Lab Skills for Switch-Mode Power Supply Evaluation—Part 1: Measuring Voltage Ripple and Switching Node

Abstract

Lab skills are essential to characterize and validate the exceptional performance of Analog Devices’ power converter products. Without accurate measurement techniques, engineers evaluating different solutions don’t have the necessary information to make an informed decision. This article discusses how to carefully select passive probes for measurement, how to optimize measurement methods, and alternative measurement techniques to further aid in diagnosing converter operations. Reference setups are shown to illustrate how typical data sheet figures are captured such as voltage ripple and switching waveforms.

Introduction

When evaluating a switch-mode power supply (SMPS), there are many key parameters to assess and many factors to consider when observing a particular measurement. Making these measurements accurately is necessary to ensure design decisions are not led astray by faulty data, and proper oscilloscope probe selection is the first step.

Oscilloscope Probe Selection

An oscilloscope is a powerful tool that engineers use to evaluate the performance of an SMPS. However, it cannot be overstated how important proper measurement technique is to oscilloscope capture accuracy. The first step is selecting the right oscilloscope probe for the measurement being taken.

Passive probes are versatile and useful for accurate signal measurements, and they contain no active circuitry and do not require external power. While less disruptive measurements for extremely sensitive circuits can be achieved with active probes, they are also more complex and expensive. This section will explore oscilloscope measurements using passive 10× probes, passive 1× probes, and coaxial cables, with a brief discussion on how to determine which of the three is appropriate for certain measurements.

Oscilloscope Probes: 10×

10× probes are the general-purpose standard probe for most modern oscilloscopes. They are designated 10× because they have a 10:1 attenuation ratio and reduce the signal being measured by a factor of 10. The oscilloscope’s display adjusts for this attenuation to display the correct voltage measurement, but this should be verified by the user because oscilloscopes may vary. The probe has a large internal impedance, which, along with the oscilloscope impedance, divides the voltage signal by 10, as shown in Figure 1. This high impedance reduces loading on the oscilloscope and allows the probe to measure high voltages typically in the hundreds of volts. The typical bandwidth for 10× probes reaches hundreds of megahertz. The Keysight N2873A 10× probe, for example, is rated for 400 V DC voltage and up to 500 MHz.¹

Figure 1. (a) Keysight N2873A 10× probe and (b) its simplified circuit.

Care must also be taken to check the voltage derating curve vs. the frequency of the probe. A voltage derating curve shows the maximum voltage the probe can measure while the signal has a certain frequency. As frequency increases, the maximum voltage that the probe can pass will decrease. The curve in Figure 2 shows an example of N2873A.

Figure 2. An N2873A 10× probe voltage vs. a frequency derating plot from its data sheet by Keysight.

To ensure measurement accuracy in the kilohertz frequency range, many 10× probes have a built-in adjustable compensation screw, as shown in Figure 3. This probe adjustment allows users to match the probe capacitance with the oscilloscope’s input capacitance, C_t/(C_in + C_p) = 1 MΩ/9 MΩ. Once adjusted, the voltage measurement will have the correct frequency response, ensuring that fast edges are measured correctly without overshoot or undershoot distortion. The probe should be calibrated each time it is connected to a new oscilloscope input to account for slight differences in the analog front ends, even when switching between different channels of the same oscilloscope. To calibrate the probe, connect it to the oscilloscope’s on-board square wave generator and adjust the probe’s compensation screw until the observed voltage signal appears square with little to no overshoot or rounding, as shown in Figure 4.

Figure 3. How to tune 10× probe compensation with a screwdriver tool.

Figure 4. (a) Sharp square, (b) overdamped, and (c) underdamped oscilloscope calibration measurement captures.

The 10× probes are especially suited for measuring switch-node voltage in SMPSs, and can also be used to measure V_IN, V_OUT, and high frequency signal-level waveforms. The 10× probe is recommended for these measurements because of its ability to pass higher frequencies and voltages.

Oscilloscope Probes: 1×

Another common passive probe is the 1× probe, as shown in Figure 5. These probes have a 1:1 attenuation ratio and do not attenuate the voltage signal. These probes have a much lower impedance compared to the 10× probe and do not require user calibration. The oscilloscope does not adjust its scale for the measurement, allowing the screen capture to have a finer resolution—typically down to 1 mV/div.

Figure 5. (a) A 1× probe (Fluke PM9001) and (b) its simplified circuit.

These probes have a voltage range in the tens of volts. The bandwidth for these probes is typically rated to the tens of megahertz. A Fluke PM9001 1× probe, for example, has an impedance of less than 1 kΩ, a maximum voltage of 30 V, and a bandwidth of 15 MHz.² A 1× probe is recommended for low voltage and low frequency measurement. When measuring voltages in SMPSs, this probe is recommended for small signal tests such as input or output voltage ripples as long as the frequency of these ripples is in the low megahertz.

Coaxial Cables

The 10× and 1× probes connect to the oscilloscope through cables with coaxial construction. Coaxial cables have a center conductor that is used to transmit a signal and an outer metal braided mesh layer to shield the cable and the signal from electromagnetic interference (EMI). The coaxial cable itself can also be used as a voltage probe with proper terminations, as shown in Figure 6. These cables typically have a 50 Ω characteristic impedance, so the oscilloscope input impedance must be set to 50 Ω.

Figure 6. A coaxial cable with a BNC connector.

