When it comes to PCB boards, many people think of them as something that can be seen everywhere around us, from all kinds of household appliances and computer components to various digital products. Almost all electronic products use PCB boards. So what exactly is a PCB board?
A PCB stands for Printed Circuit Board, a base plate with circuitry for mounting electronic components. It's created by printing resist-etched circuitry onto a copper-plated base plate and then etching it to reveal the circuitry.
PCBs can be classified into single-layer, double-layer, and multi-layer boards. Various electronic components are integrated onto PCBs. In the most basic single-layer PCB, components are concentrated on one side, while wires are concentrated on the other. This requires drilling holes in the board so that connectors can pass through to the other side; therefore, component connectors are soldered to the other side. For this reason, the front and back sides of such a PCB are called the component side and the solder side, respectively.
A double-layer PCB can be viewed as two single-layer PCBs bonded together, with electronic components and traces on both sides. Sometimes it's necessary to connect a single trace from one side to the other, which is achieved through vias. Vias are small holes on a PCB, filled or coated with metal, that connect to traces on both sides. Many computer motherboards now use 4-layer or even 6-layer PCBs, while graphics cards generally use 6-layer PCBs. Many high-end graphics cards, like the NVIDIA GeForce 4 Ti series, use 8-layer PCBs; these are what are called multi-layer PCBs. Connecting traces between layers on multi-layer PCBs can also be achieved using vias.
Because PCBs are multilayered, sometimes vias don't need to penetrate the entire PCB. These are called buried vias and blind vias, as they only penetrate a few layers. Blind vias connect several inner PCB layers to the surface PCB without penetrating the entire board. Buried vias only connect to the inner PCB layers, so they are not visible from the surface. In multilayer PCBs, the entire layer is directly connected to ground and power. Therefore, we classify the layers as signal layers, power layers, or ground layers. If components on the PCB require different power supplies, such PCBs usually have two or more power and ground layers. The more layers a PCB uses, the higher the cost. Of course, using more layers on a PCB greatly helps provide signal stability.
The professional PCB manufacturing process is quite complex. Take a 4-layer PCB as an example. Most motherboard PCBs are 4-layer. During manufacturing, the middle two layers are first rolled, cut, etched, and oxidized/plated. These four layers are the component layer, power layer, ground layer, and solder layer, respectively. These four layers are then rolled together to form a single motherboard PCB. Next, holes are drilled and through-holes are made. After cleaning, the outer two layers' circuitry is printed, copper plating is applied, etching is performed, testing is conducted, solder mask is applied, and silkscreen is printed. Finally, the entire PCB (containing many motherboards) is stamped into individual motherboard PCBs, tested, and then vacuum-packed.
If the copper foil is not properly laid during PCB manufacturing, it may not adhere well, potentially leading to short circuits or capacitive effects (which can cause interference). Vias on the PCB also require careful attention. If the vias are not centered but off-center, they can cause uneven matching or contact with power or ground planes, creating potential short circuits or grounding issues.
Copper wire wiring process
The first step in the fabrication process is establishing the wiring between the components. We use negative transfer to transfer the working film onto the metal conductor. This technique involves covering the entire surface with a thin layer of copper foil and removing any excess. Additional transfer is a less common method, where copper wires are only applied where needed, but we won't discuss that here. The photoresist is made from a photosensitive agent that dissolves under light. There are many ways to process photoresist on copper surfaces, but the most common method is to heat it and roll it over the photoresist-containing surface. It can also be sprayed on as a liquid, but dry film application offers higher resolution and allows for the creation of finer conductors. The light shield is simply a template for the PCB layers during manufacturing.
Before the photoresist on the PCB is exposed to UV light, a light shield is placed over it to prevent certain areas of the photoresist from being exposed. These areas covered by the photoresist will become wiring. After the photoresist is developed, the remaining bare copper areas are etched. The etching process can involve immersing the board in an etching solvent or spraying the solvent onto the board. Ferric chloride is commonly used as an etching solvent. After etching, the remaining photoresist is removed.
