How to measure the noise in a switching power supply (SMPS)?

Noise on a switched mode power supply (SMPS) can sometimes become very bad. I’m evaluating the voltage noise on a simple low-cost switching power supply (SMPS) and it is almost down because of the poor reputation of these power supplies in terms of noise.

Noise on a switched mode power supply (SMPS) can sometimes become very bad. I’m evaluating the voltage noise on a simple low-cost switching power supply (SMPS) and it is almost down because of the poor reputation of these power supplies in terms of noise.

Output noise in switching regulators

In terms of its nature, the output of nSMPS will have some switching noise. After all, they are designed to use a pulse width modulation (or pulse frequency modulation) signal to switch the current from a higher DC power source, and then use a 2-pole LC filter to filter it.

The switching action of the MOSFET produces an alternating cycle, in which the first current flows into the Inductor, and then the inductor discharges. This leads to large dI/dt and large voltage spikes. We look forward to this noise. This is a question of how effective the LC filter is in preventing these large voltage spikes from being transmitted to the rest of the circuit.

The typical output voltage of the SMPS will show ripple at the switching frequency. An important indicator is how much ripple there is when there is no load on the regulator and when a typical load resistance is loaded in the application.

Measuring noise in switching power supplies

I recently had a low noise application and I wanted to try a very low cost 3.3 V SMPS; it only requires a load current of 50 mA. I have an evaluation board, I use a 5 V wall power supply to connect to the power supply and measure the output with a simple 10× probe. My measurement configuration is shown in Figure 1.

How to measure the noise in a switching power supply (SMPS)?

Figure 1. Use a 10x probe to measure the output voltage rail.

The DC level is good at 3.3 V. With the 12-bit resolution and large offset capability of my Teledyne LeCroy HDO 8108 oscilloscope, I can offset this voltage so that I can amplify the ripple noise and also look for slow DC drift. Figure 2 shows the measured voltage noise on a scale of 10 mV/div.

How to measure the noise in a switching power supply (SMPS)?

Figure 2. Measurement noise on the SMPS output using a 10× probe, with a scale of 10 mV/div.

The 20 microsecond period of the switcher-corresponding to a switching frequency of 50kHz-is obvious. The triangular pulse is expected from the charge and discharge cycle of the inductor current. However, in addition to this expected characteristic, there are two types of high-frequency noise. There is 10 mV peak-to-peak noise in the flat area, and the spike noise sometimes reaches 60 mV peak-to-peak.

The high frequency noise and sharp noise spikes are disturbing. This is not filtered out by the 2-pole LC filter. If I use this kind of power supply, how can I ensure that my circuit board can maintain sufficient functionality despite these noises?

However, it turns out that this noise is not actually voltage noise on the power supply output. In my detection, all radio frequencies are radio frequencies.

Distinguish between voltage noise and RF pickup

The large dI/dt passing through the inductor in the LC filter results in a large magnetic field generated near the SMPS. Any loop with a low-inductance path will generate a magnetically induced current, which generates a voltage that we measure with an oscilloscope.

I made a loop antenna with a 10x probe connected to the SMPS lead to pick up these spikes. Your first thought may be, but does the tip of the 10× probe have a 9MΩ resistance? Isn’t this a large impedance that prevents any AC current from being induced in the loop?

The tip has a 9MΩ resistor, but there is also a 10 pF parallel capacitor, which is part of the equalizer circuit through which high-frequency current flows. At 100 MHz, the impedance of a 10 pF capacitor is only 160Ω, which is very low.

In order to test whether these noises are really the idea of ​​the RF pickup in the probe rather than the actual noise on the power rail, I soldered a small SMA connector to the output of the circuit board to reduce the loop antenna area and radiation sensitivity area. In addition, I added another 10x probe near the SMPS output voltage, but used a second probe with the tip shorted to ground. This setup allows me to use a 10x probe to measure the output rail at the same time, measure the output rail through the SMA connector, and local RF noise (the probe is picked up, the tip is shorted to the ground ICfans). As shown in Figure 3.

