22 power supply R&D issues summarized by senior engineers

okhxyyo

22 power supply R&D issues summarized by senior engineers [Copy link]

Question 1: Why do we often choose 65K or 100K (near these frequency bands) as the switching frequency for the flyback power supply that we use most in low-power applications? What are the reasons for this restriction? Or in what cases can we increase the switching frequency? Or reduce the switching frequency?

Why do switching power supplies often choose a switching frequency of around 65K or 100K? Some people would say that IC manufacturers all produce such ICs, and of course there are reasons for this. What does the switching frequency of each power supply determine? We

should think about the reasons from here. Some people also say that EMC is not easy to pass when the frequency is high. Generally speaking, this is true, but this is not inevitable. EMC is related to frequency, but it is not inevitable. Imagine that our power supply switching frequency is increased, what is the direct impact? Of course, the MOS switch loss increases, because the number of switches per unit time increases.

What will happen if the frequency is reduced? The switching loss is reduced, but the energy provided by our energy storage device in a single cycle will increase, which will inevitably require a larger transformer magnetism and a larger energy storage inductor. Selecting around 65K to 100K is a more appropriate experience compromise, and the power supply is a compromise between the rationalization and compromise.

If in special circumstances, the input voltage is relatively low, the switching loss is already very small, and you don’t care about this switching loss, then we can increase the switching frequency to reduce the size of the magnetic device.

How to choose the switching frequency of the appropriate IC? Why is the switching frequency of mainstream ICs roughly in this range? What is the switching frequency related to? I am talking about the general situation. I don’t want to go into a rut. Many ICs have different frequencies. I want to diverge everyone’s thinking to pay attention to these issues! The

general situation I want to talk about here is mainly what the switching frequency is related to, how to choose the appropriate switching frequency, why there are so many mainstream ICs and switching frequencies, and it is not necessary, but a general situation, so that the novice area can understand the general behavior. Of course, the switching power supply can be used in any way it wants, and it must be used reasonably.

1. How do you know that 65 or 100KHZ is generally chosen as the switching frequency of the switching power supply? (After investigating the mainstream ICs of major manufacturers, these two are more common. Of course, there are some around this range, and some have adjustable switching frequencies)

2. How do you find out that the switching frequency of the switching power supply is indeed 65KHZ or 100KHZ during work? (From a design perspective, most power supplies use this range)

3. Are there more than two pictures of testing 65KHZ and 100KHZ frequencies? (More than two pictures, meaningless)

4. Do you know that the switching power supply can work at 1.5HZ. (You think it is necessary to talk about this, there is nothing wrong with work, you are just being picky, remember to be picky when doing technology, then can you talk about why the general power supply does not work at 1.5HZ, it is meaningful to say this, it is meaningless for you to make a 1.5HZ power supply)

Reminder: Technicians must remember to be picky, we are not campus research schools, we need to combine theory with practice, and the products we make are meaningful products!

Question 2: Why do we often design the switching frequency in zone 2 in LLC? Why not zone 1 and zone 3? What are the factors that restrict it? Or what are the consequences if zone 1 and zone 3 are selected as the switching frequency?

The principle of LLC is to use the increase of inductive reactance of inductive load as the switching frequency increases to adjust the output voltage, that is, PFM modulation. And the turn-on loss of MOS tube ZVS is smaller than ZCS, and zone 1 is the capacitive load zone, which is naturally not desirable. Then in the third zone, the switching frequency is greater than the resonant frequency, which is still the inductive load zone. In theory, there is no problem for MOS to achieve ZVS, which is true. However, we cannot ignore the turn-off of the output diode on the secondary side. That is, when the primary MOS tube is turned off, the resonant current does not decrease to be equal to the excitation current, and the secondary rectifier diode is softly turned off. This is also the reason why we usually do not choose the third zone.

We cannot just design according to the experience of our predecessors, but we must know that there is a reason for such design!

Question 3: What will happen when the duty cycle of our flyback is greater than 50%? What are the good aspects? What are the bad aspects?

What does it mean when the duty cycle of the flyback is greater than 50%, and what factors does the duty cycle affect ? First: If the duty cycle is too large, the first thing that will be caused is an increase in the turns ratio, and the stress of the main MOS tube will inevitably increase. Generally, MOS tubes below 600V or 650V are selected for flyback, considering the cost. A duty cycle that is too large is bound to be unbearable.