1. Wiring width and current
The width should generally not be less than 0.2mm (8mil).
On high-density, high-precision PCBs, the spacing and line width are typically 0.3mm (12mil).
When the copper foil thickness is around 50µm, the conductor width is 1–1.5mm (60mil) = 2A.
Public spaces are generally 80 mils, and more attention should be paid to applications with microprocessors.
2. What frequency exactly qualifies as a high-speed board?
A signal is considered a high-speed signal when its rise/fall time is less than 3 to 6 times its transmission time.
For digital circuits, the key is to look at the steepness of the signal edges, that is, the rise and fall times of the signal.
According to the classic book *High Speed Digital Design*, a signal is considered high-speed if the time it takes for it to rise from 10% to 90% is less than six times the conductor delay. That is, even an 8kHz square wave signal is considered high-speed if its edges are steep enough, and transmission line theory should be applied during wiring.
3. PCB board stacking and layering
Four-layer boards can be stacked in several different sequences. The advantages and disadvantages of each sequence are explained below:
First scenario
GND
S1+POWER
S2+POWER
GND
The second scenario
SIG1
GND
POWER
SIG2
The third scenario
GND
S1
S2
POWER
Note: S1 signal routing is on layer 1, S2 signal routing is on layer 2; GND is the ground layer, POWER is the power layer.
The first scenario is arguably the best among four-layer boards. Because the outer layer is the ground plane, it provides EMI shielding, and the power plane is very close to the ground plane, resulting in low power supply internal resistance and optimal performance. However, this first scenario cannot be used when the board density is high. This is because the integrity of the first ground plane cannot be guaranteed, leading to poor signal quality on the second layer. Furthermore, this structure is not suitable for boards with high overall power consumption.
The second scenario, which is the most common approach, is also unsuitable for high-speed digital circuit design due to its board structure. Maintaining low power supply impedance is difficult in this structure. For example, consider a 2mm board with a required Z0 = 50 ohms, an 8mil trace width, and a 35µm copper foil thickness. This results in a 0.14mm gap between the signal layer and ground layer, and a 1.58mm gap between the ground and power layers. This significantly increases the power supply's internal resistance. In this structure, since radiation is directed into space, a shielding plate is needed to reduce EMI.
In the third scenario, the signal quality on layer S1 is the best, followed by layer S2. It provides EMI shielding, but has a relatively high power supply impedance. This board is suitable for applications where overall board power consumption is high and the board itself is a source of interference or is located very close to a source of interference.
4. Impedance matching
The amplitude of the reflected voltage signal is determined by the source reflection coefficient ρs and the load reflection coefficient ρL.
ρL=(RL-Z0)/(RL+Z0) and ρS=(RS-Z0)/(RS+Z0)
In the above formula, if RL=Z0, then the load reflection coefficient ρL=0. If RS=Z0, the source reflection coefficient ρS=0.
Since the impedance Z0 of a typical transmission line should generally meet the requirement of around 50Ω, while the load impedance is typically in the range of several thousand to tens of thousands of ohms, achieving impedance matching at the load end is quite difficult. However, the impedance at the signal source end (output) is usually relatively small, around ten ohms.
Therefore, achieving impedance matching at the source is much easier. If a resistor is connected in parallel at the load, the resistor will absorb some signal, which is detrimental to transmission (my understanding). When choosing a TTL/CMOS standard 24mA drive current, its output impedance is approximately 13Ω. If the transmission line impedance Z0 = 50Ω, then a 33Ω source-end matching resistor should be added. 13Ω + 33Ω = 46Ω (approximately 50Ω; weak underdamping helps with signal setup time).
When choosing different transmission standards and drive currents, the matching impedance will differ. In high-speed logic and circuit design, we strongly recommend adding source-side matching resistors for critical signals such as clock and control signals.
Even if the signal is connected in this way, it will still be reflected back from the load end. Because of the impedance matching at the source end, the reflected signal will not be reflected back.
5. Precautions for power cord and ground wire layout
Keep power cables as short as possible, run them in a straight line, and ideally in a tree-like pattern rather than a loop.
