Inventory of classic problems in switching power supply design

Many engineers who have never used switching power supply design may have a certain fear of it, such as worrying about the interference of switching power supply, PCB layout, parameters and type selection of components, etc. In fact, as long as you understand it, it is very convenient to use switching power supply design.

A switching power supply generally consists of two parts: a switching power supply controller and an output. Some controllers integrate MOSFET into the chip, which makes it easier to use and simplifies the PCB design, but reduces the design flexibility.

The switch controller is basically a closed-loop feedback control system, so there is generally a sampling circuit for feedback output voltage and a control circuit for the feedback loop. Therefore, the design of this part is to ensure an accurate sampling circuit and to control the feedback depth, because if the feedback loop responds too slowly, it will have a great impact on the transient response capability.

The output design includes output capacitors, output inductors, and MOSFETs. The selection of these components is basically to balance performance and cost. For example, a high switching frequency allows the use of a small inductor value (which means a small package and low cost), but a high switching frequency will increase interference and switching losses on the MOSFET, thereby reducing efficiency. A low switching frequency will have the opposite effect.

The selection of the ESR of the output capacitor and the Rds_on parameter of the MOSFET is also very critical. A small ESR can reduce the output ripple, but the capacitor cost will increase, and good capacitors will be expensive. The driving capability of the switching power supply controller should also be paid attention to. Too many MOSFETs cannot be driven well.

Generally speaking, suppliers of switching power supply controllers will provide specific calculation formulas and usage plans for engineers to refer to.

How to debug a switching power supply circuit?

There are some experiences I can share with you:

(1) The output of the power supply circuit is connected to the board through a low-resistance, high-power resistor. In this way, the power supply circuit can be debugged first without soldering the resistor to avoid the influence of the subsequent circuit.

(2) Generally speaking, the switching controller is a closed-loop system. If the output deterioration exceeds the controllable range of the closed loop, the switching power supply will not work properly. Therefore, in this case, the feedback and sampling circuits need to be carefully checked. In particular, if a large ESR output capacitor is used, a lot of power ripple will be generated, which will also affect the operation of the switching power supply.

Why grounding?

The introduction of grounding technology was originally a protective measure to prevent electrical or electronic equipment from being struck by lightning. The purpose was to introduce the lightning current generated by lightning into the earth through lightning rods, thereby protecting buildings. At the same time, grounding is also an effective means to protect personal safety. When the phase line caused by some reason (such as poor insulation of wires, aging of lines, etc.) touches the equipment shell, a dangerous voltage will be generated on the equipment shell, and the generated fault current will flow through the PE line to the earth, thus playing a protective role. With the development of electronic communications and other digital fields, it is far from meeting the requirements to only consider lightning protection and safety in the grounding system. For example, in a communication system, the interconnection of signals between a large number of devices requires that each device must have a reference "ground" as the reference ground for the signal. Moreover, with the complexity of electronic equipment, the signal frequency is getting higher and higher. Therefore, in the grounding design, electromagnetic compatibility issues such as mutual interference between signals must be given special attention, otherwise, improper grounding will seriously affect the reliability and stability of the system operation. Recently, the concept of "ground" has also been introduced in the signal return technology of high-speed signals.

Definition of Grounding

In the modern concept of grounding, for line engineers, the term usually means "the reference point of the line voltage"; for system designers, it is often a cabinet or rack; for electrical engineers, it means the green safety ground wire or the ground. A more common definition is "grounding is a low-impedance path for current to return to its source." Note that the requirements are "low impedance" and "pathway."

Common ground symbols

PE, PGND, FG - protection ground or casing; BGND or DC-RETURN - DC -48V (+24V) power supply (battery) return; GND - working ground; DGND - digital ground; AGND - analog ground; LGND - lightning protection ground.

Proper grounding method

There are many ways to ground, including single-point grounding, multi-point grounding and mixed types of grounding. Single-point grounding is divided into series single-point grounding and parallel single-point grounding. Generally speaking, single-point grounding is used for simple circuits, grounding distinction between different functional modules, and low-frequency (f10MHz) circuits require multi-point grounding or multi-layer boards (complete ground plane layers).

Introduction to signal return and cross-split

For an electronic signal, it needs to find a path with the lowest impedance for the current to flow back to the ground, so how to deal with this signal return becomes very critical.

First, according to the formula, the radiation intensity is proportional to the loop area, that is, the longer the path the return current needs to take, the larger the loop formed, and the greater the interference to external radiation. Therefore, when laying out the PCB, the area of ​​the power supply loop and the signal loop should be reduced as much as possible.

Second, for a high-speed signal, providing good signal return can ensure its signal quality. This is because the characteristic impedance of the transmission line on the PCB is generally calculated with reference to the ground layer (or power layer). If there is a continuous ground plane near the high-speed line, the impedance of this line can remain continuous. If there is no ground reference near a section of the line, the impedance will change, and the discontinuous impedance will affect the integrity of the signal. Therefore, when routing, the high-speed line should be allocated to the layer close to the ground plane, or one or two ground lines should be run in parallel next to the high-speed line to provide shielding and nearby return.

Third, why do we say that when routing, try not to cross the power split? This is because after the signal crosses different power layers, its return path will be very long and easily interfered with. Of course, it is not strictly required not to cross the power split. For low-speed signals, it is okay because the interference generated can be ignored compared to the signal. For high-speed signals, you must carefully check and try not to cross it. You can adjust the routing of the power supply part.

Why should we separate analog ground and digital ground, and how to separate them?

Both analog and digital signals need to return to the ground, because digital signals change quickly, which will cause a lot of noise on the digital ground, and analog signals need a clean ground reference to work. If the analog and digital grounds are mixed together, the noise will affect the analog signal.

Generally speaking, analog ground and digital ground should be processed separately, and then connected together through thin traces, or connected at a single point. The general idea is to try to block the noise on the digital ground from reaching the analog ground. Of course, this is not a very strict requirement that the analog ground and digital ground must be separated. If the digital ground near the analog part is still very clean, they can be combined.

How to ground the signals on the board?

For general devices, it is best to ground them nearby. After adopting a multi-layer board design with a complete ground plane, it is very easy to ground general signals. The basic principle is to ensure the continuity of the routing, reduce the number of vias, and be close to the ground plane or power plane, etc.

How to ground the interface components of a single board?

Some boards have external input and output interfaces, such as serial port connectors, network port RJ45 connectors, etc. If their grounding design is not good, it will affect normal operation, such as bit errors and packet loss in network port interconnection, and will become an external electromagnetic interference source, sending the noise inside the board to the outside. Generally speaking, an independent interface ground will be separated out, and the connection with the signal ground is connected by a thin trace, and a 0 ohm or small resistance resistor can be connected in series. The thin trace can be used to block the noise on the signal ground from passing to the interface ground. Similarly, the filtering of the interface ground and interface power supply should also be carefully considered.

How to ground the shield of a cable with a shield?

The shielding layer of the shielded cable should be connected to the interface ground of the board instead of the signal ground. This is because there are various noises on the signal ground. If the shielding layer is connected to the signal ground, the noise voltage will drive the common mode current to interfere outward along the shielding layer. Therefore, poorly designed cables are generally the largest noise output source of electromagnetic interference. Of course, the premise is that the interface ground must also be very clean.