What is ripple?
Since DC regulated power supplies are generally formed from AC power through rectification and voltage regulation, they inevitably contain some AC component in the DC stable quantity. This AC component superimposed on the DC stable quantity is called ripple. The composition of ripple is quite complex. Its form is generally a harmonic wave similar to a sine wave with a frequency higher than the power frequency, or a pulse wave with a very narrow width. The requirements for ripple vary depending on the application.
Ripple can be represented by effective value or peak value, and can be expressed as an absolute quantity or a relative quantity. For example, if a power supply is operating in a regulated state with an output of 100V 5A, and the measured effective value of the ripple is 10mV, this 10mV is the absolute quantity of the ripple. The relative quantity, i.e., the ripple factor = ripple voltage / output voltage = 10mV / 100V = 0.01%, which is equal to one ten-thousandth.
Why is your power supply ripple so high?
A user tested the ripple of the 5V output signal of their switching power supply using a 500MHz bandwidth oscilloscope and found that the peak-to-peak value of the ripple and noise reached over 900mV, while the nominal peak-to-peak value of the switching power supply's ripple was <20mV. Although the user's circuit board has an LDO further regulating the output of the switching power supply, the user believes that the measured result is too high and unreliable, and hopes to find the cause of the problem.
01
Problem Analysis
Excessive power supply ripple during testing is usually related to the probe used and the connection method of the front end. First, the user's probe connection method was checked, and it was found that a long alligator clip ground wire was used, as shown in the left image below, and the grounding point was clamped to the fixing screw of the single board, resulting in a relatively large ground loop. Since a large ground loop introduces more spatial electromagnetic radiation noise from the switching power supply and ground loop noise, it was replaced with a short grounding spring pin, as shown in the right image below.
Actual testing revealed a significant improvement in the peak-to-peak value of the ripple noise. However, the peak-to-peak value of the ripple noise still exceeded 40 mV, which is considerably different from the <20 mV claimed by the switching power supply manufacturer.
Further inspection revealed that the user was using the 10:1 passive probe that comes standard with the oscilloscope. (See the image below.)
A 10:1 probe attenuates the measured signal by a factor of 10 before sending it to the oscilloscope, which then amplifies the signal by a factor of 10. The advantage of this type of probe is that the matching circuitry increases the probe bandwidth to several hundred MHz and expands the oscilloscope's range. However, it's not particularly advantageous for measuring small signals. If the measured signal amplitude is already small, a 10-fold attenuation might cause it to be drowned out by the oscilloscope's noise floor, and even with a 10-fold amplification, the signal-to-noise ratio remains unchanged. Therefore, for measuring power supply ripple noise, a probe with a small attenuation ratio, such as a 1:1 probe, should be used whenever possible. So, a 1:1 passive probe was found. Although the bandwidth of a 1:1 passive probe is not high (usually tens of MHz), its small attenuation ratio makes it very suitable for small signal testing.
The image below shows a comparison of test results between a 1:1 passive probe and a 10:1 probe under different bandwidth limitations. As you can see, using the 1:1 probe with a 20MHz bandwidth limit results in a peak-to-peak ripple noise level of less than 10mV, significantly better than the 10:1 probe's results. The 1:1 probe's test results show a clear ripple waveform, meeting the user's expectation of power supply ripple noise <20mV. Furthermore, we can also see that bandwidth limitation has a certain effect on improving the peak-to-peak noise level.
02
Summary of problems
This is a typical power supply ripple testing problem. By using a short ground connection, replacing the probe with one having a low attenuation ratio, and implementing bandwidth limiting, we significantly improved the ripple noise test results. Generally, the factors affecting power supply ripple test results, in order of importance, are as follows:
1. Length of front-end connection wires and ground loops: Long ground loops will pick up more electromagnetic radiation and ground noise from the switching power supply, so the shortest possible ground connection should be used.
2. Probe attenuation ratio: Probes with a large attenuation ratio will make small signal amplitudes even weaker, or even drowned out by the oscilloscope's noise floor. Therefore, probes with an attenuation ratio of 1:1 should be used as much as possible.
3. Bandwidth Limitation: Many electromagnetic noises and oscilloscope noise floors are broadband. Setting an appropriate bandwidth limit can filter out additional noise. Many power supply ripple noise testing applications use a 20MHz bandwidth limit, while some chips require measurements up to 80MHz or 200MHz.
4. Measurement Range: Power supply ripple testing is usually performed at a small range (e.g., 10mV/division or 20mV/division). The larger the range, the higher the oscilloscope's noise floor. However, some oscilloscopes have limited bias range, and at small ranges, they may not be able to pull the measured DC voltage signal back to near the center of the screen for measurement. Therefore, the oscilloscope's AC coupling function is often used to isolate the DC voltage before performing ripple noise testing.
5. Input Impedance: Many oscilloscopes offer 50-ohm and 1M-ohm input impedance options. Generally, the oscilloscope has a lower noise floor at 50-ohm input impedance. However, oscilloscopes automatically switch to 1M-ohm impedance when connected to most passive probes. The 50-ohm input impedance can only be set when connecting active probes or coaxial cables.
Before conducting actual testing, it's a good practice to check the system noise floor under the current equipment and settings. The five waveforms in the image below show the noise floor results using a 500MHz S-series oscilloscope with different probe and bandwidth settings. From top to bottom, the waveforms are: 50 ohms input impedance, 1:1 probe, 500MHz bandwidth; 1M ohms input impedance, 1:1 probe, 20MHz bandwidth; 1M ohms input impedance, 1:1 probe, 500MHz bandwidth; 1M ohms input impedance, 10:1 probe, 20MHz bandwidth; 1M ohms input impedance, 10:1 probe, 500MHz bandwidth. The peak-to-peak noise floor ranges from less than 1mV to nearly 30mV, highlighting the importance of probe, bandwidth, and input impedance settings during testing.
If a suitable low-attenuation probe is unavailable, a 50-ohm coaxial cable can be used to create a custom probe as follows: Connect one end of the cable to an oscilloscope, set to a 50-ohm input impedance; strip the other end of the cable, solder the shield to the ground of the circuit under test, and connect the center conductor to the power signal being measured via a DC blocking capacitor. The advantages of this method are low cost and low attenuation ratio, but the disadvantages are poor consistency and difficulty in controlling the DC blocking capacitor parameters and bandwidth. In addition, in recent years, oscilloscope manufacturers have introduced probes specifically designed for power supply ripple testing. These probes combine advantages such as low attenuation ratio (1.1:1), high bandwidth (hardware 2GHz, bandwidth limit can be set in software), impedance matching that balances measurement needs and noise (the probe itself has a 50kΩ DC input impedance, but the oscilloscope output is a 50Ω input impedance spectrum), short ground wire (providing a soldering front end with very low loop inductance), large bias range (up to ±24V), and the ability to simultaneously test ripple and DC voltage. These probes are suitable for users with high requirements for power supply ripple measurement.