The second point: It is very important that many people know that slope compensation is required, otherwise the loop will oscillate. However, this is also conditional. The generation of the right plane zero point needs to work in CCM mode. If it is designed in DCM mode, this problem will not exist. This is also one of the reasons why low power is designed in DCM mode. Of course, we can overcome this problem by designing a sufficiently good loop compensation.

Of course, in special cases, it is also necessary to design the duty cycle to be greater than 50%. The energy transferred per unit cycle increases, which can reduce the switching frequency and achieve the purpose of improving efficiency. If the flyback is to increase efficiency, this method can be considered.

Question 4: If the flyback power supply wants to achieve a certain efficiency, what aspects should be considered? Quasi-resonance? Synchronous rectification?

A major disadvantage of flyback is the efficiency problem. What ways can be considered to improve efficiency? Reducing losses is inevitable. The loss points are switch tubes, transformers, and output rectifier tubes. These are the three main parts.

As for the switch tube, we know that flyback is mainly hard switching of PWM modulation. Switching loss is a major difficulty for us. Fortunately, the emergence of soft switching has seen hope. Flyback cannot achieve full resonance like LLC, so it can only develop towards quasi-resonance (resonance in part of the time period). There are many such ICs on the market. Our company uses NCP1207 more often. After the MOS tube is turned off, the VCC voltage is detected at the first foot before the next opening, and then the next cycle is opened after a set time.

How to minimize the loss of the transformer, the perfect use of the transformer will be involved in the following issues.

Synchronous rectification generally outputs large currents, and the secondary side rectifier diode, even if Schottky is used, will still lose a lot. At this time, synchronous rectification MOS is used to replace Schottky diodes. Some people will say that this is more expensive than LLC or forward. Of course, there is no best, only more suitable.

Question 5: How is the conduction of the power supply formed? What are the conduction paths? Commonly used methods? What things affect the radiation of the power supply? How to do high-power EMC.

The power supply conduction measurement method is to receive high-frequency interference (generally 150K to 30M) from the inside of the power supply through the input ports L, N, and PE.

To solve the conduction, it is necessary to figure out which ways to reduce the interference received by the port.

As shown in the figure: There are generally two modes: L, N differential mode components, and common mode components through the PE ground loop. Some frequencies have both differential and common modes.

Through filtering: Generally, secondary common mode is used with Y capacitors to filter. The selection method and skills are also very important, and the layout has a great impact. Generally, a low U inductor is placed near the port, preferably nickel-zinc material, specifically for high frequencies, and the winding method uses double wires in parallel to reduce differential mode components. The post-stage generally places a larger inductance, around 4MH to 10MH, which is just an empirical value, and it needs to be matched with the Y capacitor. The X capacitor also needs to be close to the port to filter the differential mode, and is generally placed in the middle of the secondary common mode. Place the Y capacitor, and the wiring needs to be thickened when the capacitor is laid out, and it cannot be externally mounted, otherwise the effect is very poor. (These are just a fuss on the input filter network)

Of course, you can also start from the source. Conduction is the result of radiation coupling into the line. Reducing the switch radiation can also bring benefits to conduction. Several places that affect radiation generally include the MOS tube opening speed, the rectifier tube conduction and shutdown, the transformer, and the PFC inductor, etc. The design of these circuits needs to compromise with other aspects and will not be elaborated in detail.

Some experience and skills: For high-power EMC, it is generally necessary to increase shielding, which is effective immediately. There are generally several options for shielding locations:

First: Shielding between the input EMI circuit and the switch tube, which has a great effect on EMC. Many filters that are ineffective generally use this method and are generally very effective.

Second: Shielding of the primary and secondary of the transformer. Generally, if there is space for the transformer design, it is best to add shielding.

Third: The location of the radiator can serve as a good shield. Reasonable layout and utilization, and the choice of radiator grounding are also very important.

Fourth: There are generally several simple methods to determine the location of the radiation source, which may not be completely accurate. For reference, if the input line magnetic ring is good for EMC, it is generally the primary MOS tube. If the output line magnetic ring is effective for EMC, it is generally the secondary output rectifier tube, especially the high frequency greater than 100M. You can consider adding capacitors or common-mode inductors to the output.

Of course, there are many other details and skills, especially in terms of layout loops, which will be explained separately for LAYOUT later.



Question 6: What factors do we need to consider when choosing a topology? What are the usage environments and advantages and disadvantages of various topologies?

