In electronics theory, when electric current flows through a conductor, a magnetic field is formed around the conductor; when alternating current passes through a conductor, an alternating electromagnetic field is formed around the conductor, which is called an electromagnetic wave.
When the frequency of electromagnetic waves is lower than 100khz, they will be absorbed by the ground and cannot form effective transmission. However, when the frequency of electromagnetic waves is higher than 100khz, they can propagate in the air and be reflected by the ionosphere at the outer edge of the atmosphere, forming long-distance transmission capability. Therefore, alternating current that changes less than 1000 times per second is called low-frequency current, and alternating current that changes more than 1000 times per second is called high-frequency current, and radio frequency is such a high-frequency current. Radio Frequency is referred to as RF.
RF circuits are composed of passive components, active devices, and passive networks. The frequency characteristics of components used in RF circuits are different from those in low-frequency circuits. In addition to the different frequency characteristics of components from low-frequency circuits, the characteristics of RF circuits in the field of electronic technology are also different from those of low-frequency circuits. Under high-frequency conditions, stray capacitance and stray inductance have a great impact on the circuit. In low-frequency circuits, these stray parameters have little effect on the performance of the circuit. As the frequency increases, the impact of stray parameters becomes greater and greater. In the tuners of early VHF-band television receivers and the front-end circuits of communication receivers, the influence of stray capacitance was so great that it was no longer necessary to add additional capacitors.
In addition, the circuit has a skin effect under RF conditions. Unlike DC, where the current flows throughout the conductor, the current flows on the surface of the conductor under high frequency conditions. As a result, the AC resistance at high frequencies is greater than the DC resistance.
Another problem in high-frequency circuits is the electromagnetic radiation effect. As the frequency increases, when the wavelength is comparable to the circuit size 12, the circuit will become a radiator. At this time, various coupling effects will occur between circuits and between circuits and the external environment, thus leading to many interference problems.
RF PCB design, like electromagnetic interference (EMI) issues, has always been the most difficult part for engineers to control. Although there are still many uncertainties in RF PCB design, there are still certain rules to follow in RF PCB design. The following will discuss various issues related to RF PCB partition design.
Five major experience conclusions
1. RF circuit layout principles
When designing RF layout, the following general principles must be met first:
(1) Isolate the high-power RF amplifier (HPA) and low-noise amplifier (LNA) as much as possible. In simple terms, keep the high-power RF transmitting circuit away from the low-power RF receiving circuit;
(2) Ensure that there is at least one whole ground in the high-power area of the PCB board, preferably without vias. Of course, the larger the copper foil area, the better.
(3) Chip and power supply decoupling is also extremely important;
(4) RF output usually needs to be far away from RF input;
(5) Sensitive analog signals should be kept as far away from high-speed digital signals and RF signals as possible;
2. Physical partition, electrical partition, and design partition
Physical partitioning mainly involves issues such as component layout, orientation and shielding; electrical partitioning can be further decomposed into partitions for power distribution, RF routing, sensitive circuits and signals, and grounding.
1. Physical partition problem
The quality of RF design can be seen from the layout of components. The most effective technique is to first fix the components located on the RF path and adjust their orientation to minimize the length of the RF path, keep the input away from the output, and separate high-power circuits and low-power circuits as far as possible.
The most effective circuit board stacking method is to arrange the main ground plane (main ground) on the second layer below the surface layer, and run the RF line on the surface layer as much as possible. Minimizing the size of the vias on the RF path can not only reduce the path inductance, but also reduce the number of cold solder joints on the main ground and reduce the chance of RF energy leaking to other areas within the stacked board.
RF and IF traces should be crossed as much as possible, and a ground plane should be placed between them as much as possible. The correct RF path is very important for the performance of the entire PCB board, which is why component layout usually takes up most of the time in mobile phone PCB board design. In mobile phone PCB board design, the low noise amplifier circuit can usually be placed on one side of the PCB board, and the high power amplifier on the other side, and finally connected to the antenna on the RF end and the baseband processor end on the same side through a duplexer. Some skills are needed to ensure that the straight through hole does not transfer RF energy from one side of the board to the other side. The commonly used technique is to use blind holes on both sides. The adverse effects of the straight through hole can be minimized by arranging the straight through hole in an area on both sides of the PCB board that is not affected by RF interference.
