Ten questions and ten answers about oscilloscopes

Oscilloscopes Digital oscilloscopes have always been a good helper for engineers to design and debug products. However, with the development of computers, semiconductors and communication technologies, the signal clock speed of circuit systems is getting faster and faster, and the signal rise time is getting shorter and shorter, resulting in increasingly prominent digital errors caused by the integrity of the underlying analog signal. In response to these new test challenges, oscilloscope suppliers have continuously launched digital oscilloscopes with better performance. However, in order to accurately and quickly analyze system signals, there are many new factors that must be considered when measuring. For example, whether the instrument speed can keep up with the changes in the measured signal, whether the bandwidth is sufficient, whether the measurement method will introduce interference, and even whether the probe used is suitable.

Question 1: Each oscilloscope has a frequency range, such as 10M, 60M, 100M..., the oscilloscope I use is nominally 60MHz, can it be understood that it can measure up to 60MHz? But when I use it to measure a 4.1943MHz square wave, it can't be measured. What's the reason?

Answer: A 60MHz bandwidth oscilloscope does not mean that it can measure 60MHz signals well. According to the definition of oscilloscope bandwidth, if you input a 60MHz sine wave with a peak-to-peak value of 1V to a 60MHz bandwidth oscilloscope, you will see a 0.707V signal on the oscilloscope (30% amplitude measurement error). If you test a square wave, the reference standard for selecting an oscilloscope should be the signal rise time, oscilloscope bandwidth = 0.35/signal rise time × 3, and your rise time measurement error is about 5.4%.

The probe bandwidth of the oscilloscope is also very important. If the bandwidth of the system composed of the oscilloscope probe used, including its front-end accessories, is very low, the bandwidth of the oscilloscope will be greatly reduced. If a 20MHz bandwidth probe is used, the maximum bandwidth that can be achieved is 20MHz. If a connecting wire is used at the front end of the probe, the probe performance will be further reduced, but it should not have much impact on square waves of about 4MHz because the speed is not very fast.

In addition, you should also look at the oscilloscope manual. For some 60MHz oscilloscopes, the actual bandwidth will drop sharply to below 6MHz under the 1:1 setting. For a square wave of about 4MHz, its third harmonic is 12MHz and the fifth harmonic is 20MHz. If the bandwidth is reduced to 6MHz, the signal amplitude will be greatly attenuated. Even if you can see the signal, it is definitely not a square wave, but a sine wave with attenuated amplitude.

Of course, there may be many reasons for not being able to measure the signal, such as poor probe contact (this phenomenon is easy to eliminate). It is recommended to connect a function generator with a BNC cable to check whether the oscilloscope itself has any problems and whether the probe has any problems. If there are any problems, you can contact the manufacturer directly.

Question 2: Some instantaneous signals are lost in a moment. How to capture and reproduce them?

Answer: Set the oscilloscope to single acquisition mode (set the trigger mode to Normal, set the trigger condition to edge trigger, and adjust the trigger level to an appropriate value, and then set the scan mode to single mode). Note that the storage depth of the oscilloscope will determine the time you can collect the signal and the maximum sampling rate that can be used.

Question 3: In PLL, period jitter can measure the quality of a design, but it is very difficult to measure it accurately. Are there any methods and techniques?

Answer: When using an oscilloscope, pay attention to whether its own jitter-related indicators meet your test needs, such as the trigger jitter indicators of the oscilloscope itself. At the same time, pay attention to the use of different probes and probe connection accessories. If the system bandwidth of the oscilloscope cannot be guaranteed, the measurement results will be inaccurate. In addition, regarding the measurement of PLL setting time, you can use an oscilloscope + USB-GPIB adapter + software options to complete it, or you can use a relatively cheap modulation domain analyzer.

Question 4: Why can't my oscilloscope sometimes capture the amplified current signal?

Answer: If the signal does exist, but the oscilloscope can sometimes capture it and sometimes not, this may be related to the settings of the oscilloscope. Usually, you can set the oscilloscope trigger mode to Normal, set the trigger condition to edge trigger, adjust the trigger level to an appropriate value, and then set the scan mode to single mode. If this method still doesn't work, there may be something wrong with the instrument.

Question 5: How to measure power supply ripple?

A: You can first use an oscilloscope to capture the entire waveform, then zoom in on the ripple part of interest to observe and measure (automatic measurement or cursor measurement is possible), and also use the oscilloscope's FFT function to analyze from the frequency domain.

Question 6: How can a new digital oscilloscope be used for microcontroller development?

A: The I2C bus signal generally operates at a rate of no more than 400Kbps, and recently several Mbps chips have also appeared. Some oscilloscopes do not need to consider the impact of different rates when setting trigger conditions, but for other buses, such as the CAN bus, you need to first set the actual current operating rate of the CAN bus on the oscilloscope so that the oscilloscope can correctly understand the protocol and trigger correctly. If you want to further analyze the Inter-IC bus signal, such as protocol-level analysis, you can use a logic analyzer, but it is relatively expensive.

Question 7: Questions about the comparison between analog and digital oscilloscopes: 1. Which of the analog and digital oscilloscopes has more advantages when observing the details of the waveform (for example, observing parasitic waveforms below 1% at zero crossings and peaks)? 2. Digital oscilloscopes generally provide online display of RMS values. What is their accuracy?

