Oscilloscope memory depth application

In the Zhiyuan Electronic Oscilloscope WeChat user exchange group, I saw some information like this: a user needs to use a multi-node high-speed serial communication bus in a project, using CPLD as the transceiver controller , and the bus design speed is 5Mbps. Due to the high bit rate, he wants to test the stability of the bus signal. I happened to have a ZDS2022 oscilloscope on hand, because I had learned that the storage depth of the ZDS2022 oscilloscope is as high as 112Mpts, so I wanted to try it. In order to test the stability of the signal, I deliberately set a frame of data to a large size, 5Mbps, and transmit for 3 seconds. Set the time base to 500ms, and want to record all the waveforms and analyze them slowly.

　　After setting the storage depth to automatic, he zoomed in and observed the acquired waveform under the same signal source and different time bases, and found that the waveforms were different. What happened?

　　Sampling Rate and Nyquist Sampling Principle

　　In fact, the essence of this problem involves the sampling rate of the oscilloscope. "Sampling" means collecting sample points, which is to collect discrete digital signal sample points in the continuous analog signal. By measuring the voltage waveform amplitude at equal time intervals and converting this voltage into digital information represented by eight bits of binary. The smaller the time interval between the sampled voltages, the closer the reconstructed waveform is to the original waveform.

　　Figure 1: Equally spaced sampling

　　Figure 2 Schematic diagram of oscilloscope sampling analysis

　　When sampling, the Nyquist sampling principle must be followed to avoid waveform distortion. The Nyquist sampling principle states that for a signal with a maximum frequency fmax, the equidistant sampling frequency fs must be twice the maximum frequency fmax, so that a unique signal can be reconstructed without waveform aliasing.

　　Figure 3 Schematic diagram of several ratios of fs and fmax

　　Since the premise of the Nyquist principle is based on infinite time and continuous signals, but no oscilloscope can provide infinite time record length (the maximum number of points that an oscilloscope can provide is directly affected by the memory depth), it is usually not enough to use a sampling rate twice the highest frequency component. In practical applications, it is usually 5 times or even higher. In order to ensure the accuracy of the measurement, the oscilloscope is usually required to maintain a high sampling rate.

　　Sampling rate and memory depth

　　When measuring waveforms with an oscilloscope, there is an important formula: storage depth = total capture time x sampling rate. The capture time is determined by the sampling window. When a horizontal time base gear is set, the total capture time is determined. If I want a higher sampling rate, I must ensure that there is a large enough storage depth. Considering this, the ZDS2022 oscilloscope provides you with a manually settable storage depth of up to 112Mpts in the [Horiz] button. You can choose the appropriate storage depth according to your needs. There is an option for automatic storage depth in the ZDS2022 oscilloscope, which is set to automatic by default. When you select automatic storage depth, an automatic maximum value option will appear, and the automatic maximum value is set to 1.4Mpts by default.

　　Automatic storage depth and automatic maximum value

　　Now let me share with you what this automatic storage depth and automatic maximum value are all about.

　　When the storage depth is set to automatic, the current storage depth of the oscilloscope is the waveform data on the current screen. If the waveform data on the current screen is greater than the automatic maximum value, the storage depth at the bottom right of the screen will remain at the automatic maximum value. Otherwise, once the storage depth is insufficient, the sampling rate will be sacrificed. When the sampling rate is reduced to a level that does not conform to the Nyquist sampling principle, problems will occur in the collected waveform, just as this user communicated with us in the group. In fact, he did not set a large storage depth at the beginning. When the horizontal time base was set to 500ms/div, the sampling rate was reduced to 200KSa/s because the storage depth was only set to 1.4Mpts.

　　Table 1 Comparison of waveforms at different time bases when the storage depth is 1.4Mpts

　　After setting the trigger conditions, press the [Single] key to zoom in the waveform to a horizontal time base of 2ms/div. Since the waveform is collected at a sampling rate of 200KSa/s, and the design speed of the bus under test is 5Mbps, it does not meet the Nyquist sampling principle, so you will see a segmented waveform. In fact, at a low sampling rate, if the time base is very small, the waveform at the "disconnected" position will not be visible due to the small amount of data per frame, and the whole looks like a "disconnected" waveform. However, the actual bus waveform is not disconnected in this way.

　　So the user promptly raised his doubts in our WeChat oscilloscope user exchange group. After the R&D colleagues patiently answered his questions, he set the storage depth to 56Mpts in the [Horiz] key and tried to set the horizontal time base to 200ms/div. At this time, the sampling rate was 20MSa/s, which is 4 times the bus speed. The displayed waveform was enlarged to a horizontal time base of 500ns/div, and the waveform was not distorted.

　　Table 2 Comparison of waveforms at different time bases when the storage depth is 56Mpts

　　In fact, in this case we can set the storage depth to 112Mpts. The greater the storage depth, the greater the sampling rate will be at the same capture time. Once the sampling rate is very high, the reconstructed waveform on the screen will be more accurate.