Sometimes it is not possible to ensure sufficient isolation between multiple circuit blocks. In this case, it is necessary to consider using a metal shield to shield the RF energy in the RF area. The metal shield must be soldered to the ground and must be kept at an appropriate distance from the components, so it takes up valuable space on the PCB board. It is very important to ensure the integrity of the shield as much as possible. The digital signal line entering the metal shield should be routed on the inner layer as much as possible, and it is best that the PCB layer below the routing layer is the ground layer. The RF signal line can go out from the small gap at the bottom of the metal shield and the wiring layer at the ground gap, but as much ground as possible should be laid around the gap, and the ground on different layers can be connected together through multiple vias.
3. Chip and power decoupling
Many RF chips with integrated linear circuits are very sensitive to power supply noise. Usually, each chip needs to use up to four capacitors and an isolation inductor to ensure that all power supply noise is filtered out. An integrated circuit or amplifier often has an open-drain output, so a pull-up inductor is required to provide a high-impedance RF load and a low-impedance DC power supply. The same principle also applies to decoupling the power supply at the inductor end.
Some chips require multiple power supplies to work, so you may need two or three sets of capacitors and inductors to decouple them separately. Inductors are rarely placed in parallel because this will form an air-core transformer and induce interference signals with each other. Therefore, the distance between them should be at least equivalent to the height of one of the devices, or they should be arranged at right angles to minimize their mutual inductance.
4. Electrical zoning principles
The principles of electrical partitioning are largely the same as physical partitioning, but there are some additional factors. Some parts of the phone use different operating voltages and are controlled by software to extend the battery life. This means that the phone needs to run on multiple power supplies, which brings more problems to isolation.
The power supply is usually introduced from the connector and immediately decoupled to filter out any noise from outside the circuit board, and then distributed after passing through a set of switches or regulators. The DC current of most circuits on the mobile phone PCB is quite small, so the trace width is usually not a problem. However, a separate high-current line as wide as possible must be run for the power supply of the high-power amplifier to minimize the transmission voltage drop. In order to avoid too much current loss, multiple vias are required to transfer current from one layer to another. In addition, if the power pin of the high-power amplifier is not adequately decoupled, the high-power noise will radiate to the entire board and cause various problems.
Grounding of high power amplifiers is critical and often requires a metal shield. In most cases, it is also critical to keep the RF output away from the RF input. This also applies to amplifiers, buffers, and filters. In the worst case, amplifiers and buffers have the potential to self-oscillate if their outputs are fed back to their inputs with the proper phase and amplitude. In the best case, they will be stable over all temperature and voltage conditions.
In fact, they may become unstable and add noise and intermodulation signals to the RF signal. If the RF signal line has to loop back from the input of the filter to the output, this may seriously damage the bandpass characteristics of the filter. In order to obtain good isolation between the input and output, firstly, a circle of ground must be laid around the filter, and secondly, a ground must be laid in the lower area of the filter and connected to the main ground around the filter. It is also a good idea to keep the signal line that needs to pass through the filter as far away from the filter pins as possible.
In addition, the grounding of all parts of the board must be done very carefully, otherwise a coupling channel will be introduced. Sometimes you can choose to run single-ended or balanced RF signal lines. The principles of cross-interference and EMC/EMI also apply here. Balanced RF signal lines can reduce noise and cross-interference if they are routed correctly, but their impedance is usually higher, and a reasonable line width must be maintained to obtain an impedance that matches the signal source, routing, and load. The actual wiring may be somewhat difficult. Buffers can be used to improve isolation because they can divide the same signal into two parts and use them to drive different circuits. In particular, the local oscillator may require a buffer to drive multiple mixers.
When a mixer reaches common-mode isolation at RF frequencies, it will not work properly. Buffers can isolate impedance changes at different frequencies very well so that circuits do not interfere with each other. Buffers are very helpful in designs because they can be placed right after the circuits that need to be driven, making high-power output traces very short. Since the input signal level of the buffer is relatively low, they are not likely to interfere with other circuits on the board. Voltage-controlled oscillators (VCOs) convert changing voltages into changing frequencies, a feature used for high-speed channel switching, but they also convert tiny amounts of noise on the control voltage into tiny frequency changes, which adds noise to the RF signal.