How to measure the noise in a switching power supply (SMPS)?

Figure 3. Two 10× probes and a coaxial 1× connection are used to measure the voltage noise on the SMPS output.

How to measure the noise in a switching power supply (SMPS)?
Figure 4. The measured voltage on the SMPS output. All channels are within the same 10 mV/div range.

Probe attenuation affects SNR

There are two important observations. First, the general noise level of a 1× coaxial cable is much lower than that of a 10× probe. This is actually because the 10× probe is not a 10× probe, it is a 0.1× probe. It attenuates the signal by a factor of 10 and reduces its amplitude by 20 dB. When we measure small signal levels, such as tens of millivolts, the measured voltage is very sensitive to the amplifier noise of the oscilloscope.

Most oscilloscopes are smart enough to recognize that a 10× probe is attached to the channel. They will automatically adjust the displayed voltage scale to compensate for the ten-fold attenuation and Display the tip voltage. Therefore, when the oscilloscope displays a signal on a 10 mV/div scale, it actually uses a 1 mV/div scale on the amplifier. What we have seen is that the peak-to-peak noise of the tip reaches almost 10 mV. In fact, the peak-to-peak noise of the oscilloscope amplifier is about 1 mV.

The coaxial cable connected with SMA is actually a 1× probe. The trace is also displayed on a scale of 10 mV/div. In this case, the 1 mV peak-to-peak amplifier noise is more or less contained in the line width of the trace.

This demonstrates an important best measurement practice: when we observe low-amplitude signals, such as power rail noise, any 10-fold attenuation probe will reduce our SNR by 20 dB. When counting every dB, do not use attenuation probes.

Coaxial connection with oscilloscope probe

The second observation is that there are no large and sharp spikes in the coaxial connection, but in the two 10× probe measurements. Since one of the probes did not even touch the track output, this strongly indicates that the spike noise is caused by the RF pickup and not the voltage noise on the SMPS output.

This shows the second important best measurement practice: When measuring low-amplitude signals, use a measurement setup that is as close as possible to the coaxial connection to reduce the loop area of ​​the probe and its effectiveness as an antenna.

If we implement these two best measurement practices, we have 30 mV peak-to-peak ripple noise in the 3.3 V rail. This is 1% ripple, which is very suitable for low-cost SMPS. In addition, high-frequency noise is greatly reduced, and short-term transients—actually as RF pick-up noise rather than rail voltage noise—are no longer displayed as part of the switch’s output signal.

Frequency domain noise

As long as I use a ground plane close to my power supply and signal paths, this is an important best semiconductor design practice, so SMPS-powered equipment and signals on my board will only see the harmonic SMPS generated by 50 kHz.

Using a direct coaxial, low-noise connection, I measured the noise spectrum on the SMPS power rail. An example is shown in Figure 5.

How to measure the noise in a switching power supply (SMPS)?

Figure 5. The noise spectrum on the power rail. Top is a time-varying spectrogram, over 10 seconds, showing a very stable amplitude. In this ratio, 0 dBmV is 1 mV amplitude noise.

The peak in the spectrum is the 50 kHz harmonic of the switching frequency. The amplitude of the first harmonic is about 10 dBmV, or 3 mV. This is much smaller than the 30mV peak-to-peak voltage measured in the time domain. This is because ripple noise has such a low duty cycle. There are not too many sine waves in the short-term triangular pulses of the first harmonic. A large number of higher harmonics represent the strange shape of the waveform in the time domain and its high frequency content.

All switching noises are below the 10μV amplitude at approximately 3 MHz. For my application, this is an acceptable noise level, which is actually very low for this low-cost SMPS.

in conclusion

This article discusses important considerations about the voltage noise actually produced by a switching power supply, and introduces two best measurement methods that can help you make accurate oscilloscope measurements on the output rail of a switching regulator.

The Links:   PM400DVA060 HSD150SX84-D

Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *