PCB Design Technical Q&A

The Chinese electronics market is developing faster and faster, and the speed of electronic product replacement is beyond our expectation. Many different products can appear in almost the same period of time. In the field of electronic product design, what issues should our engineers pay attention to? The reporter visited many design companies and interviewed the technicians of Shenzhen Core Valley Technology Co., Ltd. (hereinafter referred to as Core Valley Technology). The following is the conversation between the reporter and the technicians of Core Valley Technology Company:

(Note: XinGu Technology is mainly engaged in electronic product solutions. Its reverse technology research and development division is currently the largest R&D center in the world. It is mainly engaged in PCB design, solution design, chip design, chip decryption, MCU decryption, IC decryption, electronic product imitation, PCB copying, and a series of reverse businesses)

About the selection of hybrid circuit PCB materials and wiring considerations

Q: In today's wireless communication devices, the RF part often uses a miniaturized outdoor unit structure, and the RF part, IF part, and low-frequency circuit part of the outdoor unit are often deployed on the same PCB. What are the material requirements for such PCB wiring? How to prevent interference between RF, IF, and low-frequency circuits?

Answer: Hybrid circuit design is a big problem, and it is difficult to have a perfect solution. Generally, we lay out and route the RF circuit as an independent single board in the system, and there will even be a special shielding cavity. Moreover, the RF circuit is generally single-sided or double-sided, and the circuit is relatively simple. All of this is to reduce the impact on the distributed parameters of the RF circuit and improve the consistency of the RF system. Compared with the general FR4 material, the RF circuit board tends to use a high-Q value substrate. The dielectric constant of this material is relatively small, the distributed capacitance of the transmission line is small, the impedance is high, and the signal transmission delay is small.

In hybrid circuit design, although RF and digital circuits are made on the same PCB, they are generally divided into RF circuit area and digital circuit area, and the layout and wiring are done separately. Ground vias and shielding boxes are used in between.

About the methods and rules of input and output connection

Q: In modern high-speed PCB design, in order to ensure the integrity of the signal, it is often necessary to terminate the input or output of the device. What are the termination methods? What factors determine the termination method? What are the rules?

Answer: Termination is also called matching. Generally, there are source-end matching and terminal matching according to the matching position. Among them, source-end matching is generally resistor series matching, and terminal matching is generally parallel matching. There are many ways, including resistor pull-up, resistor pull-down, Thevenin matching, AC matching, and Schottky diode matching. The matching method is generally determined by the BUFFER characteristics, topology, level type and judgment method, and the signal duty cycle, system power consumption, etc. must also be considered. The most critical issue in digital circuits is the timing issue. The purpose of adding matching is to improve the signal quality and obtain a certain signal at the moment of judgment. For level-valid signals, the signal quality is stable under the premise of ensuring the establishment and holding time; for delay-valid signals, the signal change delay speed meets the requirements under the premise of ensuring the monotonicity of the signal delay.

What should I pay attention to when dealing with wiring density?

Q: If the size of the circuit board is fixed, and the design needs to accommodate more functions, it is often necessary to increase the routing density of the PCB. However, this may lead to increased mutual interference between the routings, and the impedance cannot be reduced if the routing is too thin. What are the techniques for high-speed (>100MHz) high-density PCB design?

A: When designing high-speed and high-density PCBs, crosstalk interference is indeed something that needs special attention, because it has a great impact on timing and signal integrity. Here are a few things to note: 1. Control the continuity and matching of the characteristic impedance of the traces. 2. The size of the trace spacing. The spacing commonly seen is twice the line width. Through simulation, you can know the impact of trace spacing on timing and signal integrity and find the minimum tolerable spacing. The results of different chip signals may be different. 3. Choose an appropriate termination method. 4. Avoid the same routing direction of the upper and lower adjacent layers, or even have traces overlapped up and down, because this crosstalk is greater than the situation of adjacent traces on the same layer. 5. Use blind/buried vias to increase the routing area. However, the production cost of the PCB board will increase. In actual implementation, it is indeed difficult to achieve complete parallelism and equal length, but it should be done as much as possible. In addition, differential termination and common mode termination can be reserved to mitigate the impact on timing and signal integrity.

About Impedance Matching in PCB Design

Q: When designing high-speed PCB, impedance matching must be considered to prevent reflection. However, the PCB processing technology limits the continuity of impedance and simulation cannot be achieved. How to consider this issue when designing the schematic? In addition, regarding the IBIS model, I wonder where a relatively accurate IBIS model library can be provided. Most of the libraries we download from the Internet are not very accurate, which greatly affects the reference of simulation.