5. Solve the noise problem
First, the desired bandwidth of the control line may range from DC to 2MHz, and it is almost impossible to remove such a wide bandwidth of noise through filtering; second, the VCO control line is usually part of a feedback loop that controls the frequency, which may introduce noise in many places, so the VCO control line must be handled very carefully. Make sure that the ground of the lower layer of the RF trace is solid, and all components are firmly connected to the main ground and isolated from other traces that may bring noise.
In addition, make sure that the power supply of the VCO is adequately decoupled. Since the RF output of the VCO is often a relatively high level, the VCO output signal can easily interfere with other circuits, so special attention must be paid to the VCO. In fact, the VCO is often placed at the end of the RF area, and sometimes it also requires a metal shield. The resonant circuit (one for the transmitter and the other for the receiver) is related to the VCO, but it also has its own characteristics. Simply put, the resonant circuit is a parallel resonant circuit with a capacitive diode, which helps set the VCO operating frequency and modulate the voice or data onto the RF signal. All VCO design principles also apply to the resonant circuit. Because the resonant circuit contains a considerable number of components, a wide distribution area on the board, and usually operates at a very high RF frequency, the resonant circuit is usually very sensitive to noise.
Signals are usually arranged on adjacent pins of the chip, but these signal pins need to work with relatively large inductors and capacitors, which in turn requires that these inductors and capacitors be located very close to each other and connected back to a control loop that is very sensitive to noise. This is not easy to achieve.
The automatic gain control (AGC) amplifier is also a problem area. Both the transmit and receive circuits have AGC amplifiers. The AGC amplifier can usually filter out noise effectively, but because the mobile phone has the ability to handle rapid changes in transmit and receive signal strength, the AGC circuit is required to have a fairly wide bandwidth, which makes it easy for the AGC amplifier on some key circuits to introduce noise. The design of the AGC line must follow good analog circuit design techniques, which are related to very short op amp input pins and very short feedback paths, both of which must be away from RF, IF or high-speed digital signal routing.
Likewise, good grounding is essential, and the chip's power supply must be well decoupled. If you must run a long line at the input or output, it is best to do it at the output, which usually has much lower impedance and is less likely to induce noise. Generally, the higher the signal level, the easier it is to introduce noise into other circuits.
In all PCB designs, it is a general principle to keep digital circuits away from analog circuits as much as possible, which also applies to RF PCB design. Common analog ground and ground used to shield and separate signal lines are usually equally important, so careful planning, thoughtful component layout and thorough layout evaluation are very important in the early stages of design. RF lines should also be kept away from analog lines and some critical digital signals. All RF traces, pads and components should be filled with as much ground copper as possible and connected to the main ground as much as possible. If the RF trace must pass through the signal line, try to lay a layer of ground connected to the main ground along the RF trace between them. If this is not possible, make sure they are cross-crossed, which can minimize capacitive coupling, and at the same time, try to lay more ground around each RF trace and connect them to the main ground.
Additionally, minimizing the distance between parallel RF traces can minimize inductive coupling. Isolation is best achieved when a solid, monolithic ground plane is placed directly under the surface layer on the first layer, although other approaches can work with careful design. On each layer of the PCB, place as many ground planes as possible and connect them to the main ground plane. Place traces as close together as possible to increase the number of ground planes on internal signal layers and power distribution layers, and adjust traces appropriately so that you can place ground connection vias to isolated ground planes on the surface layer. Avoid creating free ground planes on each PCB layer because they can pick up or inject noise like a small antenna. In most cases, if you can't connect them to the main ground, then you're better off removing them.
3. Several aspects should be paid attention to when designing PCB boards
1. Processing of power supply and ground wire
Even if the wiring in the entire PCB board is completed well, the interference caused by the inconsiderate consideration of the power supply and ground wires will reduce the performance of the product and sometimes even affect the success rate of the product. Therefore, the wiring of the power and ground wires should be taken seriously to minimize the noise interference generated by the power and ground wires to ensure the quality of the product. Every engineer engaged in the design of electronic products understands the cause of the noise between the ground wire and the power wire. Now we will only describe the noise reduction method:
(1) It is well known that decoupling capacitors are added between the power supply and the ground line.
(2) Try to widen the width of the power and ground wires. It is best if the ground wire is wider than the power wire. The relationship between them is: ground wire > power wire > signal wire. Usually the signal wire width is: 0.2~0.3mm, the thinnest width can reach 0.05~0.07mm, and the power wire is 1.2~2.5 mm. For the PCB of digital circuits, a wide ground wire can be used to form a loop, that is, to form a ground network (the ground of analog circuits cannot be used in this way)
(3) Use a large copper layer as the ground wire, and connect all unused areas on the printed circuit board to the ground as the ground wire. Or make a multi-layer board, with the power supply and ground wire occupying one layer each.