Answer: 1) When observing parasitic waveforms below 1%, the observation accuracy is not very good, whether it is an analog oscilloscope or a digital oscilloscope. The vertical accuracy of an analog oscilloscope is not necessarily higher than that of a digital oscilloscope. For example, the vertical accuracy of an analog oscilloscope with a bandwidth of 500MHz is ±3%, which is not more advantageous than a digital oscilloscope (usually with an accuracy of 1-2%). In addition, the automatic measurement function of a digital oscilloscope is more accurate than the manual measurement of an analog oscilloscope for details.

2) For the amplitude measurement accuracy of an oscilloscope, many people measure it by the number of A/D bits. In fact, it will change with the bandwidth of the oscilloscope you use, the actual sampling rate setting, etc. If the bandwidth is not enough, the amplitude measurement error itself will be very large. If the bandwidth is sufficient and the sampling setting is very high, the actual amplitude measurement accuracy is not as good as the accuracy when the sampling rate is low (you can sometimes refer to the user manual of the oscilloscope, which may give the actual number of effective bits of the A/D of the oscilloscope at different sampling rates). In general, the accuracy of oscilloscope in measuring amplitude, including RMS value, is often not as good as that of multimeter. Similarly, it is not as good as frequency counter in measuring frequency.

Question 8: What is the significance of glitch trigger indicator (e.g. 5ns)? If there is a 100MHz oscilloscope, the square wave signal measured is about 10M, and it is a square wave with a duty cycle of about 1:1. Imagine that a 10M square wave has a positive or negative pulse width of 50ns. In what cases can the 5ns performance be really used?

Answer: There are generally two typical applications for glitch/pulse width triggering. One is to synchronize circuit behavior, such as using it to synchronize serial signals, or for applications with very serious interference, when the edge trigger cannot be used to correctly synchronize the signal, pulse width triggering is an option; the other is to find abnormal phenomena in the signal, such as narrow glitches caused by interference or competition. Since the abnormality appears occasionally, it must be captured by glitch triggering (there is also a peak detection method, but the peak detection method may be limited by its maximum sampling rate, so it can generally only be viewed but not measured). In the example mentioned in the question, if the pulse width of the object being measured is 50ns, and there is no problem with the signal, that is, there is no signal distortion or narrowing caused by interference, competition, etc., then the edge trigger can be used to synchronize the signal without using glitch trigger. Depending on the application, the 5ns indicator may not be used. Generally, users set the pulse width trigger to 10ns to 30ns.

Question 9: When choosing an oscilloscope, the bandwidth is generally considered the most. So under what circumstances should the sampling rate be considered?

A: It depends on the object being measured. Under the premise of satisfying the bandwidth, it is hoped that the minimum sampling interval (the inverse of the sampling rate) can capture the signal details you need. There are some empirical formulas for sampling rate in the industry, but they are basically derived for the bandwidth of the oscilloscope. In practical applications, it is best not to use an oscilloscope to measure signals of the same frequency. If the bandwidth of the oscilloscope selected for a sine wave should be more than 3 times the frequency of the sine signal being measured, the sampling rate should be 4 to 5 times the bandwidth, which is actually 12 to 15 times the signal; if it is other waveforms, ensure that the sampling rate is sufficient to capture signal details. If you are using an oscilloscope, you can verify whether the sampling rate is sufficient by the following method: stop the waveform, enlarge the waveform, and if you find that the waveform has changed (such as certain amplitudes), it means that the sampling rate is not enough, otherwise it is fine. In addition, you can also use point display to analyze whether the sampling rate is sufficient.

Question 10: How do you understand "When assessing whether the waveform sampling rate is sufficient, stop the waveform, enlarge the waveform, and if you find that the waveform has changed (such as certain amplitudes), it means that the sampling rate is not enough, otherwise it is fine. You can also use point display to analyze whether the sampling rate is sufficient."?

A: I have the honor to do actual measurements for users and have personally experienced this phenomenon. The object being measured was a signal that looked very random and changed at a high speed, and the user set the trigger level to about -13V. After the waveform was acquired, when I wanted to zoom in on the measurement details, I found that when I changed the oscilloscope time base (SEC/DIV) setting, the signal amplitude suddenly became smaller. I changed the oscilloscope to dot display at the time and found that it seemed that the number of points (storage depth) was not enough. However, after comparing the dot display and the vector display, I found that if the vector display has a certain degree of credibility, then there is a sudden change in the signal in the current two sampling intervals (the inverse of the sampling rate), but it has not been collected (the sampling interval is not fine enough, that is, the sampling rate is not high enough). I changed to an oscilloscope with the same memory depth but a higher sampling rate, and found that the problem disappeared.

The memory depth also affects the actual maximum sampling rate that the oscilloscope can use. Too shallow a memory depth may be a problem, because the memory depth may limit the maximum sampling rate that can be actually used, but in fact, the sampling rate is not enough and the signal details are lost. If the memory depth is not deep enough, the actual sampling rate may not be high, which has little to do with the indicators provided by the manufacturer.