide, which may exceed 500MHz or even reach around 2GHz. For such a broadband signal, although the traditional spectrum analyzer can see the signal spectrum, it is limited by the real-time analysis bandwidth and it is difficult to perform effective demodulation analysis on the modulation quality of the signal. In this case, a broadband oscilloscope can be used. Vector demodulation software is used to perform signal demodulation analysis. In order to simplify the analysis of the problem, in the following chapters we will take a QPSK modulated signal with a carrier frequency of 5.2GHz and a data rate of 50MBaud as an example to introduce how to use an oscilloscope to perform demodulation analysis of radio frequency and microwave signals.

First of all, we can directly observe the time domain waveform of the signal under test on an oscilloscope. The envelope shape of the signal is related to the type and roll-off factor of the shaping filter at the transmitter. Experienced engineers can roughly estimate the envelope shape of the signal through the envelope shape of the signal. The power, modulation rate and modulation method of the signal, but other than that, it is difficult to quantitatively evaluate the quality of the signal modulation.

In order to further demodulate and analyze the modulation quality of this signal, we can use the corresponding VSA (VectorSignal Analyzer: Vector Signal Analysis) software. This software can be installed on the oscilloscope, or it can be installed on an external PC and control the oscilloscope via a network cable or USB cable. After completing the connection with the address setting of the oscilloscope, the VSA software can control the oscilloscope and send the waveform data collected by the oscilloscope to the software for resampling and signal analysis. Its main working principle is shown in the figure below. It can be seen that the VSA software mainly uses a high-bandwidth oscilloscope to directly sample the radio frequency or even microwave band signals, and demodulates the signal according to the completely opposite process of signal modulation, and then extracts the signal from the time domain (the demodulated time domain Analyze the signal from various angles such as waveform), frequency domain (signal spectrum), code domain (demodulated data), modulation domain (modulation quality), etc.

Under normal circumstances, by setting the center frequency point, sweep bandwidth, reference power level, etc., you can roughly see the spectrum of the signal and the resampled time domain waveform. Through this setting, you can also confirm whether there are strong enough signal components in the frequency band to be analyzed. For example, in the figure below, we set the center frequency of the spectrum to 5.2GHz, the Span to 100MHz, and the reference level to 0dbm. , you can see that there is an obvious bulge in the middle of the spectrum, and the entire occupied bandwidth is around 60~70MHz. Therefore, you can confirm that the frequency and other information are set correctly, and the signal has been sent out normally.

In the figure, the spectrum span Span=100MHz, the resolution bandwidth ResBW=1MHz, and the sampling time length is about 3.8us. In order to better observe the details of the spectrum, we usually adjust the resolution bandwidth ResBW, or need to collect a longer time period for signal analysis. We can reduce the minimum resolution bandwidth ResBW that can be set by increasing the number of points involved in spectrum analysis. Since the acquisition length in the time domain is inversely proportional to ResBW, the length of time acquisition can be indirectly controlled by adjusting the number of spectrum points and the setting of ResBW. During the adjustment process, it should be noted that when different windowing types are used in the FFT process, the corresponding acquisition time length under the same ResBW may be different due to different window coefficients. The relationship between the sampling time length, window type, ResBW, and number of spectrum points is shown in the figure below.

The picture below shows the spectrum and time domain waveforms after increasing the number of spectrum analysis points and reducing the resolution bandwidth. It can be seen that since the resolution bandwidth ResBW is reduced to 30kHz, the resolution and details of the spectrum are clearer, and the length of the sampled time domain waveform is also longer.

If there are no problems with the spectrum and time domain waveform of the signal, you can turn on the digital modulation function to perform vector demodulation analysis on the signal. The signal processing flow of vector demodulation analysis is shown in the figure below.

The functions and effects of each key step are described as follows:

· Frequency difference compensation: In the process of vector signal demodulation, for the signal sampled by the ADC, the signal in the frequency band of interest is first filtered out through a digital filter, and then the carrier wave is summed according to the set center frequency point and symbol rate. Symbol locking, and frequency errors at the transmitter and receiver are measured and compensated. The frequency difference between the transmitter and the receiver may cause a continuous phase deviation, which is expressed as a rotation of the constellation in the demodulation result (as shown in the figure below).

