How to use an oscilloscope to assist your design?

Q1: The bandwidth of TDS3032B is 300MHz, and the sampling frequency is 2.5G/s, which is 8 times the bandwidth. Is there a fixed relationship between bandwidth and sampling frequency? We also have an oscilloscope from another manufacturer, with a bandwidth of 100MHz and a sampling frequency of only 200MHz. Why is the bandwidth sampling frequency ratio of the two oscilloscopes so different?

A1: Bandwidth is the most important indicator of an oscilloscope, because there is an ADC in a digital oscilloscope, and its sampling rate theoretically needs to meet the Nyquist sampling theorem, that is, each cycle of the highest frequency signal of the measured signal theoretically needs to sample at least 2 points, otherwise it will cause aliasing. But in reality it also depends on many other factors, such as the waveform reconstruction algorithm. Tektronix oscilloscopes use advanced waveform reconstruction algorithms, and only 2.5 points are needed for each cycle of the measured signal to reconstruct the waveform. Some oscilloscopes use linear interpolation algorithms, which may require 10 points. Generally, a sampling rate of 4-5 times the bandwidth can accurately reproduce the waveform.

Tektronix's TDS3000B series is a "real-time sampling" oscilloscope, that is, its single-shot bandwidth (the ability to capture a single signal) = repetitive bandwidth. The single-shot bandwidth of the other oscilloscope you mentioned is obviously less than 100MHz. You can take a look at its indicators.

Q2: Under the condition of a certain bandwidth, does it make no sense to have a too high sampling frequency? Because the bandwidth is already fixed. Unless you need to capture glitches.

A2: Bandwidth is the basic condition to limit the capture of high-frequency components of the measured signal. Using Tektronix's oscilloscope, only 2.5 points are needed for each cycle of the measured signal to reconstruct the waveform to the maximum extent. Some other oscilloscopes require more than 4 samples/cycle, that is, a 100MHZ bandwidth oscilloscope needs a sampling rate of at least 400MS/s for a single acquisition. Some oscilloscopes even need 10 points (linear interpolation technology) to ensure that the acquired signal is meaningful.

Q3: Are there digital amplifiers now? Is the penetration rate high?

A3: The digital amplifiers we often talk about now are generally called Class D amplifiers. They are based on PWM and delta-sigma modulation technology. They are widely used in some applications that require high efficiency but low THD, such as bass amplifiers and speaker amplifiers in mobile phone designs.

Q4: Regarding glitch measurement, I have consulted relevant technicians before, and the answer I got was that the smallest glitch that an oscilloscope can capture is the sampling rate of the oscilloscope. Do all oscilloscopes follow this rule? Will the pre-filter of the oscilloscope have an impact on it?

A4: It cannot be asserted that all oscilloscopes are like this. For example, some oscilloscopes reach 1GS/s, but the bandwidth is only 60MHz. Obviously, it is impossible to capture 1ns glitches. In fact, the ability to capture glitches depends not only on bandwidth and sampling rate, but also on waveform capture rate, that is, the number of waveforms that can be captured per second. For details, please refer to Tektronix's application article on DPO.

Q5: After the oscilloscope is grounded, the MOSFET explodes. What is the reason?

A5: To ensure personal safety during testing and obtain good measurement results, the ground wires of all probes of the oscilloscope are generally connected to the case and to the ground wire of the oscilloscope power cord. Therefore, when you measure the waveform of the MOSFET tube in the power supply, if any point is not grounded, there will be problems.

Cutting the ground wire can prevent short circuit problems in the MOSFET tube test, but it will also bring some other test problems, such as the oscilloscope case being charged, and the oscilloscope case distributed parameters affecting the measured signal. The solution is to use a differential probe, such as Tektronix's P5205, which can measure the so-called differential signal when neither of the two test points is grounded.

Q6: The I/Q signal I measured is the baseband. The purpose of my measurement is to compare and distinguish good mobile phones from bad mobile phones. But I can't see any signal when I measure. Is there a problem with the oscilloscope settings?

A6: IQ signal: Is your test point high impedance? Is the signal amplitude very small? Even the Tektronix oscilloscope can only do 1mV/grid. Adding a 10:1 probe is equivalent to 10mV/grid. Generally, the XY display function of TDS3000B can be used to measure IQ signals, or if you have a clock, use the XYZ display function. If you still have questions, please contact me directly, and I will arrange for Tektronix engineers to discuss and solve this problem with you.