When the 50 Ω impedance setting is applied in the oscilloscope, the maximum voltage that the scope can support is then limited, typically to 5 V. This limit protects the oscilloscope from excess load. Although coaxial cable measurements are limited to a low voltage range, they pass a much higher bandwidth. A Pomona 2249-C-12 BNC-BNC cable has a bandwidth of 4 GHz, higher than that of many oscilloscope probes.³

In order for coaxial cables to be used for SMPS measurements, the printed circuit board (PCB) itself must be equipped with the appropriate matching connector. This means that the board must be designed in advance to incorporate the connector or have enough space to insert one. There are several styles of coaxial cables varying by gauge and connector type that are suitable for demo board measurements.

Many ADI μModule^® demo boards utilize 10 mm diameter BNC connectors. The female BNC-BNC is commonly used for oscilloscope measurements because of its sturdy, reliable construction, and wide availability. These coaxial cables are typically used to measure V_OUT ripple in a dynamic load circuit on the output of an SMPS. Subminiature Version B (SMB) and U.FL connectors are two types of connectors used especially for high frequency measurements. These two examples are also good for PCBs where space is a critical concern because of their small footprint.

Probe Selection Summary

The 10× probe is recommended for capturing high voltage and some high frequency measurements. The 1× probe is recommended for small voltage and low frequency measurements. Coaxial cables are useful for low voltage and high frequency measurements but require a connector at the test point on the PCB. These differences are summarized in Table 1.

Table 1. 10×, 1×, and BNC Coaxial Cable Usage Summary 10× Probe
	10× Probe (Keysight N2873A)	1× Probe (Fluke PM9001)	Coaxial Cable (Pomona 2249-C-12)
Voltage Range	400 V	30 V	5 V (oscilloscope limited)
Bandwidth	500 MHz	15 MHz	4 GHz
Allows Calibration	Yes	No	No
Requires PCB Connector	No	No	Yes
Recommended Usage	High frequency, high voltage	Low frequency signal, small signal voltage	High frequency signal, small voltage

Measuring Voltage Ripple

Output voltage ripple is a primary specification for power supplies and is a major parameter used to compare different designs. When measuring a ripple that is tens of millivolts or smaller, the location on the board and the method of measurement can significantly influence the results. Coaxial cables can be used to measure the input or output ripple if the voltage is less than 5 V, and if the proper connectors are available on the evaluation board. Otherwise, a 10× probe and a 1× probe are both good options. However, the standard ground clip of these probes would form a long measurement loop that will reduce the accuracy of measurement by increasing the probe’s impedance. Pigtail leads are recommended to take this measurement and should make solid contact with the probe tip and ground ring as shown in Figure 7. The difference between test results with a standard probe ground connection and pigtail leads can be seen in Figure 8. Standard bus wire can be used to construct pigtails by wrapping them into a tightly looped coil. Be sure to avoid creating the scope pigtail on the probe itself because this could damage the probe.

Figure 7. Probe measurement connection with scope pigtails.

Figure 8. V_OUT ripple (a) with normal probe ground clip vs. (b) with scope probe pigtails.

Output voltage ripple is typically measured across the device’s output voltage sensing point. In μModule demo boards, the optimized minimal ripple is captured on a ceramic capacitor underneath the board, directly beneath the SMPS output. By measuring on a ceramic capacitor, the ripple from the capacitor’s equivalent series resistance (ESR) is less than when measured across a bulk capacitor. By measuring on the bottom of the board, opposite the converter, the board itself provides shielding from EMI noise. In other applications, users may be interested in measuring output voltage ripple at a point far from the converter such as where load is applied. The shape of the output voltage ripple will differ here vs. next to the converter because of parasitic effects from the PCB and the amount of capacitance at this measurement point.

Input voltage ripple should be measured across the input voltage capacitor closest to the IC, as this directly measures the input voltage seen by the IC. Oscilloscope probe pigtails should be used in this measurement for the same reasons described above.

Measuring Switch-Node Waveforms

Proper measurement of the switch node is critical because improper measurement techniques can yield incorrect waveforms, most commonly by introducing false ringing, which can derail debug efforts. To prevent this, use short pigtail leads to ground the oscilloscope probe. Often there will not be an easily accessible ground next to the switch node. To remedy this, scrape off a small patch of solder mask from the ground plane next to the switch node and install a pigtail on the newly exposed ground to use for the measurement. Take care not to short the switch node to ground while doing this. The difference between switch-node waveforms with a standard probe ground connection and with pigtail leads can be seen in Figure 9.

Figure 9. A switch-node waveform measured (a) with a standard scope probe ground lead and (b) with pigtail connections.

In some μModule parts, the switch node is integrated into the part to reduce the solution’s area and is not directly accessible, such as for the LTM8050, a single-channel, 2 A stepdown converter. For these parts, the switching node waveform shape can still be viewed by placing a floating scope probe above the μModule package. This will take a coupled measurement of the waveform shape that can be used to view the switching frequency and check for cycle to cycle stability. This coupled waveform will not give an accurate voltage magnitude measurement.

Figure 10. Example of getting SW node waveform from μModule package by floating the probe over the package, taking a coupled measurement.

The best probe to measure the switch node is typically a 10× probe, as the signal is typically higher frequency or amplitude than a 1× probe will be able to measure.

Conclusion

SMPSs are the go-to power circuit for most applications requiring voltage increases or significant voltage decreases, and for good reason—they offer the optimal combination of noise performance, efficiency, and solution size for many situations. There is a plethora of decisions to be made when designing an SMPS circuit, and accurate measurements of key parameters are needed to inform these decisions and ensure you are making the right choice.