1 About Routing of High-Speed Differential Signals
Q: When high-speed differential signal pairs are routed close and parallel to each other on a PCB, many benefits are brought by the mutual coupling of the two wires under impedance matching conditions. However, some views hold that this will increase signal attenuation and affect transmission distance. Is this true and why? On evaluation boards of some large companies, some high-speed routing is as close and parallel as possible, while the distance between the two wires is intentionally made uneven. I do not know which one has better effect. The signal of our product is about 1GHz, and the impedance requirement is 50 ohms. When calculating with software, is the differential pair also calculated as 50 ohms or 100 ohms? Can a matching resistor be added between the differential pairs at the receiving end? Thanks!
A: The reasons for high-frequency signal energy attenuation are conductor loss of the conductor itself, including skin effect, and dielectric loss. The influence of these two factors on signal attenuation can be seen in the analysis of transmission line effects through electromagnetic theory. The coupling of differential pairs will affect their respective characteristic impedances and make them smaller. According to the voltage division principle, this will make the voltage sent by the signal source to the wires a little lower. As for the theoretical analysis of signal attenuation caused by coupling, I have not seen it, so I cannot comment. The routing method of differential pairs should be properly close and parallel. The so-called proper closeness is because the spacing will affect the value of differential impedance, which is an important parameter for designing differential pairs. Parallel routing is also required to maintain the consistency of differential impedance. If the distance between the two wires is uneven, the differential impedance will be inconsistent, which will affect SI 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 the coupling between the two differential wires, which is related to the wire spacing. As for the question of whether to calculate the differential impedance as 50 ohms or 100 ohms, it is actually very simple, which mainly depends on the requirements of your bus or design platform. For example, generally, the impedance requirement of USB 2.0 is 90 ohms, while that of SATA is 100 ohms, and the requirements of Intel's X86 motherboards are a little different. So it all depends on the actual situation. As for the calculation method, it can be calculated by simulation software. The most commonly used one is Polar SI9000.
2
Q: To improve anti-interference performance, in addition to separating analog ground and digital ground and connecting them only at the power supply, and thickening ground wires and power wires, I hope experts can give some good opinions and suggestions!
A: In addition to separating and isolating grounds, attention should also be paid to the power supply of the analog circuit part. If it shares the power supply with the digital circuit, it is better to add a filter circuit. In addition, digital signals and analog signals should not be staggered, especially not crossing the split ground (moat).
3 About Copper Pouring and Grounding in Blank Areas of Signal Layers in High-Speed PCB Design
Q: In high-speed PCB design, copper can be poured in blank areas of signal layers. So is it better to ground the copper pouring of multiple signal layers, or half ground and half power?
A: In most cases, copper pouring in blank areas is grounded. Only when pouring copper next to high-speed signal wires, attention should be paid to the distance between copper pouring and signal wires, because the poured copper will reduce the characteristic impedance of the routing a little. Attention should also be paid not to affect the characteristic impedance of other layers, such as in the stripline structure.
4 Matching of High-Speed Signal Wires
Q: In high-speed board layout, why are high-speed signal wires (such as CPU data and address signal wires) required to be matched? What hidden dangers will be caused if they are not matched? What factors determine the matching length range (i.e., the time lag difference of signal wires), and how to calculate it?
A: The main reason for requiring characteristic impedance matching of routing is to avoid the influence of reflection caused by transmission line effect on SI and flight time. That is to say, if there is no matching, the signal will be reflected and its quality will be affected. The length range of all routing is set according to timing requirements. There are many factors affecting signal delay time, and routing length is only one of them. The length of some signal wires must be within a certain range, which is determined by calculating the timing margin under the transmission mode (common clock or source synchronous) used for the signal, and then determining the allowable error of routing length.
5 How to Solve Signal Integrity Problems in High-Speed Design
Q: How to solve signal integrity problems in high-speed design?
A: Signal integrity is basically an impedance matching problem. The factors affecting impedance matching include the architecture of the signal source, output impedance, characteristic impedance of routing, characteristics of the load end, topology, etc. The solution is to rely on termination and adjust the topology of routing.
6 How to Solve Some Theoretical Conflicts in Actual Routing
Q: In actual routing, many theories conflict with each other; for example: a. Handling the connection of multiple analog/digital grounds: Theoretically, they should be isolated from each other, but in actual miniaturized and high-density routing, due to space limitations or absolute isolation, the routing of small-signal analog grounds will be too long, making it difficult to achieve the theoretical connection method. My approach is: split the ground of the analog/digital functional module into a complete isolated island, and connect the analog/digital grounds of the functional module to this isolated island. Then connect the isolated island to the "main" ground through a moat. I wonder if this approach is correct? b. Theoretically, the connection between the crystal oscillator and the CPU should be as short as possible. Due to structural layout reasons, the connection between the crystal oscillator and the CPU is relatively long and thin, so it is interfered and unstable. How to solve this problem from routing? There are many such problems, especially considering EMC and EMI issues in high-speed PCB routing, there are many conflicts, which is very troublesome. May I ask how to solve these conflicts? Thanks a lot!
A: a. Basically, it is correct to split and isolate analog/digital grounds. It should be noted that signal routing should not cross the split area (moat) as much as possible, and the return current path of power and signals should not be made too large. b. The crystal oscillator is an analog positive feedback oscillation circuit. To obtain a stable oscillation signal, it must meet the specifications of loop gain and phase, and the oscillation specifications of this analog signal are easily interfered, even if ground guard traces are added, the interference may not be completely isolated. Moreover, if the distance is too far, the noise on the ground plane will also affect the positive feedback oscillation circuit. Therefore, the distance between the crystal oscillator and the chip must be as close as possible. c. Indeed, there are many conflicts between high-speed routing and EMI requirements. But the basic principle is that the resistors, capacitors or ferrite beads added for EMI must not cause some electrical characteristics of the signal to fail to meet the specifications. Therefore, it is better to solve or reduce EMI problems by routing and PCB stack-up techniques first, such as routing high-speed signals on inner layers. Finally, resistors, capacitors or ferrite beads are used to reduce the damage to the signal.
About Maxipcb
Maxipcb empowers innovators to turn cutting-edge technologies into reality.
We offer one-stop solutions for design, simulation, testing, PCB manufacturing, component procurement and SMT assembly, enabling efficient development, rapid deployment and risk control across the full product lifecycle.Serving the world in communications, industrial automation, aerospace, automotive, semiconductor and beyond, we build a safer, more connected future together.