Board system interconnects include chip-to-board, PCB-board interconnects, and three types of interconnects such as signal input/output between the PCB and external devices.
1. Interconnection between chip and PCB board
Pentium IV and high-speed chips containing a large number of input/output interconnects have already been introduced. As far as the chip itself is concerned, its performance is reliable, and its processing speed can already reach 1 GHz. The most exciting aspect of the recent GHz Interconnect Symposium (www.az.ww.com) is that the methods for dealing with the increasing number of I/Os and frequency issues are well known. The main problem with chip-to-PCB interconnects is that too high an interconnect density can cause the basic structure of the PCB material to become a limiting factor in the growth of interconnect density. An innovative solution was proposed at the conference, which uses local wireless transmitters inside the chip to transfer data to adjacent circuit boards.
Regardless of whether or not this solution is effective, the participants are very clear: In terms of high-frequency applications, IC design technology is far ahead of PCB design technology.
2. PCB board interconnection
The techniques and methods for high-frequency PCB design are as follows:
1. Use a 45° angle at the corner of the transmission line to reduce the return loss (Figure 1);
2. Use a high-performance insulated circuit board with tightly controlled levels of insulation constant values. This method facilitates effective management of the electromagnetic field between the insulating material and adjacent wiring.
3. To perfect PCB design specifications for high-precision etching. Consider the specified total line width error of +/- 0.0007 inches, manage the undercuts and cross-sections of the wiring shapes, and specify the wiring sidewall plating conditions. The overall management of the wiring (wire) geometry and the coating surface is very important for solving the skin effect problems associated with microwave frequencies and implementing these specifications.
4. The presence of tapped inductance in the protruding leads avoids the use of leaded assemblies. In high frequency environments, it is best to use surface mount components.
5. For signal vias, avoid using the pth process on the sensitive board because it can cause lead inductance at the via. If a via on a 20-layer board is used to connect layers 1 to 3, the lead inductance can affect layers 4 to 19.
6. To provide a rich ground plane. The use of molded holes to connect these ground planes prevents the influence of 3-dimensional electromagnetic fields on the circuit board.
7. To select the electroless nickel plating or immersion gold plating process, do not use the HASL method for electroplating. This plated surface can provide better skin effect for high-frequency currents (Figure 2). In addition, fewer leads are needed for this high solderable coating, helping to reduce environmental pollution.
8. Solder mask prevents solder paste flow. However, due to the uncertainty of the thickness and the unknown nature of the insulation properties, covering the entire board surface with the solder mask material will result in a large change in the electromagnetic energy in the microstrip design. Solder dams are generally used as solder masks.
If you are not familiar with these methods, consult an experienced design engineer who has worked on military microwave circuit board designs. You can also discuss with them the range of prices you can afford. For example, using a copper-backed coplanar microstrip design is more economical than a stripline design, and you can discuss this with them to get better advice. Good engineers may not be accustomed to cost considerations, but their advice is also quite helpful. It is now a long-term task to train young engineers who are not familiar with RF effects and lack experience in dealing with RF effects.
In addition, other solutions can also be used, such as improving the computer type to have RF effect processing capabilities.
3. PCB and external device interconnection
It can now be assumed that we have solved all signal management issues on the board as well as on the various discrete component interconnects. So how do you solve the signal input/output problems from the circuit board to the remote device wires? Trompet Electronics, an innovator of coaxial cable technology, is working to solve this problem and has made some important advances (Figure 3). Also, look at the electromagnetic field given in Figure 4. In this case, we manage the transition from microstrip to coaxial cable. In coaxial cables, the ground layers are circularly interwoven and spaced evenly. In microstrip, the ground plane is below the active line. This introduces some edge effects that need to be understood, predicted, and considered at design time. Of course, this mismatch also results in return loss. This mismatch must be minimized to avoid noise and signal interference.
The management of impedance issues within the board is not a negligible design issue. The impedance begins at the surface of the board, passes through a solder joint to the connector, and ends at the coaxial cable. Since the impedance varies with frequency, the higher the frequency, the more difficult it is to manage the impedance. The problem of using higher frequencies to transmit signals over wideband appears to be a major problem in design.