2. Common ground processing for digital circuits and analog circuits
Now many PCBs are no longer single-function circuits (digital or analog circuits), but are a mixture of digital circuits and analog circuits. Therefore, when wiring, it is necessary to consider the mutual interference between them, especially the noise interference on the ground line. The frequency of digital circuits is high, and the sensitivity of analog circuits is strong. For signal lines, high-frequency signal lines should be as far away from sensitive analog circuit devices as possible. For ground lines, the entire PCB has only one node to the outside world, so the problem of digital and analog common ground must be handled inside the PCB. In fact, the digital ground and analog ground are separated inside the board. They are not connected to each other, but at the interface where the PCB is connected to the outside world (such as plugs, etc.). There is a short circuit between the digital ground and the analog ground. Please note that there is only one connection point. There are also non-common grounds on the PCB, which is determined by the system design.
3. Signal lines are laid on the electrical (ground) layer
When wiring a multi-layer printed circuit board, there are not many unfinished wires in the signal layer. Adding more layers will cause waste and increase the workload of production, and the cost will increase accordingly. To solve this contradiction, you can consider wiring on the power (ground) layer. First, consider using the power layer, and then the ground layer. Because it is best to preserve the integrity of the ground layer.
4. Treatment of connecting legs in large area conductors
In large-area grounding (electricity), the legs of common components are connected to it. The treatment of the connection legs needs to be comprehensively considered. In terms of electrical performance, it is better for the pads of the component legs to be fully connected to the copper surface, but there are some hidden dangers in the welding and assembly of components, such as: ① Welding requires a high-power heater. ② It is easy to cause cold solder joints. Therefore, taking into account both electrical performance and process requirements, a cross-shaped pad is made, which is called heat shield, commonly known as thermal pad. In this way, the possibility of cold solder joints caused by excessive heat dissipation in the cross section during welding can be greatly reduced. The treatment of the legs of the power (ground) layer of the multilayer board is the same.
5. The role of network system in wiring
In many CAD systems, wiring is determined by the network system. If the grid is too dense, the number of paths will increase, but the stepping is too small, and the amount of data in the drawing field is too large, which will inevitably have higher requirements for the storage space of the equipment, and will also have a great impact on the computing speed of computer-related electronic products. Some paths are invalid, such as those occupied by the pads of the component legs or by the mounting holes and fixed holes. If the grid is too sparse, too few paths will have a great impact on the wiring rate. Therefore, a grid system with reasonable density is required to support the wiring. The distance between the two legs of a standard component is 0.1 inches (2.54 mm), so the basis of the grid system is generally set to 0.1 inches (2.54 mm) or an integer multiple of less than 0.1 inches, such as: 0.05 inches, 0.025 inches, 0.02 inches, etc.
4. High-frequency PCB design skills and methods
1. The transmission line corners should be 45° to reduce return loss
2. Use high-performance insulating circuit boards with strictly controlled insulation constant values at different levels. This method is conducive to the effective management of the electromagnetic field between the insulating material and the adjacent wiring.
3. Improve the PCB design specifications for high-precision etching. Consider specifying a total line width error of +/-0.0007 inches, managing the undercut and cross-section of the wiring shape, and specifying the wiring sidewall plating conditions. Overall management of the wiring (conductor) geometry and coating surface is very important for solving the skin effect problems related to microwave frequencies and achieving these specifications.
4. Avoid using components with leads because of the tapped inductance of the protruding leads. In high-frequency environments, it is best to use surface-mount components.
5. For signal vias, avoid using the through-hole processing (PTH) process on sensitive boards because this process will cause lead inductance at the vias.
6. Provide abundant grounding layers. Use molded holes to connect these grounding layers to prevent the influence of 3D electromagnetic fields on the circuit board.
7. Choose electroless nickel plating or immersion gold plating process, and do not use HASL method for electroplating.
8. The solder mask prevents the flow of solder paste. However, due to the uncertainty of thickness and unknown insulation performance, covering the entire board surface with solder mask will lead to a large change in the electromagnetic energy in the microstrip design. Solder dam is generally used as the electromagnetic field of the solder mask.