I don't know what you think of the first step of designing a power supply? I think so. Carefully study the technical indicators of customers, convert them into the specifications of the power supply, and communicate the indicators with customers. Different indicators mean design difficulty and cost, which also has a great impact on the questions I raised. When choosing the topology, we consider the power supply indicators combined with the cost. What are the characteristics of the commonly used topologies?

Here we mainly talk about isolation. Non-isolated applications are limited, and of course, the cost is the lowest.

Flyback characteristics: Applicable to less than 150W, in theory, it is rarely used when it is greater than 75W, not to mention very special situations. Flyback is a bit low-cost and easy to debug (compared to half-bridge and full-bridge). It is mainly unidirectional excitation of the magnetic core, power is limited, and efficiency is not high, mainly due to hard switching, large leakage inductance and other reasons. The efficiency of the full voltage range (85V-264V) is generally below 80%, and it is easy to reach 80% for a single voltage.

Characteristics of forward: moderate power, can be used for small and medium power, power is generally below 200W, of course, can be used for very high power, but it is not often done, because forward and flyback are unidirectionally excited, and large magnetic core volume is required for high power. Of course, there are also two transformers connected in series and parallel. Note that only general situations are discussed to avoid misleading newcomers. Forward has some advantages, moderate cost, of course higher than flyback, and higher efficiency than flyback, especially using active clamping for primary side absorption to reuse leakage inductance energy.

Half-bridge: LLC resonant half-bridge is currently more popular, small and medium power, high power all-in-one type. (Generally greater than 100W and less than 3KW). Characteristics The cost is higher than the flyback forward, because one more MOS tube (bidirectional excitation) and one rectifier tube are used, the control IC is also expensive, and the loop design is complex (generally using op amps, especially current loops). Advantages: soft switching, good EMC, extremely high efficiency, higher than forward pulse. I have made 960W LLC, and the efficiency can reach more than 96% (full voltage) (of course, PFC adopts bridgeless mode). I don't recommend other half-bridges, at least I won't use them. It is difficult to achieve soft switching with the older asymmetric bridges. LLC was used more before it matured, but it is rarely used now. At least Emerson and other large companies tend to prefer LLC. It is generally not wrong to follow the mainstream.

Full bridge: generally used for more than 2KW, the first choice is phase-shifted full bridge, features, bidirectional excitation, small MOS tube stress, half of LLC stress. For high power, especially when the input voltage is high, a phase-shifted full bridge is generally used, and LLC is used for low input voltage. The cost is particularly high, and 2 more MOS are used than LLC. This is not the first priority. The main reason is that the drive is complex, and the general IC drive capability cannot reach it. The drive needs to be amplified and driven by an isolation transformer. This is the other side of the high cost.

Push-pull: It is used in high power, especially in high power occasions with low input voltage. It is characterized by high voltage stress, but of course low current stress. Whether full bridge or push-pull is used for high power generally depends on the input voltage. The transformer has one more winding, and the tube stress requirement is high. Of course, the magnetic bias mentioned often also needs to be overcome. I have never used this, and it does not involve power supply, so it is difficult to use it.

Question 7: When considering the cost of power supply, where should we start?

When designing power supply, cost evaluation is essential. At present, customers have suppressed the cost of power supply very low. All major competitors are fighting price wars. Everyone can make power supply. It depends on who can make it cheaper to win orders. From which aspects can we start to benefit Chen Ben:

First: technical indicators. The higher the technical indicators of the power supply, the higher the cost. If your power supply cost is high, then you can sell your performance indicators. If there are more performance requirements, the circuit will increase and the cost will naturally be higher. It is also the capital for talking with customers.

Second: material procurement cost. Why do large companies have high profits in power supply? It is nothing more than that they have a superior procurement platform, large procurement volume, low material cost, and of course lower cost. If purchasing is not considered, as an engineer, you must figure out the cost of different materials, such as using SMDs instead of plug-ins (for example, plug-in resistors are more expensive than SMDs), using domestic products instead of Taiwanese products, and using Taiwanese products instead of Japanese products. The price difference here is not cheap. (For example, Japanese capacitors are several times more expensive than domestic capacitors!!! Of course, there are also differences in quality;)

Third: important components that affect cost: transformers, inductors, MOS tubes, capacitors, optocouplers, diodes and other semiconductor devices, ICs, etc. The prices of transformers wound by different transformer manufacturers vary greatly. MOS tube stress and thermal resistance are sufficient, and the cost of IC solutions, etc.

