How to Measure Ripple & Noise in Power Supplies

Hello and welcome to CUI In The Lab. I'm Ron Stull, Power Supply Marketing Engineer at CUI, and today we'll be talking to you about power supply ripple and noise measurements. Specifically we'll be taking a look at probing techniques. We'll explore both good and bad techniques so that you can see the impact it has on your end results.

Before we get to testing, a little background. Ripple and noise is the AC waveform on top of the DC output voltage, and is typically measured in volts peak-to-peak. As the name implies, there are two components. The ripple is caused by the charging and discharging of storage elements in each switching cycle. For example in a flyback converter, the load current is supplied entirely by the output capacitors during one half of the cycle. During this time, the output voltage will fall as the capacitors discharge. In the other half of the cycle, energy stored in the transformer is transferred to the output, supplying both the load and recharging the output capacitors. During this time, the voltage will rise until the cycle repeats. Because of its relation to the switching period, the ripple will be of the same frequency as the switching converter, or some multiple of it, while the ripple is an inherent and predictable attribute of a power supply. The noise is caused by parasitics and the rapid rates of change of current that occur within the power supply.

Noise usually presents itself as high-frequency voltage spikes on the ripple waveform. The ripple and noise is typically very small compared to the DC output voltage - it's common for it to be specified at 1% or less of the nominal output. For this reason, any noise that gets coupled into the measurement can corrupt it. It's essential that proper probing techniques are used in order to obtain valid measurement results.

To perform this measurement, we will need the following equipment: we'll need our power supply that we'll be testing and it's datasheet. Today we'll be testing CUI's compact internal power supply: the VOF-15C, it's a 215 watt, 12 volt output, with a ripple and noise rating of 100 millivolts. Next, we have our AC source, which will provide power to our power supply, over electronic load to load it with a constant current source, and our oscilloscope to record the measurement. We also have a few scope probes to show the difference that probing technique can have on your measurement.

Before we perform this measurement, we need to take a look at the datasheet and see what conditions are specified for this test. So now we'll set up the scope and apply the 20 megahertz bandwidth limit. Now we'll add a 10 microfarad electrolytic and a 1 microfarad ceramic capacitor to the output, where we plan to probe. As mentioned earlier, how you probe the output can have a large impact on your results. If your probe connections create a large loop, it can pick up external noise. The larger the loop, the more noise that can get picked up, so for good probing, the number-one rule is to minimize the probe area. This is generally done first by avoiding the use of the ground clips and leads that come with the oscilloscope probes.

One acceptable probing technique that we call the paperclip method wraps a wire around the ground connection, with a lead coming out close to the tip to minimize the loop area. Another acceptable probing technique is the tip and barrel method, which exposes the ground barrel and makes the connection to ground directly rather than through an external lead. This minimizes the loop area even further. So now, let's take a look at an example of bad probing. Here we'll use the probe clips that came with the probe to make our connection. These are simple to use, but lead to a large loop area which will pick up noise.

We have another power supply here, which we'll use to show an example of the noise getting picked up when we turn it on. You can see that we haven't even turned our power supply under test on, and we're already picking up a significant amount of noise. And after we turn our power supply on, the noise remains, making it difficult to see our ripple and distorting our results. Now we'll add our paperclip probe. You can see that the loop area is significantly reduced with this method, and we're not seeing any significant external noise being picked up, even with our power supply nearby. Now we'll turn the power supply on so you can see the difference between the two methods.

Now we'll perform the test using the tip and barrel method. You can see that our loop area is slightly larger due to the location of the connectors. This method can be difficult as it's not always an easy place to access and place your probe across.

Now we'll take a look at all three at the same time. You can see that in the case where we've removed the probe leads, we're getting good results, with only minor differences between the two cases, but, in the case where we've left the probe lead, we're still picking up external noise. Sliding around the second power supply, you can see the external noise being picked up by the probe with the large loop. Our probe where we employed good probing methods is not picking up this noise and it's a good indication that what we're seeing is really the ripple and noise and nothing external to the power supply.

So to summarize, ripple and noise measurements can be very sensitive and in order to obtain accurate results, it's important to pay close attention to the conditions outlined in the datasheet and to the loop area created by the probe.

For more information on ripple and noise measurements, please read CUI's blog and white paper on this topic.