Q7: Our company previously used passive probes (such as P6139A, bandwidth 500M) with oscilloscopes such as TDS744 and TDS745. After purchasing an active probe (P6237), from the test waveform (especially when measuring high-frequency signals), the test results of the two are quite different. From the probe parameters, we know that the input capacitance of the active probe is <1pF, while that of the passive probe is about 10pF. It seems that the test results of the active probe can better reflect the true situation of the signal. Since the passive probe has a great attenuation on high-frequency signals, what is the significance of its 500M bandwidth? When we test the signal, how do we choose to use an active or passive probe according to the test situation?

A7: Your P6139A probe plus Tektronix's 500MHz oscilloscope can still achieve a typical bandwidth of 500MHz, but as you said, the input capacitance is different. This capacitance will produce a load effect on the signal to be measured, causing signal ringing and shape changes. Therefore, using an active probe at this time can reflect the true situation of the signal. In fact, when using a probe, we must not only consider bandwidth, but all these factors must be considered when measuring high-frequency signals: bandwidth/rise time, dynamic range, load effect, grounding effect, and resonance effect. Especially

when using P6139A, you must also consider the influence of the ground wire. The ground wire on the probe will also cause ringing. When measuring high-frequency signals, the length of the ground wire should be shortened as much as possible. In addition

, the P6247 you use is an active differential probe, and the influence of common mode may also be a factor.

The main reason for choosing a passive probe is its large dynamic range. For example, the P6139A can measure signals from millivolts to hundreds of volts, while the P6247 can only measure +-8.5V signals. In addition, the price of active probes is also a factor.

For details, please refer to Tektronix's "ABC of Probes" and "Digital Signal Integrity Testing"

Q8: When viewing waveforms on an oscilloscope, is it better to use external triggering or self-triggering? What is the difference between the two?

A8: The usual triggering of an oscilloscope is edge triggering, which has two triggering conditions, trigger level and trigger edge. That is, when the rising edge (or falling edge) of the signal reaches a certain level (trigger level), the oscilloscope triggers. The oscilloscope will only use external triggering when there is a problem with the self-triggering of the signal. There is no question of which is better. This problem may usually be that the signal is relatively complex and there are many points that meet the triggering conditions. It is impossible to trigger at the same position every time to obtain a stable display. At this time, an external trigger is needed.

Q9: I found a rather strange question. Why does a seemingly ordinary oscilloscope probe often sell for a very high price? Ask the experts, what are the main factors that determine the price of an oscilloscope probe? How long is the service life of a general oscilloscope probe? Does it need to be calibrated regularly?

A9: There are many types of oscilloscope probes with different performances, such as high voltage, differential, active high-speed probes, etc. Tektronix has nearly 100 different probes, and the prices range from a few hundred RMB to nearly 10,000 US dollars. The main determinants of price are of course bandwidth and function. The probe is the part of the oscilloscope that contacts the circuit. A good probe can provide the fidelity required for testing. To achieve this, even passive probes must have a lot of passive device compensation circuits (RC networks) inside. It is difficult to say how long the oscilloscope probe will last, depending on your use environment and method. The standard does not have clear measurement regulations for probes, but for passive probes, at least when replacing probes or switching channels, probe compensation adjustment must be performed. All active probes should be preheated for at least 20 minutes before use, and some active probes and current probes require zero drift adjustment. In short, only by using the probe well can you use your oscilloscope well. For details, please refer to Tektronix's "Probe ABC". If you need it, you can send an email to Tektronix to request it. china.mktg@tektronix.com

Q10: When we are measuring the conducted disturbance of the off-board signal line, we found two large noise signals at two specific frequencies (one is 659K and the other is 1.977K). Our preliminary analysis shows that it is caused by the switching power supply chip on the board, but we are not very clear about how to use an oscilloscope to measure such noise signals. Experts please give us some advice!

A10: There are several factors to consider when an oscilloscope can test noise signals: 1. The amplitude of the measured signal, whether it is a small signal, the oscilloscope can test uA-level signals with the probe; 2. The frequency of the measured signal; 3. Improper connection of the probe will generate noise and affect the test results.

Q11: How to use TDS3052B to measure the maximum value of the modulated wave with a carrier frequency of tens of K and a modulating wave frequency of the power supply frequency?

A11: Your power frequency input may be a low frequency of 50Hz/60Hz, and the carrier frequency is tens of K. A power frequency cycle is about 20ms. If the oscilloscope needs to observe a 20ms signal, the duration acquisition window of the oscilloscope must be at least 2ms/div ×10 grids. At the same time, the sampling rate of the oscilloscope is determined based on the carrier signal of tens of K. Finally, the required acquisition memory length can be estimated to determine whether it can meet the test requirements. (end)