Ground loop problem: For digital circuits, the circulating current caused by a ground loop is only in the tens of millivolts range. The interference immunity threshold for TTL circuits is 1.2V, and for CMOS circuits it can reach half the power supply voltage. This means that ground loop current will not adversely affect the circuit's operation. Conversely, if the ground line is not closed, the problem will be greater because the pulsed power supply current generated during the operation of digital circuits will cause an imbalance in ground potential at various points. For example, in my actual measurement, the ground current of the 74LS161 during inversion was 1.2A (measured with a 2Gsps oscilloscope, with a ground current pulse width of 7ns).
Under the impact of a large pulse current, if a branched ground plane (25mil linewidth) is used, the potential difference between various points on the ground plane will reach the level of hundreds of millivolts. However, with a ground loop, the pulse current is distributed to various points on the ground plane, greatly reducing the possibility of interference with the circuit. Using a closed ground plane, the measured maximum instantaneous potential difference of the ground planes of various devices is one-half to one-fifth that of the open ground plane. Of course, the measured data of circuit boards with different densities and speeds vary greatly. What I said above refers to the level of the Z80Demo board that comes with Protel99SE. For low-frequency analog circuits, I believe that the power frequency interference after the ground plane is closed is induced from space, which cannot be simulated or calculated in any way.
If the ground wire is not closed, no ground eddy current will be generated. What is the theoretical basis for Beckhamtao's claim that "but the open-loop ground wire will result in a larger power frequency induced voltage"? To give two examples, seven years ago I took over a project for someone else: a precision pressure gauge using a 14-bit A/D converter, but the actual measured accuracy was only 11 bits. Upon investigation, it was found that there was 15mVp-p power frequency interference on the ground wire. The solution was to break up the analog ground loop on the PCB and use flying leads to create a branched distribution of the ground wire from the front-end sensor to the A/D converter. Later, the PCBs of the mass-produced model were re-manufactured according to the flying lead routing, and no problems have occurred since.
In the second example, a friend of mine, an audiophile, built his own amplifier, but it consistently had an AC hum at the output. I suggested he cut the ground loop, and the problem was solved. Afterwards, this friend consulted the PCB diagrams of dozens of "famous Hi-Fi amplifiers" and confirmed that none of them used a ground loop in their analog sections.
6. Printed Circuit Board Design Principles and Anti-interference Measures
Printed circuit boards (PCBs) are the supporting components for circuit elements and devices in electronic products. They provide electrical connections between these components and devices. With the rapid development of electronics technology, PCB density is increasing. The quality of PCB design greatly affects its anti-interference capability. Therefore, when designing PCBs, general PCB design principles must be followed, and anti-interference design requirements must be met.
General principles of PCB design
To achieve optimal performance in electronic circuits, the placement of components and the routing of wires are crucial. To design high-quality, low-cost PCBs, the following general principles should be followed:
1. Layout
First, consider the PCB size. If the PCB is too large, the printed lines are long, impedance increases, noise immunity decreases, and cost increases; if it's too small, heat dissipation is poor, and adjacent lines are easily interfered with. After determining the PCB size, determine the location of special components. Finally, based on the functional units of the circuit, arrange all components.
The following principles should be followed when determining the location of special components:
(1) Minimize the connections between high-frequency components to reduce their distributed parameters and mutual electromagnetic interference. Components susceptible to interference should not be placed too close to each other, and input and output components should be kept as far apart as possible.
(2) There may be a high potential difference between some components or wires. The distance between them should be increased to prevent accidental short circuits caused by discharge. High-voltage components should be placed in places that are not easily accessible by hand during debugging.
(3) Components weighing more than 15g should be fixed with a bracket before soldering. Large, heavy components that generate a lot of heat should not be mounted on the printed circuit board, but on the chassis base plate of the whole machine, and heat dissipation should be considered. Thermistors should be kept away from heat-generating components.
