Oscilloscope Basics - 10 Power Integrity Measurement Tips

Summary of Power Integrity Measurement Tips and Tricks

Tip 1: Use 50 Ω oscilloscope inputs—they generally have the lowest noise, and you can start with a null measurement to get a feel for how noisy your oscilloscope measurement system is before moving on to power supply noise measurements.

Tip 2: Use only as much bandwidth as necessary.

Tip 3: Try to use a probe with less attenuation, preferably a 1:1 probe.

Tip 4: Use probe offset to amplify the signal.

Tip 5: If you choose a DC blocker, you must use it flexibly.

Tip 6: Note the loading effect on the power supply when connected through the 50 Ω terminals of the oscilloscope (50 Ω DC). Tip 7: Use the FFT to gain another view of analysis.

Tip 8: Trigger on suspected noise sources and use averaging to remove extraneous noise.

Tip 9: Use enough bandwidth to capture troublesome transients and noise.

Tip 10: Get the most out of  the N7020A power supply probe .

For decades, the electronics industry has been advancing at a rapid pace, driven by Moore’s Law, with increasingly powerful and feature-rich products that enable all of us to enjoy a more modern life. The explosive growth of affordable microcontrollers means that more and more devices (such as home appliances, cars, medical devices, wearable devices, IoT, smartphones, and the cloud) are working under the control of microcontrollers, providing better performance and more diverse features. However, designers of these products are also faced with the challenge of supplying “clean” power to the devices and circuits in their products. To this end, the industry has invested a lot of time, manpower, and equipment to focus on designing the power distribution network (PDN) in modern products. Designers often use real-time oscilloscopes to measure the DC power in these products. This application note introduces practical techniques for measuring and analyzing DC power supplies and discusses how to select and evaluate tools used in DC power supply measurements.

PDN and Power Integrity

Power integrity (PI) is a common term in the electronics industry that refers to the analysis of how effectively the power supply in a system supplies and converts power to the load. The passive components and interconnects (including semiconductor packages) from the power supply to the load form the power distribution network (PDN), and the power output of the power supply is supplied to the load through this PDN. Power integrity analysis typically includes measurements from DC to several gigahertz. Common power integrity measurements are:

– PARD — Periodic and Random Interference, a term widely used in the industry, is defined as the deviation of the DC output from its mean value, with all other parameters held constant. It is a measure of the unwanted AC and noise components that remain after the DC output has passed through conditioning and filtering circuits, expressed in either RMS or peak-to-peak values. Peak-to-peak measurements are more common in bandwidths between 20 Hz and 20 MHz. PARD-like variations below 20 Hz are often referred to as drift.

– Load response – refers to static or transient loads and is a measure of the ability of a power supply to maintain output within specified thresholds under predetermined load conditions. It typically involves measuring the power supply’s transient recovery time, which is the time it takes for the power supply to recover from a transient to within a predefined stable frequency band when a load change occurs.

– Noise – refers to the deviation of the DC supply from its nominal value. Noise can include random noise (such as thermal noise) and spurious signals (such as switching coupling or PARD and load response from adjacent circuits).

Power Issues

As component density and speed continue to increase in successive generations of products being designed, the importance of "clean" power increases. DC power supply deviations are probably the largest cause of clock and data jitter in digital systems. The voltage drop from the power supply to a digital device can reduce the propagation delay through the gating circuitry within the device, resulting in reduced timing margins or even bit failures. To combat this, power supply tolerances are being squeezed to 5% or less.

As the switching speeds and slew rates of digital devices have increased dramatically, the potential for switching noise to enter the power supply has also increased. This noise appears over the bandwidth of the switching currents, which can easily exceed 1 GHz.

Reducing signal amplitude in digital systems allows faster switching, but this also requires a reduction in the noise margin of the power supply.

Improving efficiency or reducing power consumption is another reason to require tighter control of power supply tolerances. If the power supply tolerance was previously 10% and is now reduced to 5%, the power consumption of the design can also be reduced by up to 5%.

Next, designers are challenged to measure the smaller, faster AC signals that reside on the DC source.

DC power supply noise

Ideally, a DC power supply would not generate any noise. So where does the noise come from?