Other aspects lead to cost issues: device heat sinks, the size is appropriate, and more is a waste of money. PCB layout, it is a waste of money to use a single-sided board as a double-sided board. PCB layout technology, choose a reasonable process with low processing cost and high production efficiency.

Question 8: Power supply loop design, which parts of the power supply affect the power supply loop? What indicators determine a good loop?

The loop design of power supply has always been a difficult point. Why? Because there are too many major influencing factors, theoretical calculations are difficult to be accurate, and simulations are also based on idealized models. Here we only talk about some influencing factors of loop design, and understand the loop from a qualitative perspective and how to perform loop compensation.

The loop is based on the need for feedback when the input and output fluctuate. The loop will correspondingly inform the control IC to adjust and maintain the stability of the output. The power supply loop is generally a series negative feedback, some are voltage series negative feedback (under CC mode), and some are current series negative feedback (under CV mode).

So what will affect the loop? Zeros and poles in the circuit. Zeros generally cause the gain to increase, causing a 90-degree phase shift (the right half plane zero will cause a -90-degree phase shift). Poles generally cause the gain to decrease, causing a -90-degree phase shift, and left half plane poles will cause system oscillation. Therefore, we need to use zero-pole compensation to reasonably regulate our loop. For the low-frequency part, in order to meet sufficient gain, zero compensation is generally introduced, and for high-frequency interference, pole compensation is generally introduced to offset and reduce high-frequency interference.

The principle of loop stability is: 1. At the crossover frequency (that is, the frequency when the gain is zero dB), the phase margin of the system is greater than 45 degrees.

2. When the phase reaches -180 degrees, the gain margin is greater than -12dB. 3. Avoid entering the crossover frequency too quickly. The slope of the curve near the crossover frequency is -1.

For general flyback circuits: 1. The output filter capacitor that generates zero points can increase the loop gain. (Generally around 4K at the intermediate frequency, it is good for the gain and does not require compensation)

2. If working in CCM mode, a right half plane zero point will also be generated. In the high frequency band, pole compensation can be used. This is generally difficult to compensate, so try to avoid it. Let the crossover frequency be less than the right half plane zero frequency (about 15K, which will change with the load). Select 3. The load will generate low-frequency poles. Use low-frequency zero points to compensate. 4. The LC filter will generate low-frequency poles, and zero point compensation is required. It is important to know which zero poles are good or bad, and compensate them in a targeted manner.

The compensation circuit is relatively simple for the power supply loop. Generally, type 2 compensation is used for op amps, and some will use type 3 compensation, which is rarely used.

Question 9: What are the soft switching forms for various topologies? How is soft switching achieved?

Soft switching is currently used frequently. First, it can improve sub-efficiency, and second, it can benefit EMC. Many topologies have begun to use soft switching. Even flyback has introduced quasi-resonance to achieve soft switching for high efficiency. This has been mentioned in the previous question. LLC's soft switch also mentioned the implementation conditions in the previous question, but the specific implementation process was not detailed. Here I will share my understanding of soft switching. Implementation conditions

and process: Using soft switching requires two elements, one is C and the other is L to achieve resonance (of course, it can also be multi-resonant). Resonance will produce a sine wave, and the sine wave can achieve zero crossing. If it is series resonance, it belongs to voltage resonance, and parallel resonance belongs to current resonance.

Secondly, the difference between soft switching and hard switching is: during the hard switching process, the voltage and current overlap, and the soft switch is either zero current (ZCS) or zero voltage (ZVS). The soft switch of MOS tube can use junction capacitance or parallel capacitance, and then connect inductor to achieve series ZVS, such as quasi-resonant flyback, active clamping absorption circuit, and soft switching towards full bridge. There is also LC parallel ZCS, but it is rarely used because the loss of MOS tube ZVS is less than ZCS. LLC belongs to the series-parallel type, but we use the ZVS area. (In the dead zone, the resonant current passes zero. Before the upper tube is soft-turned on, the lower tube junction capacitor is charged first, and the upper tube is soft-turned on)

Question 10: What kind of transformer is the most perfect and applicable? What does the transformer determine and what does it affect?

Designing a transformer is one of the core points of various topologies. The quality of transformer design affects all aspects of the power supply. Some transformers cannot work, some are inefficient, some are difficult to do EMC, some have high temperature rise, some will saturate in extreme cases, and some cannot pass safety regulations. It is necessary to comprehensively consider various factors to design the transformer.

Where should I start designing a transformer? Generally speaking, the size of the magnetic core is selected according to the power. Experienced people can refer to their own designs. Inexperienced people can only calculate according to the AP algorithm. Of course, a certain margin must be left, and finally the design is tested by experiment.

