If you don’t pay attention to details, even the most expensive oscilloscope will not be able to measure high-speed signals accurately

With the rapid development of electronic technology, the frequency of communication signals is getting higher and higher, and the signal quality requirements are becoming more and more stringent. Is it enough to just choose an expensive oscilloscope to measure these high-speed signals? In fact, if some details are not paid attention to, even the most expensive oscilloscope may not be able to measure accurately!

1. Bandwidth selection

When measuring high-speed signals, the bandwidth of the test system must be considered first. The bandwidth of the test system includes the bandwidth of the probe and the bandwidth of the oscilloscope. To measure a 100MHz signal, is it enough to use an oscilloscope with a bandwidth of 100MHz? Some users may not have a clear understanding of the concept of bandwidth. They think that an oscilloscope with a bandwidth of 100MHz can measure a 100MHz signal, but this is not the case. The bandwidth refers to the frequency when the sine wave signal attenuates to -3dB, and the digital signals we generally measure are not positive waves, but close to square waves. The bandwidth requirements of the two are different.

According to Fourier transform, square waves can be decomposed into sine waves of odd multiple frequencies. For example, a 1MHz square wave is composed of 1MHz, 3MHz, 5MHz, 7MHz, etc. sine waves superimposed. The following figure shows the response of square wave signals under different filters. They are the square wave responses when the filter is set to the square wave fundamental frequency, the 3rd harmonic spectrum, the 5th harmonic frequency, and the 7th harmonic frequency.

Figure 1 Filtering with a cutoff frequency equal to the square wave frequency

Figure 2 Filtering with the cutoff frequency being the third harmonic frequency of a square wave

Figure 3: Filtering with the cutoff frequency being the 5th harmonic frequency of a square wave

Figure 4: Filtering with the cutoff frequency being the 7th harmonic frequency of a square wave

It can be seen that in order to obtain more complete square wave information, at least 5 harmonic components are required, and if more accurate information is to be obtained, more harmonic components need to be measured.

Therefore, when selecting the bandwidth of the oscilloscope and probe, you should choose a bandwidth that is at least the 5th harmonic frequency of the square wave signal being measured.

2. Probe Selection

An oscilloscope cannot measure signals directly. The signal must be transmitted to the oscilloscope through a physical connection. This physical connection is the probe. Probes are crucial for high-speed signal measurement. Common passive probes generally include 1:1 probes and 10:1 probes. In addition to different attenuation ratios, these two probes also produce great differences in high-speed signals. To explain this problem, we need to discuss a key feature of the probe - the loading effect.

Ideally, we hope that the impedance of our measuring equipment is infinite, so that the connection of the test equipment will not have any impact on the system under test, thus ensuring the authenticity of the measurement. Unfortunately, the measurement system cannot have infinite input impedance. At this time, what impact will the connection of the measuring equipment have on the system under test? Assume that the system under test is as shown in the figure below.

Figure 5 Equivalent schematic diagram of the system under test

It can be seen that the voltage at the measuring point is:

When an oscilloscope is used for measurement, due to the input resistance and parasitic capacitance of the oscilloscope, the equivalent circuit diagram at this time will be as shown in the figure below:

Figure 6 Schematic diagram of probe access equivalent

It can be seen that the voltage at the measuring point is:

Where Rin is the input impedance, Cin is the parasitic capacitance, and s represents the frequency. It can be seen that the voltage at the test point has changed, which has caused the signal itself to change before and after the probe is connected. From the formula, it can be seen that the larger the Rin, the smaller the impact on the signal. The term 1/(Cin×s) is the reciprocal of the product of the parasitic capacitance and the frequency of the measured signal. The higher the test signal frequency, the greater the impact of this term. To reduce the impact of this term, the only way is to reduce the capacitance of the parasitic capacitance Cin as much as possible.

