The generation and measurement of switching power supply ripple

With the switch of SWITCH, the current in the Inductor L fluctuates up and down the effective value of the output current. Therefore, a ripple with the same frequency as SWITCH will also appear at the output end, and the commonly referred to as ripple refers to this. It is related to the capacity of the output Capacitor and ESR. The frequency of this ripple is the same as the switching power supply, ranging from tens to hundreds of KHz.

I am working ON a switching power supply very recently, and the output ripple is a more troublesome aspect, so take this opportunity to summarize a little. I found some information and sorted it out, including the generation, measurement and suppression of ripple.

The generation of switching power supply ripple

Our ultimate goal is to reduce the output ripple to a tolerable level. The fundamental solution to this goal is to avoid the generation of ripple as much as possible. First of all, we must know the type and reason of the ripple of the switching power supply.

With the switch of SWITCH, the current in the Inductor L fluctuates up and down the effective value of the output current. Therefore, a ripple with the same frequency as SWITCH will also appear at the output end, and the commonly referred to as ripple refers to this. It is related to the capacity of the output capacitor and ESR. The frequency of this ripple is the same as the switching power supply, ranging from tens to hundreds of KHz.

In addition, SWITCH generally uses bipolar transistors or mosfets, no matter which is, there will be a rise time and a fall time when it is turned on and off. At this time, a noise with the same frequency as the SWITCH rise and fall time or an odd multiple of the frequency will appear in the circuit, generally tens of MHz. Similarly, at the moment of reverse recovery of diode D, its equivalent circuit is the series connection of resistance, capacitance and inductance, which will cause resonance, and the generated noise frequency is also tens of MHz. These two kinds of noise are generally called high-frequency noise, and the amplitude is usually much larger than the ripple.

If it is an AC/DC converter, in addition to the above two kinds of ripple (noise), there is also AC noise. The frequency is the frequency of the input AC power supply, which is about 50-60 Hz. There is also a kind of common mode noise, which is caused by the equivalent capacitance generated by many switching power supplies that use the shell as a heat sink. Because I am doing research and development, I have less contact with the latter two noises, so I will not consider it for the time being.

Switching power supply ripple measurement

Basic requirements: use oscilloscope AC coupling, 20MHz bandwidth limit, unplug the ground wire of the probe

1. AC coupling is to remove the superimposed DC voltage to get an accurate waveform.

2. Turning on the 20MHz bandwidth limit is to prevent the interference of high-frequency noise and prevent the wrong results from being measured. Because the high frequency component has a large amplitude, it should be removed when measuring.

3. Unplug the ground clip of the oscilloscope probe and use a ground ring to measure, in order to reduce interference. Many departments do not have a ground ring. If the error allows, use the probe’s ground clamp to measure directly. But this factor should be considered when judging whether it is qualified.

Another point is to use a 50Ω terminal. According to the Yokogawa oscilloscope data, the 50Ω module removes the DC component and measures the AC component. However, few oscilloscopes are equipped with this kind of special probe. In most cases, the standard probes of 100KΩ to 10MΩ are used for measurement, and the impact is temporarily unclear.

The above are the basic precautions when measuring switching ripple. If the oscilloscope probe is not in direct contact with the output point, it should be measured with a twisted pair or 50Ω coaxial cable.

When measuring high frequency noise, use the full passband of the oscilloscope, which is generally in the order of hundreds of megabytes to GHz. Others are the same as above. Different companies may have different testing methods. In the final analysis, you must be clear about your test results. The second is to get customer approval.

About oscilloscope:

Some digital oscilloscopes cannot accurately measure ripple due to interference and memory depth. At this time, the oscilloscope should be replaced. In this regard, although the bandwidth of the old analog oscilloscope is only tens of megabytes, the performance is better than that of the digital oscilloscope.

Tektronix has special software for measuring the above two kinds of ripple (noise) separately, please refer to Reference 5. Similarly, you can also take a look at the relevant knowledge about the grounding of the oscilloscope and the power supply test.