In this case, we are managing the transition from microstrip to coax. In coax, the ground planes are woven in a circular pattern and are evenly spaced. In microstrip, the ground plane is underneath the active lines. This introduces certain edge effects that need to be understood, predicted, and accounted for during design. Of course, this mismatch also results in return loss, which must be minimized to avoid noise and signal interference.
5. Electromagnetic compatibility design
Electromagnetic compatibility refers to the ability of electronic equipment to work in a coordinated and effective manner in various electromagnetic environments. The purpose of electromagnetic compatibility design is to enable electronic equipment to suppress various external interferences, so that electronic equipment can work normally in a specific electromagnetic environment, while reducing the electromagnetic interference of electronic equipment itself to other electronic equipment.
1. Choose a reasonable wire width
Since the impact interference generated by transient current on the printed lines is mainly caused by the inductance component of the printed conductor, the inductance of the printed conductor should be minimized. The inductance of the printed conductor is proportional to its length and inversely proportional to its width, so short and fine conductors are beneficial for suppressing interference. The signal lines of clock leads, row drivers or bus drivers often carry large transient currents, so the printed conductors should be as short as possible. For discrete component circuits, the requirements can be fully met when the printed conductor width is about 1.5mm; for integrated circuits, the printed conductor width can be selected between 0.2 and 1.0mm.
2. Use the right wiring strategy
Using equal routing can reduce wire inductance, but the mutual inductance and distributed capacitance between wires increase. If the layout allows, it is best to use a tic-tac-toe mesh wiring structure. The specific method is to wire horizontally on one side of the printed circuit board and vertically on the other side, and then connect them with metallized holes at the cross holes.
3. Effectively suppress crosstalk
In order to suppress the crosstalk between the wires on the printed circuit board, long-distance equal routing should be avoided as much as possible when designing the wiring, and the distance between the wires should be as far as possible, and the signal wires should not cross the ground wires and power wires as much as possible. Setting a grounded printed wire between some signal wires that are very sensitive to interference can effectively suppress crosstalk.
4. Key points for wiring to avoid electromagnetic radiation
In order to avoid electromagnetic radiation generated when high-frequency signals pass through printed conductors, the following points should be noted when wiring printed circuit boards:
(1) Minimize the discontinuity of printed conductors, for example, the conductor width should not change suddenly, the conductor corner should be greater than 90 degrees, and loop routing is prohibited.
(2) Clock signal leads are most likely to generate electromagnetic radiation interference. They should be routed close to the ground loop and the driver should be close to the connector.
(3) The bus driver should be close to the bus it is intended to drive. For those leads that leave the printed circuit board, the driver should be close to the connector.
(4) The data bus should be wired with a signal ground wire between every two signal wires. It is best to place the ground loop close to the least important address lead because the latter often carries high-frequency current.
5. Suppress reflection interference
In order to suppress the reflection interference appearing at the terminal of the printed line, the length of the printed line should be shortened as much as possible and a slow circuit should be used, except for special needs. If necessary, terminal matching can be added, that is, a matching resistor with the same resistance value is added to the ground and power supply ends at the end of the transmission line. According to experience, for TTL circuits with generally faster speeds, terminal matching measures should be adopted when the printed line is longer than 10cm. The resistance value of the matching resistor should be determined according to the maximum value of the output drive current and absorption current of the integrated circuit.
6. Use differential signal line routing strategy during circuit board design
Differential signal pairs that are routed very close to each other will also be tightly coupled to each other. This mutual coupling will reduce EMI emissions. Usually (of course there are some exceptions) differential signals are also high-speed signals, so high-speed design rules usually apply to the routing of differential signals, especially when designing signal lines of transmission lines. This means that we must be very careful in designing the routing of signal lines to ensure that the characteristic impedance of the signal line is continuous and constant along the signal line.
During the layout and routing of differential line pairs, we hope that the two PCB lines in the differential line pair are completely consistent. This means that in practical applications, we should make every effort to ensure that the PCB lines in the differential line pair have exactly the same impedance and the wiring length is also completely consistent. Differential PCB lines are usually always routed in pairs, and the distance between them is kept constant at any position along the direction of the line pair. Normally, the layout and routing of differential line pairs are always as close as possible.
Note: This article is reposted from the Internet and is only for technical sharing. If there is any infringement, please inform us.
|