The following figure shows the model of ×1 probe:

Figure 7 ×1 probe model diagram

Since the probe must have a length of wire, otherwise it will be inconvenient to measure, and the length of the wire is generally more than 1 meter. This leads to a very large parasitic capacitance, about 100pF. When measuring high-frequency signals, it will produce a large load effect. Let's take a look at the ×10 probe model:

Figure 8 ×10 probe model

It can be seen that the input capacitance Cin of the ×10 probe is a series connection of 10pF and the following capacitor. According to the capacitor series connection formula, Cin must be less than 10pF, which is much smaller than the input capacitance of the ×1 probe, and Rin has increased to 10MΩ. Therefore, the ×10 probe has smaller parasitic capacitance and higher input resistance, which greatly reduces the load effect of the probe.

Therefore, when measuring high-speed signals, you need to choose a probe with ×10 or higher attenuation input impedance.

3. Choice of grounding method

In traditional usage, the grounding method of the oscilloscope is the long grounding clip wire. This grounding method is indeed a simple and convenient grounding method, but it is not a rigorous and accurate grounding method.

Figure 9 Schematic diagram of grounding clamp

Since the ground clip is relatively long, it will form a parasitic inductance Lgnd. As the clip length increases, this inductance will also increase, and this loop inductance will resonate with the input capacitance Cin of the oscilloscope probe. This causes the amplitude-frequency characteristic of the oscilloscope to become uneven, resulting in inaccurate measurements. The figure below shows the equivalent circuit when using the ground clip.

Figure 10 Equivalent circuit diagram of grounding clamp

The figure below is the spectrum characteristic curve simulated by this equivalent circuit:

Figure 11 Spectrum characteristic curve

It can be seen that at frequencies above 60MHz, the amplitude has produced an overshoot of more than 3dB, and when it reaches about 100M, the overshoot reaches the maximum amplitude. Therefore, if a ground clamp is used, measuring signals above 60MHz will produce relatively large distortion. The correct way should be to use a ground spring. The ground spring has a very small inductance and can greatly increase the bandwidth of the probe.

Figure 12 Schematic diagram of grounding spring

4. Measurement location selection

For high-frequency signals, PCB traces can no longer be simply treated as short-circuits, but the effects of propagation delays, signal reflections, etc. on the lines must be considered. The reason why traditional low-speed signals can ignore the effects of PCB traces is that their wavelengths are long and the length of PCB traces can be ignored, so they are treated as concentrated components. However, the wavelength of high-frequency signals is short, and the length of PCB traces can no longer be ignored. Signals must also be considered from a wave perspective. The following figure shows the waveforms of the same signal measured at the source and terminal:

Figure 13: Differences in measurements at different locations

The reason for this is that electrical signals are transmitted on PCBs like waves. Their propagation speed is generally half the speed of light. Therefore, there will be a delay in the propagation of signals on PCBs, and reflections will occur according to the change in characteristic impedance. In the above figure, the terminal device of the signal is not terminated, so when the signal reaches the terminal, a reflected wave will be generated, which will be reflected back to the source end. After the delay on the PCB, the reflected wave and the transmitted signal will be superimposed, thus generating a waveform at the source end. Similarly, not only at the source end, but also on the entire transmission line, the transmitted signal and the reflected signal will be superimposed. The difference is that the phase difference between them is different, and the superimposed waveform is also different.

It can be seen that the choice of measurement point location will lead to huge differences in measurement results. Therefore, when measuring high-speed signals, the closer the measurement point is to the terminal device, the better, so that the waveform of the signal received by the terminal device can be truly measured.

Summarize

This article points out some precautions for measuring high-speed signals, which are summarized as follows:

1. When selecting the bandwidth of the oscilloscope and probe, you must select a bandwidth that is at least the 5th harmonic frequency of the square wave signal being measured.

2. When measuring high-speed signals, you need to select a probe with ×10 or higher input impedance.

3. When choosing a grounding method, the grounding loop inductance should be reduced as much as possible, such as using a grounding spring, so that the bandwidth of the measurement system can be truly utilized.

4. When measuring high-speed signals, the closer the measurement position is to the terminal device, the better, so that you can truly measure what kind of signal the terminal device receives.

The above is the main content of this article. The ZDS4054 Plus oscilloscope has a bandwidth of 500MHz and can truly measure 100MHz square wave signals. It can achieve 9pF input capacitance with the probe, greatly reducing the measurement load effect and can measure most high-speed signals well.