Suppression of switching power supply ripple

For switching ripple, there must be both theoretically and practically. There are usually three ways to suppress or reduce it:

1. Increase inductance and output capacitor filtering

According to the formula of the switching power supply, the current fluctuation in the Inductor is inversely proportional to the inductance value, and the output ripple is inversely proportional to the output capacitance value. Therefore, increasing the inductance value and the output capacitance value can reduce the ripple.

Similarly, the relationship between output ripple and output capacitance: vripple=Imax/(Co?f). It can be seen that increasing the value of the output capacitor can reduce the ripple.

The usual practice is to use aluminum electrolytic Capacitors for the output capacitors in order to achieve the purpose of large capacity. However, electrolytic capacitors are not very effective in suppressing high-frequency noise, and the ESR is relatively large, so a ceramic capacitor will be connected in parallel next to it to make up for the lack of aluminum electrolytic capacitors.

At the same time, when the switching power supply is working, the Voltage Vin at the input terminal does not change, but the current changes with the switch. At this time, the input power supply will not provide current well. Usually, it is close to the current input terminal (in the case of BucK type, near SWITcH), and parallel capacitors are used to provide current.

The above method has limited effect on reducing ripple. Because of the volume limitation, the inductance will not be too large; the output capacitance is increased to a certain extent, there is no obvious effect on reducing the ripple; increasing the switching frequency will increase the switching loss. Therefore, this method is not very good when the requirements are relatively strict. For the principle of switching power supply, you can refer to various switching power supply design manuals.

2. Two-stage filtering is to add yi-stage LC filter

The LC filter has a more obvious inhibitory effect on the noise ripple. According to the ripple frequency to be removed, a suitable inductor and capacitor are selected to form a filter circuit, which can generally reduce the ripple well.

If the sampling point is selected before the LC filter (Pa), the output voltage will decrease. Because any inductor has a DC resistance, when there is a current output, there will be a voltage drop across the inductor, resulting in a decrease in the output voltage of the power supply. And this voltage drop varies with the output current.

The sampling point is selected after the LC filter (Pb), so that the output voltage is the voltage we hope to get. But this introduces an inductor and a capacitor inside the power system, which may cause the system to be unstable. Regarding the stability of the system, many materials have been introduced, so I won’t write them in detail here.

3. After switching power supply output, connect LDO filter

This is a very effective way to reduce ripple and noise. The output voltage is constant, and the original feedback system does not need to be changed, but it is also a very costly and high power consumption method. Any LDO has an indicator: noise rejection ratio. It is a frequency-dB curve, as shown on the right is the curve of Linear Technology’s LT3024.

To reduce ripple. The PCB layout of the switching power supply is also very critical, which is a very powerful problem. There are dedicated switching power supply PCB engineers. For high-frequency noise, due to the high frequency and large amplitude, although the post-filter has a certain effect, the effect is not obvious. There are special studies in this area, and the simple way is to parallel a capacitor C or RC on the diode, or a series inductor.

4. Parallel capacitor C or RC on the diode

Parasitic parameters should be considered when the diode is turned on and off at high speed. During the diode reverse recovery period, the equivalent inductance and equivalent capacitance become an RC oscillator, generating high-frequency oscillation. In order to suppress this high-frequency oscillation, a capacitor C or RC snubber network must be connected in parallel across the diode. The resistance is generally 10Ω-100 Ω, and the capacitance is 4.7pF-2.2nF.

The value of the capacitor C or RC connected in parallel on the diode can only be determined after repeated trials. If the selection is improper, it will cause more serious oscillation.

If the requirements for high-frequency noise are strict, soft switching technology can be used. There are many books dedicated to introducing soft switches.

5. Connect the inductor after the diode (filtering)

This is also a commonly used method to suppress high-frequency noise. In view of the frequency of the noise, selecting the appropriate inductance element can also effectively suppress the noise. It should be noted that the rated current of the inductor must meet the actual requirements. The simpler approach will not be explained in detail.

summary

The above is about the ripple of the switching power supply, some of the content is summarized, if you can add some waveforms, it would be better. Although it may not be complete, it is sufficient for general applications. Regarding noise suppression, not all applications are in practice. It is important to choose an appropriate method according to your own design requirements, such as product volume, cost, development cycle, etc.