Answer: When designing high-speed PCB circuits, impedance matching is one of the design elements. The impedance value is absolutely related to the routing method. For example, whether it is on the surface layer (microstrip) or the inner layer (stripline/double stripline), the distance from the reference layer (power layer or ground layer), the routing width, the PCB material, etc. will affect the characteristic impedance value of the routing. In other words, the impedance value can only be determined after routing. General simulation software cannot take into account some impedance discontinuous routing conditions due to the limitations of the line model or the mathematical algorithm used. At this time, only some terminators (terminations), such as series resistors, can be reserved on the schematic diagram to mitigate the effect of routing impedance discontinuity. The real fundamental solution to the problem is to try to avoid impedance discontinuity during routing. The accuracy of the IBIS model directly affects the simulation results. Basically, IBIS can be regarded as the electrical characteristic data of the equivalent circuit of the actual chip I/O buffer, which can generally be converted from the SPICE model (it can also be measured, but there are more restrictions). The SPICE data is absolutely related to chip manufacturing, so the SPICE data of the same device provided by different chip manufacturers is different, and the data in the converted IBIS model will also be different. In other words, if the device of manufacturer A is used, only they have the ability to provide accurate model data of their device, because no one else knows better than them what process their device is made of. If the IBIS provided by the manufacturer is inaccurate, the only fundamental solution is to constantly ask the manufacturer to improve it.

Q: When designing high-speed PCBs, the software we use only checks the set EMC and EMI rules. From what aspects should designers consider the EMC and EMI rules? How to set the rules?

A: Generally, EMI/EMC design needs to consider both radiation and conduction. The former belongs to the higher frequency part (>30MHz) and the latter is the lower frequency part (<30MHz). So we can't just pay attention to the high frequency and ignore the low frequency part. A good EMI/EMC design must consider the device location, PCB stacking arrangement, important connection routing, device selection, etc. at the beginning of the layout. If these are not arranged better in advance, solving them afterwards will be ineffective and increase costs. For example, the location of the clock generator should not be close to the external connector as much as possible, high-speed signals should be routed on the inner layer as much as possible and pay attention to characteristic impedance matching and continuity of the reference layer to reduce reflection, the slew rate of the signal pushed by the device should be as small as possible to reduce the high-frequency component, and when selecting decoupling (decoupling/bypass) capacitors, pay attention to whether its frequency response meets the requirements to reduce the power layer noise. In addition, pay attention to the return path of the high-frequency signal current to make its loop area as small as possible (that is, loop impedance as small as possible) to reduce radiation. You can also use the method of splitting the ground layer to control the range of high-frequency noise. Finally, choose the appropriate grounding point between the PCB and the chassis ground.

Wiring tips for high-speed differential signals

Q: When high-speed differential signal lines are run close to each other and in parallel on a PCB, there will be many benefits due to the mutual coupling of the two lines under impedance matching. However, some people believe that this will increase signal attenuation and affect the transmission distance. Why? I have seen some evaluation boards of large companies that some high-speed wiring is as close and parallel as possible, while others intentionally make the distance between the two lines different. Which one will have a better effect? ​​My signal is above 1GHz and the impedance is 50 ohms. When using software to calculate, is the differential line pair also calculated as 50 ohms? Or is it calculated as 100 ohms? Can a matching resistor be added between the differential line pairs at the receiving end?

A: The reasons for the attenuation of high-frequency signal energy are the conductor's own resistance characteristics (conductor loss), including skin effect, and dielectric loss of dielectric materials. These two factors can be seen in the electromagnetic theory analysis of transmission line effect. The coupling of differential lines will affect the characteristic impedance of each other, making it smaller. According to the voltage divider principle, this will make the voltage sent to the line by the signal source smaller. As for the theoretical analysis of signal attenuation due to coupling, I have not seen it, so I cannot comment. The wiring method of differential pairs should be appropriately close and parallel. The so-called appropriate closeness is because this spacing will affect the value of differential impedance, which is an important parameter for designing differential pairs. Parallelism is also required to maintain the consistency of differential impedance. If the two lines are sometimes far and sometimes close, the differential impedance will be inconsistent, which will affect signal integrity and timing delay. The calculation of differential impedance is 2(Z11 - Z12), where Z11 is the characteristic impedance of the trace itself, and Z12 is the impedance generated by coupling between two differential lines, which is related to the line distance. Therefore, when designing a differential impedance of 100 ohms, the characteristic impedance of the trace itself must be slightly greater than 50 ohms. As for how much greater it should be, it can be calculated using simulation software. Matching resistors are usually added between differential line pairs at the receiving end, and their values ​​should be equal to the value of the differential impedance. This will improve the signal quality.

In the design of electronic products, we know that it is very difficult to design a perfect PCB board, and it needs to go through many debugging. We know that many so-called PCB reverse solution companies out there are actually copying the PCB boards they make. The so-called copying means that the technicians do not understand the principles and can only copy the same. This is why so many solutions fail. The technicians of XinGu Technology suggest that we must not be blind when choosing a technology company, and we must make a good choice after careful consideration.