Generally, the recommended small-power flybacks are EE type, EF type, EI type, ER type, and medium-to-high-power PQ types. There are also individual habits and platform differences between different companies. For large power, there is no suitable magnetic core, so two transformers can be connected in series on the primary side and in parallel on the secondary side. Different topologies have

different requirements for transformers. For example, for flyback, you need to consider what mode you need to work in and how to adjust the inductance appropriately. Especially for multi-channel outputs, you must pay attention to the load adjustment rate to meet the requirements, and the coupling effect should be good, such as parallel winding, uniform winding, and as many secondary turns as possible. The MOS tube withstand voltage determines the turns ratio, how to select the appropriate duty cycle, and how large the Bmax is (generally less than 0.35, of course 0.3 is better, and even short circuit will not be too serious saturation). Some also need to add shielding to rectify EMC. The primary and secondary shields are generally added with 2 layers, and the outer shield is 1 layer.

High-power transformers generally pay more attention to losses. Copper loss and magnetic loss need to be balanced. Air cooling and natural cooling should also be considered. The current density is appropriate. The current density of slightly larger power (greater than 150W) is relatively small (3.5-4.5), and the current density of smaller power (5.0-7.0).

It is also necessary to know what safety regulations the power supply has passed, whether the retaining wall is sufficient, and whether the interlayer tape is set reasonably. It cannot be ignored. Once the certification is required, the transformer will be changed, which will also affect the progress.

Question 11: Do we really need to be obsessed with design tools and rely on simulation?

The design tools of power supply are mainly used in the following aspects: 1. Select the core and design the transformer 2. Loop simulation design 3. Main power topology simulation 4. Analog circuit simulation 5. Thermal simulation (for high power) 6. Calculation tools (calculation books) and so on.

For newcomers, my advice is to use fewer tools, more calculations, and grasp the design process yourself, because the tools are made by people, and different people have different design habits. You can't use a fixed design model to design different power supplies.

Some simulations can be combined with design: for example, it is difficult to directly meet the design requirements after the loop design is completed. Simulation can be well verified before the test, but the simulation is not completely the same as the test, at least not too far off.

It is also necessary to be proficient in using Mathcad and Saber, but we need to understand the principles of many things. We only need to use the tools as calculators to meet our design more quickly, conveniently and efficiently. It is undoubtedly a big misunderstanding to rely purely on tools to design power supplies.

Question 12: What are the places where you can find out the quality of a power board LAYOUT?

What kind of PCB is a good PCB? At least one of the following aspects must be met: 1. The electrical performance has little interference, the key signal lines and bottom lines are reasonably routed, and the performance in all aspects is stable (the premise is that the circuit is defect-free). 2. It is conducive to EMC, low radiation, and reasonable loop. 3. It meets safety regulations and the safety distance meets the requirements. 4. It meets the process, mass production producibility, and reduces production costs. 5. It is beautiful, the layout is regular and orderly (the device is not crooked), and the routing is beautiful and beautiful, not winding.

How to achieve the above points? Let me share my experience in layout:

1. Before layout, understand the power supply specification, power supply specifications, whether there are any special requirements, and the safety standards to be passed.

Whether the structural input conditions are accurate, as well as the confirmation of the air duct, the confirmation of the input and output ports, and the main power flow.

Select the process route according to the density of the device and whether there are special devices.

2. During layout,Pay attention to reasonable layout, ensure that the four major loops are as small as possible, and predict in advance whether the subsequent routing is easy to go. The placement of the transformer basically determines the overall layout, so it must be placed carefully and in the best position. The layout flow of the EMI part is clear, and there is a clear isolation zone with other main power parts. Reduce interference from the main power switching devices. The area of each absorption loop should be as small as possible, and the length and position of the radiator should be reasonable and not block the wind duct.

3. In the routing part, whether the routing of the input EMI circuit meets the safety regulations, the distance between the primary and secondary sides, and the distance between the input and output to the ground must meet the safety regulations. Whether the thickness of the traces can meet the current size, key signals (such as driving signals, sampling signals, whether the ground wire is reasonable), driving signals should not interfere with sensitive signals (high-frequency signals); whether the sampling signals are sampled accurately and whether they will be interfered; whether the ground wire is pulled reasonably (sometimes single-point grounding is required, sometimes multi-point grounding is required, which is related to actual needs), the main power ground and signal ground are strictly separated, the primary chip ground is taken from the sampling resistor, not from the large electrolytic (especially when the sampling resistor and the large electrolytic ground are far apart), the VCC ground is returned to the large electrolytic before the stage, the secondary capacitor ground is connected to the chip, the feedback signal is also connected to the IC at a single point, and the ground is connected to the IC at a single point. The ground of the radiator must be connected to the main power ground, not the signal ground, and many other detailed requirements.