I am working on a switching power supply very recently, and the output ripple is a more troublesome aspect, so take this opportunity to summarize a little. I found some information and sorted it out, including the generation, measurement and suppression of ripple.

The generation of switching power supply ripple

Our ultimate goal is to reduce the output ripple to a tolerable level. The fundamental solution to this goal is to avoid the generation of ripple as much as possible. First of all, we must know the type and reason of the ripple of the switching power supply.

With the switch of SWITCH, the current in the inductor L fluctuates up and down the effective value of the output current. Therefore, a ripple with the same frequency as SWITCH will also appear at the output end, and the commonly referred to as ripple refers to this. It is related to the capacity of the output capacitor and ESR. The frequency of this ripple is the same as the switching power supply, ranging from tens to hundreds of KHz.

In addition, SWITCH generally uses bipolar transistors or MOSFETs, no matter which is, there will be a rise time and a fall time when it is turned on and off. At this time, a noise with the same frequency as the SWITCH rise and fall time or an odd multiple of the frequency will appear in the circuit, generally tens of MHz. Similarly, at the moment of reverse recovery of diode D, its equivalent circuit is the series connection of resistance, capacitance and inductance, which will cause resonance, and the generated noise frequency is also tens of MHz. These two kinds of noise are generally called high-frequency noise, and the amplitude is usually much larger than the ripple.

If it is an AC/DC converter, in addition to the above two kinds of ripple (noise), there is also AC noise. The frequency is the frequency of the input AC power supply, which is about 50-60 Hz. There is also a kind of common mode noise, which is caused by the equivalent capacitance generated by many switching power supplies that use the shell as a heat sink. Because I am doing research and development, I have less contact with the latter two noises, so I will not consider it for the time being.

Switching power supply ripple measurement

Basic requirements: use oscilloscope AC coupling, 20MHz bandwidth limit, unplug the ground wire of the probe

1. AC coupling is to remove the superimposed DC voltage to get an accurate waveform.

2. Turning on the 20MHz bandwidth limit is to prevent the interference of high-frequency noise and prevent the wrong results from being measured. Because the high frequency component has a large amplitude, it should be removed when measuring.

3. Unplug the ground clip of the oscilloscope probe and use a ground ring to measure, in order to reduce interference. Many departments do not have a ground ring. If the error allows, use the probe’s ground clamp to measure directly. But this factor should be considered when judging whether it is qualified.

Another point is to use a 50Ω terminal. According to the Yokogawa oscilloscope data, the 50Ω module removes the DC component and measures the AC component. However, few oscilloscopes are equipped with this kind of special probe. In most cases, the standard probes of 100KΩ to 10MΩ are used for measurement, and the impact is temporarily unclear.

The above are the basic precautions when measuring switching ripple. If the oscilloscope probe is not in direct contact with the output point, it should be measured with a twisted pair or 50Ω coaxial cable.

When measuring high frequency noise, use the full passband of the oscilloscope, which is generally in the order of hundreds of megabytes to GHz. Others are the same as above. Different companies may have different testing methods. In the final analysis, you must be clear about your test results. The second is to get customer approval.

About oscilloscope:

Some digital oscilloscopes cannot accurately measure ripple due to interference and memory depth. At this time, the oscilloscope should be replaced. In this regard, although the bandwidth of the old analog oscilloscope is only tens of megabytes, the performance is better than that of the digital oscilloscope.

Tektronix has special software for measuring the above two kinds of ripple (noise) separately, please refer to Reference 5. Similarly, you can also take a look at the relevant knowledge about the grounding of the oscilloscope and the power supply test.

Suppression of switching power supply ripple

For switching ripple, there must be both theoretically and practically. There are usually three ways to suppress or reduce it:

1. Increase inductance and output capacitor filtering

According to the formula of the switching power supply, the current fluctuation in the inductor is inversely proportional to the inductance value, and the output ripple is inversely proportional to the output capacitance value. Therefore, increasing the inductance value and the output capacitance value can reduce the ripple.