(4) The layout of adjustable components such as potentiometers, adjustable inductors, variable capacitors, and microswitches should take into account the overall structural requirements of the machine. If the adjustment is internal, it should be placed in a convenient location on the printed circuit board; if the adjustment is external, its position should be compatible with the position of the adjustment knob on the chassis panel.
(5) Space should be reserved for the positioning holes of the printing plate and the space occupied by the fixing bracket.
When arranging all components of a circuit according to its functional units, the following principles should be followed:
(1) Arrange the positions of each functional circuit unit according to the circuit flow so that the layout facilitates signal flow and keeps the signals in the same direction as much as possible.
(2) The layout should be centered around the core component of each functional circuit. Components should be arranged evenly, neatly, and compactly on the PCB. Minimize and shorten the leads and connections between components.
(3) For circuits operating at high frequencies, the distributed parameters between components must be considered. Generally, components should be arranged in parallel as much as possible. This not only improves aesthetics but also facilitates assembly, soldering, and mass production.
(4) Components located at the edge of the circuit board should generally be no less than 2mm away from the edge. The optimal shape of the circuit board is rectangular, with an aspect ratio of 3:2 or 4:3. When the circuit board size is greater than 200x150mm, the mechanical strength of the circuit board should be considered.
2. Wiring
The principles of wiring are as follows:
(1) The wires used for input and output terminals should avoid being adjacent and parallel as much as possible. It is best to add a ground wire between the wires to avoid feedback coupling.
(2) The minimum width of printed circuit traces is mainly determined by the adhesion strength between the trace and the insulating substrate and the current flowing through them. When the copper foil thickness is 0.05mm and the width is 1~15mm, the temperature will not exceed 3℃ when a current of 2A flows through it; therefore, a trace width of 1.5mm is sufficient. For integrated circuits, especially digital circuits, a trace width of 0.02~0.3mm is usually selected. Of course, wider traces should be used whenever possible, especially for power and ground lines. The minimum spacing between traces is mainly determined by the worst-case insulation resistance and breakdown voltage. For integrated circuits, especially digital circuits, the spacing can be as small as 5~8mm, provided the process allows.
(3) The bends of printed conductors are generally rounded, while right angles or included angles will affect electrical performance in high-frequency circuits. In addition, try to avoid using large areas of copper foil, otherwise, when heated for a long time, the copper foil is prone to expansion and shedding. If large areas of copper foil must be used, it is best to use a grid pattern. This is conducive to the removal of volatile gases generated by the adhesive between the copper foil and the substrate when heated.
3. Solder pads
The center hole of the solder pad should be slightly larger than the diameter of the device lead. Too large a solder pad can easily lead to cold solder joints. The outer diameter D of the solder pad is generally not less than (d+1.2) mm, where d is the lead hole diameter. For high-density digital circuits, the minimum solder pad diameter can be (d+1.0) mm.
PCB and circuit anti-interference measures
The anti-interference design of printed circuit boards is closely related to the specific circuit. Here, we will only explain some commonly used measures for PCB anti-interference design.
1. Power cord design
Based on the current rating of the printed circuit board, increase the width of the power lines as much as possible to reduce loop resistance. At the same time, ensure that the power and ground lines run in the same direction as the data transmission; this helps enhance noise immunity.
2. Grounding Design
The principle of ground wire design is:
(1) Digital ground and analog ground should be separated. If there are both logic circuits and linear circuits on the circuit board, they should be separated as much as possible. The ground of low frequency circuits should be connected in parallel at a single point as much as possible. If it is difficult to connect the actual wiring, it can be partially connected in series and then connected in parallel. High frequency circuits should use multi-point series grounding. The ground wire should be short and long. The area around high frequency components should be covered with a large area of grid-like ground foil.
(2) The grounding wire should be as thick as possible. If a very thin wire is used for the grounding wire, the grounding potential will change with the current, reducing the noise immunity. Therefore, the grounding wire should be thickened so that it can carry three times the allowable current on the printed circuit board. If possible, the grounding wire should be 2-3 mm or more.