Question 13: How much do you know about power supply components? How big is the junction capacitance of the MOS tube, and what does it affect? What is the relationship between RDS and temperature? What does Schottky reverse recovery current affect? What effects will the ESR of the capacitor bring?

There are many types of devices designed in the power supply, mainly semiconductor devices such as MOS tubes, triodes, ICs, op amps, diodes, optocouplers, etc.; magnetic devices: inductors, transformers, magnetic beads, etc.; capacitors: Y capacitors, X capacitors, ceramic capacitors, electrolytic capacitors, chip capacitors, etc.; each device has its specifications and limit parameters.

Conventional parameters are easy to grasp when we select them. For example, when selecting MOS tubes, the withstand voltage parameters will definitely be considered, the rated current will also be considered, and we will consider the on-resistance, but there are some parasitic parameters and some parameters that change with temperature, but they are rarely paid attention to, or they will only be found when problems are found. The on-resistance Rds(on) increases with the increase of temperature. When designing the loss of MOS tubes, the ambient temperature of its operation should be considered. The junction capacitance affects our turn-on loss and also affects EMC. The withstand voltage

and rated current of Schottky diodes are generally easy to pay attention to. Some parameters such as the on-voltage drop will decrease when the temperature rises, and the reverse recovery time is short, but the leakage current is large (especially considering that the leakage current has a greater impact at high temperatures), and the parasitic inductance will cause the turn-off spike to be very high.

ESR is an important parameter of capacitors. It is usually considered when calculating ripple. ESR is generally closely related to C, but the quality factors of different manufacturers also have a huge impact, so it must be clearly distinguished. For

general estimation companies, please refer to: ESR=10/(0.73 power of C). The life of capacitors will be shortened at high temperatures, the capacity will be reduced at low temperatures, the leakage current will also increase, etc.;

of course, the differences in the characteristics of devices under special circumstances are issues worth thinking about. Please think about it. It is very helpful for us to solve problems under special circumstances.

Question 14: How much do you know about magnetic materials? What are the differences between magnetic rings and magnetic cores ? In what situations are low magnetic rings and high magnetic rings used?

The importance of magnetic devices to switching power supplies is self-evident. It can be said that they are the heart of the power supply. There are many types of magnetic materials. Ferrite materials are generally used to make transformers. The main reason is that they are cheap and the maximum switching frequency can reach 1000K, which is enough for general use. Ferrite cores can be used as both main transformers and inductors, such as PFC inductors (generally made of iron silicon aluminum, with high cost performance), and energy storage inductors. Of course, in the case of high requirements, especially high power, magnetic rings are generally used. The main reason is that the inductance can be large and it is not easy to saturate. Compared with ferrite cores, the disadvantage is that the price is expensive, especially for large currents, and the winding process is more difficult. Magnetic rings are also divided into high U value and low U value, mainly due to the different materials of the magnetic rings. The appearance of high U ring magnetic rings is green. Generally, the common mode inductor of EMI circuits is selected. The inductance will be relatively large to filter low frequencies. The gray color is low U ring, the inductance is very low, and the high frequency is filtered. Generally, for EMC, the effect is generally better when used in combination!

Question 15: How is the power loss distributed? MOS tube loss? Transformer loss? In addition to DC loss, how is the AC loss of the transformer calculated?

Power loss is generally concentrated in the following aspects: 1. MOS tube turn-on loss and conduction loss. 2. Transformer copper loss; 3. Loss of the secondary rectifier tube; 4. Loss of the bridge rectifier. 5. Sampling resistor loss; 6. Absorption circuit loss; 7. Other losses: PFC inductor loss, LLC resonant inductor loss, synchronous rectifier MOS tube loss. And so on. . .

For these losses, appropriate reduction can improve efficiency. 1. For MOS tubes, you can choose fast switching speed and low on-resistance, and use soft switching in circuit class. 2. For transformers: choose a core of appropriate size. If the core is too small, the loss will be large, and it is difficult to balance copper loss and iron loss. In particular, copper loss not only has DC loss but also AC loss. AC loss is generally 2 times larger than DC loss, because the AC impedance of copper wire at high frequency is much larger than DC impedance. It must be fully estimated when calculating.