· I/Q filtering: After frequency difference compensation, the preliminary time domain waveforms of the I and Q channels can be obtained. After the I and Q waveforms are attenuated and offset compensated, and passed through the corresponding measurement filters, the final I and Q waveforms (called IQMeasure Time in the demodulation software) can be obtained. This measurement filter is used to simulate the shape of the shaped filter at the receiving end. When a root raised cosine filter is used at the transmitter, this measurement filter also uses a root raised cosine filter.

· Symbol demodulation: After obtaining the I/Q time domain waveform, the I/Q waveform can be sampled at the corresponding time point according to the symbol rate to obtain the voltage values ​​of the I and Q channels. According to the corresponding voltage value and then decoding the I/Q symbol coding table, the transmitted symbol data information can be recovered, thereby completing the data demodulation process.

· Generating reference waveforms: Obtaining the transmitted symbol data is not the final goal, because as long as the modulation quality is not particularly poor, the correct data should be obtained, so simply obtaining the demodulated data is not enough, the modulation quality also needs to be quantified analyze. In order to analyze the actual modulation quality, the VSA software will use the demodulated data as a benchmark and mathematically simulate the time domain waveform of these data after passing through an ideal distortion-free transmitter and an ideal receiver. It should be noted that the shaping filter used in this step to reconstruct the time domain waveform is usually called a reference filter and is different from the measurement filter used in step 2. The measurement filter only simulates the filter on the receiving end, while the reference filter contains the joint effects of the transmitting end filter and the receiving end filter. Therefore, if both the transmitter and the receiver use root raised cosine filters, the measurement filter is a root raised cosine filter, and the reference filter is a raised cosine filter.

· Error calculation: Once the actual measured I/Q signal waveform and the reference waveform generated with an ideal transmitter and receiver based on the same data are obtained, the two waveforms can be compared and the error calculated. For example, the time domain waveform of the error can be obtained, the error can also be decomposed into amplitude error and phase error, and the spectrum analysis of the time domain waveform of the error can also be performed.

Among the error analysis methods of measurement results, the most intuitive is the EVM (ErrorVector Magnitude: Error Vector Magnitude) indicator. The definition of EVM is as shown in the figure below. The position of the reference signal on the constellation diagram is used as the reference point. The distance between the actually measured I/Q signal and the reference point on the constellation diagram is used as the error vector. Then this error vector is The ratio of the maximum symbol amplitude as a percentage is called EVM. Each symbol has a corresponding EVM result. Usually, the variance of the EVM results of multiple symbols is taken, and its root mean square value is used as the EVM measurement result of the current modulated signal.

Therefore, in the process of signal demodulation, in addition to setting the correct center frequency point, spectrum span, and reference power level, the most important things are the modulation method, symbol rate, measurement filter type, reference filter type, and filter roll. Decreasing factor setting. Through the previous introduction, it should be easy to understand and set these parameters (as shown in the figure below).

By setting the demodulation parameters and slightly adjusting the demodulation result window, you can see the signal demodulation analysis results as shown below. It can display the spectrum of the original signal, the constellation diagram formed by the I/Q signal vector, the error vector result, the statistical analysis of the error vector, the demodulated original data information, and the signal eye diagrams of the I and Q channels and other information.

In the above demodulation results, the modulation quality of the signal can be analyzed from different angles. For example, from the EVM measurement results, the EVM value is about 4%, which is a relatively normal indicator of a transmitter; from the signal spectrum, the main power of the signal is concentrated in the range of about 70MHz near the center frequency point, which is mainly It is the effect of the baseband shaping filter; from the I/Q vector diagram, on the four constellation points of QPSK, the gathering of vector points at the sampling time is relatively dense, which corresponds to the measurement results of the EVM; from the I channel Looking at the eye diagram of the Q channel, the difference between the high and low levels of the two signals at the middle sampling time is also obvious, indicating that the inter-symbol interference at the sampling time is very small, but at the same time, the amplitude fluctuates greatly in other positions of the eye diagram. This It is determined by the characteristics of the Nyquist filter.