Similarly, the relationship between output ripple and output capacitance: vripple=Imax/(Co?f). It can be seen that increasing the value of the output capacitor can reduce the ripple.

The usual practice is to use aluminum electrolytic capacitors for the output capacitors in order to achieve the purpose of large capacity. However, electrolytic capacitors are not very effective in suppressing high-frequency noise, and the ESR is relatively large, so a ceramic capacitor will be connected in parallel next to it to make up for the lack of aluminum electrolytic capacitors.

At the same time, when the switching power supply is working, the voltage Vin at the input terminal does not change, but the current changes with the switch. At this time, the input power supply will not provide current well. Usually, it is close to the current input terminal (in the case of BucK type, near SWITcH), and parallel capacitors are used to provide current.

The above method has limited effect on reducing ripple. Because of the volume limitation, the inductance will not be too large; the output capacitance is increased to a certain extent, there is no obvious effect on reducing the ripple; increasing the switching frequency will increase the switching loss. Therefore, this method is not very good when the requirements are relatively strict. For the principle of switching power supply, you can refer to various switching power supply design manuals.

2. Two-stage filtering is to add yi-stage LC filter

The LC filter has a more obvious inhibitory effect on the noise ripple. According to the ripple frequency to be removed, a suitable inductor and capacitor are selected to form a filter circuit, which can generally reduce the ripple well.

If the sampling point is selected before the LC filter (Pa), the output voltage will decrease. Because any inductor has a DC resistance, when there is a current output, there will be a voltage drop across the inductor, resulting in a decrease in the output voltage of the power supply. And this voltage drop varies with the output current.

The sampling point is selected after the LC filter (Pb), so that the output voltage is the voltage we hope to get. But this introduces an inductor and a capacitor inside the power system, which may cause the system to be unstable. Regarding the stability of the system, many materials have been introduced, so I won’t write them in detail here.

3. After switching power supply output, connect LDO filter

This is a very effective way to reduce ripple and noise. The output voltage is constant, and the original feedback system does not need to be changed, but it is also a very costly and high power consumption method. Any LDO has an indicator: noise rejection ratio. It is a frequency-dB curve, as shown on the right is the curve of Linear Technology’s LT3024.

To reduce ripple. The PCB layout of the switching power supply is also very critical, which is a very powerful problem. There are dedicated switching power supply PCB engineers. For high-frequency noise, due to the high frequency and large amplitude, although the post-filter has a certain effect, the effect is not obvious. There are special studies in this area, and the simple way is to parallel a capacitor C or RC on the diode, or a series inductor.

4. Parallel capacitor C or RC on the diode

Parasitic parameters should be considered when the diode is turned on and off at high speed. During the diode reverse recovery period, the equivalent inductance and equivalent capacitance become an RC oscillator, generating high-frequency oscillation. In order to suppress this high-frequency oscillation, a capacitor C or RC snubber network must be connected in parallel across the diode. The resistance is generally 10Ω-100 Ω, and the capacitance is 4.7pF-2.2nF.

The value of the capacitor C or RC connected in parallel on the diode can only be determined after repeated trials. If it is not selected properly, it will cause more serious oscillation.

If the requirements for high-frequency noise are strict, soft switching technology can be used. There are many books dedicated to introducing soft switches.

5. Connect the inductor after the diode (filtering)

This is also a commonly used method to suppress high-frequency noise. In view of the frequency of the noise, selecting the appropriate inductance element can also effectively suppress the noise. It should be noted that the rated current of the inductor must meet the actual requirements. The simpler approach will not be explained in detail.

summary

The above is about the ripple of the switching power supply, some of the content is summarized, if you can add some waveforms, it would be better. Although it may not be complete, it is sufficient for general applications. Regarding noise suppression, not all applications are in practice. It is important to choose an appropriate method according to your own design requirements, such as product volume, cost, development cycle, etc.

The Links:   CM900DU-24NF BSM50GP60