(3) The grounding wire forms a closed loop. For printed circuit boards composed only of digital circuits, the grounding circuit is often arranged in a closed loop to improve noise immunity.
3. Decoupling capacitor configuration
One of the common practices in PCB design is to place appropriate decoupling capacitors at various critical locations on the printed circuit board.
The general principle for configuring decoupling capacitors is:
(1) Connect an electrolytic capacitor of 10~100uF across the power input terminal. If possible, it is better to connect one of 100uF or more.
(2) In principle, each integrated circuit chip should be equipped with a 0.01pF ceramic capacitor. If there is not enough space on the printed circuit board, a 1-10pF ceramic capacitor can be installed for every 4-8 chips.
(3) For devices with weak noise immunity and large power supply changes when turned off, such as RAM and ROM memory devices, a decoupling capacitor should be directly connected between the power line and ground line of the chip.
(4) The capacitor leads should not be too long, especially the high-frequency bypass capacitors should not have leads.
In addition, the following two points should be noted:
(1) When there are components such as contactors, relays, and buttons on the printed circuit board, large spark discharges will be generated when they are operated. The RC circuit shown in the attached figure must be used to absorb the discharge current. Generally, R is 1~2K and C is 2.2~47UF.
(2) CMOS has a high input impedance and is easily induced, so the unused terminals should be grounded or connected to a positive power supply when in use.
7. Design techniques and key points for achieving efficient automated PCB routing
Despite the power of modern EDA tools, PCB design remains challenging due to increasingly smaller PCB sizes and higher component densities. How can we achieve high routing completion rates and shorten design time? This article introduces design techniques and key points for PCB planning, placement, and routing. With increasingly shorter design times, smaller board spaces, higher component densities, extremely stringent layout rules, and large components, the work of designers is becoming more difficult. To address these challenges and accelerate product launches, many manufacturers now tend to use dedicated EDA tools for PCB design. However, dedicated EDA tools often fail to produce ideal results, cannot achieve 100% routing completion, and are often messy, requiring significant time to complete the remaining work.
There are many popular EDA software tools on the market, but they are largely similar except for the terminology and function key locations. How can these tools be used to better achieve PCB design? Careful analysis of the design and proper configuration of the software tools before routing will make the design more compliant with requirements. Below is a general design process and steps.
1. Determine the number of PCB layers
The board size and number of routing layers need to be determined early in the design process. If the design requires the use of high-density ball grid array (BGA) components, the minimum number of routing layers required for these devices must be considered. The number of routing layers and the stack-up method directly affect the routing and impedance of the traces. The board size helps determine the stack-up method and trace width to achieve the desired design effect.
For years, it has been believed that fewer circuit board layers equate to lower costs; however, many other factors influence PCB manufacturing costs. In recent years, the cost difference between multilayer boards has significantly decreased. It's best to start with more circuit layers and evenly distribute the copper during the initial design phase to avoid discovering issues like signals not conforming to defined rules and space requirements near the end of the design process, thus forcing the addition of new layers. Careful planning before design will reduce many routing problems.
2. Design rules and limitations
Automatic routing tools don't inherently know what to do. To complete a routing task, they need to operate within the correct rules and constraints. Different signal lines have different routing requirements, necessitating the categorization of all signal lines with special requirements, and these categorizations vary depending on the design. Each signal class should have a priority level, with higher priority levels enforcing stricter rules. These rules encompass trace width, maximum number of vias, parallelism, inter-signal line interactions, and layer limitations, significantly impacting the performance of the routing tool. Carefully considering design requirements is a crucial step towards successful routing.
3. Component layout
To optimize the assembly process, Design for Manufacturability (DFM) rules impose constraints on component placement. If the assembly department allows component movement, the circuitry can be appropriately optimized, facilitating automated routing. The defined rules and constraints influence the placement design.
When laying out a layout, routing channels and via areas need to be considered. These channels and areas are obvious to designers, but automatic routing tools only consider one signal at a time. By setting routing constraints and specifying the layers on which signal lines can be routed, the routing tool can complete the routing as the designer envisions.