Question 16: Thermal design in power supply, how to choose the heat sink? What needs to be considered in heat sink design?

The design of the heat sink is a key point of the switching power supply. The heat sink is mainly for our heating device with too high temperature rise. The heat sink needs to be used to reduce the thermal resistance to achieve the effect of reducing the temperature rise!

Main heating devices: rectifier bridge, MOS tube, rectifier diode, transformer, inductor, etc.

The size of the radiator is generally selected based on the power loss and the required temperature rise to calculate the thermal resistance, and the radiator of the corresponding area is selected based on the thermal resistance.

Of course, some auxiliary methods are also needed, such as applying thermal paste between the device and the heat sink, which will have some effect. For relatively small spaces, profile heat dissipation can be used, which has a small volume and a large heat dissipation area.

Special devices have special treatments: For example, the PCB board under the transformer can be hollowed out for heat dissipation, or a heat sink can be attached to the transformer with thermal conductive mud. The inductor can also be cooled by adding a copper ring, etc. . .

Question 17: How is the output filter capacitor of LLC determined? What factors affect it?

The output filter capacitor is crucial to the output ripple. Choosing a suitable filter capacitor requires consideration of cost and ripple requirements. Of course, the selection of each topology filter capacitor is based on the output ripple requirement and the ESR value corresponding to the ripple current to select the corresponding capacitor. Of course, the relationship between the capacitance and ESR of the capacitor is also very important to the quality of the capacitor. The relationship has been discussed before. When the ripple voltage is our demand, if it is generally based on the demand of 50mv, the design reserves a margin and generally chooses 10mv. (Taking into account the filtering effect of the PCB board, the low temperature capacitance of the capacitor is reduced), the ripple current calculation formula is as follows:

Question 18: How is the drive of the phase-shifted full bridge realized? What is phase shifting? What does phase shifting bring? The phase-

shifted full bridge is currently very popular in medium and high power applications, and its popularity is second only to the LLC resonant half bridge. The usage of different topologies has been compared before, and here we will specifically introduce the characteristics of the phase-shifted full-bridge.

The first characteristic of the phase-shifted full-bridge: The drive is relatively complex, resulting in a complex control circuit and high cost. The reason is that the phase-shifted full-bridge generally has 4 MOS, which requires a high driving capability, which is difficult for general ICs to achieve. The driving capability needs to be amplified by an external MOS tube, and in order to enhance reliability, an isolation transformer is generally used to drive the MOS tube.

The second characteristic of the phase-shifted full-bridge: phase shifting, why phase shifting is needed, what does phase shifting bring, and what is the difference from an ordinary full bridge. Phase shifting is aimed at the same group of MOS tubes, allowing two MOS tubes to be turned on in sequence, which can reduce switching losses. When the leading arm bridge realizes ZVS, the secondary side is in freewheeling, the primary side current is shared by the diode, the MOS tube current is also very small, and it is approximately zero current conduction. The lagging arm bridge can be turned on at zero voltage.

The third characteristic of the phase-shifted full-bridge: the working process is complex, two output power states (energy is provided by the primary side), two freewheeling states (energy is provided by the secondary side inductor and capacitor), and four dead zones (to realize the soft opening of each MOS tube I)

are just to let novices understand the phase-shifted full-bridge. As a relatively important topology of the switching power supply, what are its key points and difficulties.

Question 19: If high power pursues efficiency, how is bridgeless PFC achieved? What is the principle?

Many people have heard of bridgeless PFC, but it is not very common in actual use. The reason is that bridgeless PFC has improved efficiency compared to ordinary bridge PFC, generally only 1-2%. If it is not for the pursuit of high efficiency, it is generally not used because the cost is too high. According to the characteristics of bridgeless PFC, the rectifier bridge is not really omitted, but is used as the isolation of the positive and negative half axes of the AC input. In simple terms, it is equivalent to two ordinary PFCs, one for each positive and negative half axis of the AC, and the corresponding PFC inductor will also increase by one, the MOS tube will also increase by one, and the driver IC will also be more complicated. For high power, in order to achieve high efficiency, the detection resistor is made of transformer windings to reduce losses. I have come across a 960W bridgeless PFC+LLC with an efficiency of 96.5%, but in the end, the customer required that both AC and DC input voltages be met, and the bridgeless PFC could not work well under DC, so it was rejected.