Thermal noise is inevitable on the power supply, that is, electronic noise generated by the thermal disturbance of electrons, which will form simple Gaussian noise. Gaussian noise is usually not the largest noise source.

The main sources of noise on a DC power supply are the switching noise of the power supply itself and the noise caused by the switching current of the components in the circuit. Switching generates transient current demand. This noise caused by switching events may appear randomly in time, but they tend to be consistent with the clock in the system.

We can think of the noise on a DC power supply as a combination of "signals" such as power supply switching noise and switching current noise, and they are superimposed on the DC power supply, which makes it easier to measure and analyze.

Measurement Challenges

Since DC power supply noise has a wide bandwidth, users often prefer to use an oscilloscope when measuring this noise, because oscilloscopes not only have very wide bandwidth, but are also easy to use and readily available. Oscilloscopes can also provide a unique analytical perspective on the causes of noise, as shown in the following example.

Real-time wideband oscilloscopes and the probes they are equipped with have a certain noise floor. If the noise amplitude of the oscilloscope and probe is similar to the noise of the DC power supply, it will make DC power supply noise measurements very troublesome.

Dynamic range is another challenge in measuring DC power supply noise. The power supply output you are interested in is at a certain DC level, while the small AC signal (noise) on the power supply output you want to measure is much smaller than this DC level. You want to zoom in on the AC noise and set the oscilloscope to a more sensitive range to see the details of the noise while continuing to maintain a low noise level on the oscilloscope (see the "Little Rules of Oscilloscope Noise" section in the right column). Depending on the oscilloscope and probe you are using, you may not be able to do this completely.

Tip 1. Choose the Oscilloscope Measurement Path with the Lowest Noise

Obviously, if you are measuring noise on a DC power supply, you want your oscilloscope measurement system to be as quiet as possible to avoid affecting your measurements. Unfortunately, this is where many users get lost, not knowing that there may be better options. The oscilloscope measurement path consists of the oscilloscope being used and the oscilloscope input terminals (50 Ω or 1 MΩ).

For many oscilloscopes, the 50 Ω input path has lower noise than the 1 MΩ path. Figure 1 below shows the baseline noise of the 50 Ω input and the 1 MΩ input of a Keysight DSOS054A High Definition Oscilloscope (500 MHz, 4 channels).

Figure 1. Comparison of baseline noise for 50 Ω and 1 MΩ oscilloscope inputs of the Keysight DSOS054A.

This type of measurement is often called a null measurement and is a measurement that determines the baseline noise of an oscilloscope's measurement system. This is a sanity check similar to shorting the leads on a DMM before making a continuity or resistance measurement. It is a good idea to perform a null measurement on your entire oscilloscope measurement system, including the probe, to ensure that the oscilloscope and probe are appropriate for the power supply noise measurements you are making. To make a null measurement, simply configure the oscilloscope and probe as you would for a power supply noise measurement, then short the inputs to ground (or short the inputs on a differential probe) and begin measuring noise.

Tip 2: Reduce the noise in your measurement system by limiting the bandwidth

Is higher bandwidth better? Not necessarily. Oscilloscope and probe noise voltage is frequency dependent. By limiting the bandwidth used to that required for a given measurement, we can reduce the oscilloscope and probe noise present in the measurement. Let’s take the measurements shown in Figure 2 as an example. In these measurements, we used a Keysight MSOS804A oscilloscope (8 GHz, 10-bit ADC, 20 GSa/s) and an N7020A power supply probe (2 GHz, 1:1 attenuation) to perform the nulling measurements mentioned earlier. The measurement results are summarized in Table 1.

Table 1. Null measurement noise results at different bandwidths

Table 1. Null measurement noise results at different bandwidths

Figure 2. Baseline noise of the N7020A power supply probe and S-Series oscilloscope at different bandwidth limits.

Tip 3. Use 1:1 attenuation to reduce noise in your measurement system

Oscilloscope probes come in different attenuation ratios. You are probably most familiar with 10:1 passive probes. One advantage of using a 10:1 probe is that it allows you to measure signals that exceed the maximum input of the oscilloscope. The downside of attenuation is that the greater the attenuation ratio, the greater the amplitude of the oscilloscope noise relative to the amplitude of the signal being measured. See the "Little Rules of Oscilloscope Noise" section in the right column for more information.