4. Fan-out design
During the fan-out design phase, to enable automated routing tools to connect component pins, each pin of a surface-mount device should have at least one via so that the board can perform inner-layer interconnection, in-circuit testing (ICT), and circuit reprocessing when more connections are needed.
To maximize the efficiency of automated routing tools, use the largest possible via size and trace size, ideally with a spacing of 50 mil. Use via types that maximize the number of routing paths. When fan-out design, consider in-circuit testing. Test fixtures can be expensive and are typically ordered just before full production; adding nodes to achieve 100% testability at that point is too late.
After careful consideration and prediction, in-circuit testing can be designed in the early stages of design and implemented later in the production process. The via fan-out type is determined based on the routing path and in-circuit testing. Power and ground also affect routing and fan-out design. To reduce the inductive reactance generated by filter capacitor connections, vias should be placed as close as possible to the pins of surface-mount devices. Manual routing may be necessary in some cases, which may affect the originally planned routing path and even require a reconsideration of which via to use. Therefore, the relationship between via and pin inductive reactance must be considered, and via specifications must be prioritized.
5. Manual wiring and processing of critical signals
Although this article primarily discusses automated routing, manual routing remains an important process in printed circuit board design, both now and in the future. Manual routing helps automated routing tools complete the routing work. As shown in Figures 2a and 2b, by manually routing and fixing the selected nets, a path can be formed for automated routing.
Regardless of the number of critical signals, these signals should be routed first, either manually or using a combination of manual and automated routing tools. Critical signals typically require careful circuit design to achieve the desired performance. After routing, these signal routes should be inspected by relevant engineers; this process is relatively easy. Once the inspection is passed, these lines are secured, and then automated routing of the remaining signals can begin.
6. Automatic wiring
When routing critical signals, it's necessary to control certain electrical parameters during the routing process, such as reducing distributed inductance and EMC. The same applies to routing other signals. All EDA vendors provide a method to control these parameters. Understanding the input parameters of autorouting tools and their impact on routing quality can, to a certain extent, ensure the quality of autorouting.
A general set of rules should be used for automatic signal routing. By setting restrictions and no-route zones to limit the layers and number of vias used for a given signal, the routing tool can automatically route according to the engineer's design. If there are no restrictions on the layers and number of vias used by the automatic routing tool, it will use every layer and generate a large number of vias.
After setting the constraints and applying the created rules, automatic routing will achieve results close to expectations. Of course, some follow-up work may still be needed, and space must be ensured for other signal and network cabling. Once a portion of the design is complete, it should be fixed in place to prevent it from being affected by subsequent routing processes.
Use the same steps to route the remaining signals. The number of routing iterations depends on the complexity of the circuit and the number of general rules you define. After each type of signal is completed, the constraints on the remaining network routing decrease. However, this also means that many signal routing tasks require manual intervention. Modern autorouting tools are very powerful and can typically complete 100% of the routing. However, when the autorouting tool has not completed all signal routing, the remaining signals must be manually routed.
7. Key design considerations for automatic wiring include:
Slightly change the settings and try different routing paths;
Keeping the basic rules unchanged, try different routing layers, different printed lines and spacing widths, as well as different line widths and different types of vias such as blind vias and buried vias, and observe how these factors affect the design results;
Allow the cabling tools to process the default networks as needed;
The less important the signal, the greater the freedom that automatic routing tools have in routing it.
8. Organizing the wiring
If your EDA software can list the signal trace lengths, checking this data might reveal that some signal traces with few constraints are quite long. This problem is relatively easy to handle; manual editing can shorten the signal trace length and reduce the number of vias. During the cleanup process, you need to determine which traces are reasonable and which are not. Similar to manual routing design, automatic routing design can also be cleaned up and edited during the review process.
9. Appearance of the circuit board
In the past, designs often focused on the visual appeal of circuit boards, but things are different now. Automated circuit boards may not be as aesthetically pleasing as manually designed ones, but they meet the specified requirements in terms of electronic characteristics, and the overall performance of the design is guaranteed.