Question 20: Why is three-phase electricity used in power supplies? How is three-phase three-level achieved, and what does three-level bring?

Three-phase electricity is used more frequently in power supplies, generally in high-power situations above 1KW or tens of thousands of W. Three-phase electricity generally uses three-phase four-wires, one of which is the neutral wire, and the four wires are equivalent to being able to transmit three times the power of ordinary two-phase electricity. The greater transmission power is its biggest advantage; secondly, three-phase electricity is easy to generate, and the most common three-phase asynchronous motor can be generated simply and conveniently.
What is the three-phase three-level? Because three-phase electricity cannot directly power some electrical equipment, it needs to be converted into ordinary two-phase electricity. In the general process, three-phase PFC is used to convert to DC power, and DC power is then inverted into two-phase AC power. This involves three-level technology. The three-phase PFC rectifier does not produce ordinary positive and negative DC, but three levels, that is, positive DC, zero, and negative DC. It can also be seen from here that the stress of using three-level devices is reduced, the harmonic content is low, and the switch loss is also low, so the advantages are very prominent in high-voltage and high-power occasions.

Question 21: There are many protection circuits in the power supply. How many protections can you say at most? How to achieve it?

The reliability of the power supply is inseparable from the protection circuit. What are the common protection circuits?

1. Input undervoltage and overvoltage are very common, sampling the AC signal.

2. Output overvoltage protection, once the power switch can lock the machine, it is also helpful for the reliability of the power supply.

3. Overcurrent protection, some use constant current to overcurrent, some use power limiting to overcurrent, of course, it can also lock the machine to do it, the purpose is reliability, and there are many methods. The most reliable protection must be locked rather than hiccup!

4. Overtemperature protection, using thermistor to detect the transformer or ambient temperature, etc., to feedback to the IC to lock the machine or hiccup.

5. Short circuit protection, short circuit can hiccup, and it can also lock the machine.

These are commonly used in general power supplies, and some can be said to be necessary protection circuits. So look at the specification and choose the right IC to make the protection function more convenient. I used an LD7522 for flyback, and these functions are very good, and all can be made simply.

Question 22: Do you not understand the market when doing power supply? Where will your power supply go? Is it useless after development? It is useful only when it makes money for the boss.

Finally, it comes to the last question. The power supply market problem may not be paid much attention by general engineers, and it is wrong to focus on research and development. The success of a project is not to make it, but to make less money.

For example: you worked hard on three projects in a year and made 1 million, and another person did one project in a year, which was much easier than doing three projects, and made 10 million in a year. Which one does the boss like?

Some people say that the project is not our choice, how do we know whether it makes money or not, but we must be familiar with the characteristics of the money-making project. What kind of power supply is more popular in the market, do you know? According to the existing model of your own company, is there a gap in design with that of large companies? It is not about whether the project can be completed, but whether it can be completed in the best possible way. In fact, from the perspective of R&D, it is about how to choose the optimal topology and make a cost-saving solution.

annysky2012

Damn, this sums it up so well.

mao45

This knowledge point is exactly what I was looking for. Thank you for sharing. Thank you.

PowerAnts

Does “senior engineer” mean “highly skilled” or “long experience”? Does long experience mean good skills?

The first two questions are just bullshit, I don't want to read the rest

Question 1: Why do we often choose 65K or 100K (near these frequency bands) as the switching frequency for the flyback power supply that we use most in low-power applications? What are the reasons for this?

Correct answer: The standard frequency for the conduction test of the switching power supply is 150KHz-30M. Since the filtering cost of low frequency is relatively high, the fixed frequency PWMIC should avoid its fundamental wave and harmonics falling near 150KHz. The second harmonic of 65K is 130K, which is not picked up by the instrument, while the third harmonic is 195K, and the filtering cost is more cost-effective than 150K. Similarly, 100~120KHZ is a higher frequency choice.

Question 2: Why do we often design the switching frequency in zone 2 in LLC? Why not zone 1 and zone 3? What are the limiting factors? Or what are the consequences if zone 1 and zone 3 are selected as the switching frequency?
The author's answer mentioned: The principle of LLC is to adjust the output voltage by using the increase of inductive reactance of inductive load as the switching frequency increases, which is PFM modulation.

Obviously, this is also bullshit! PFM is variable frequency, but the pulse width remains unchanged; LLC is variable frequency, but the duty cycle is maximized, and the pulse width will change;

I don’t want to watch the rest!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!