There is a limit to how many circuits can be installed on a board. For the sake of manufacturing equipment and standard panel size, PCB itself has an upper size limit. The number of transistors on a device continues to expand, while resistors and capacitors almost disappear to the naked eye. Even with all integration and miniaturization, the complexity of the circuit will grow to an unmanageable scale. A PCB can’t solve the problem.
There are various expansion forms beyond a single PCB. Telecommunications and other network devices often appear in rack systems. The familiar 19-inch wide x 1.75-inch high rack unit is the default housing size for smaller circuits. This is what you see when you read the term server farm or data center. From there, the baseline extends to multiple rack units. Although they come in various sizes, the popular one is the seven-rack rack. It’s still 19 inches wide, but it takes up just over a foot of rack space. The vertical PCB is installed on the shelf and connected with a 19“ x 12 “backplane, which is installed on the back of the shelf.
The backplane is usually just a connector and wiring. The number of layers may be high to support a large number of connections. The thickness of this board provides the required rigidity when engaging with the 640-pin connector, while resisting anyone who tries to preheat the bottom board before welding. Therefore, press-fit connectors are standard configurations on these types of printed circuit boards.
Treating thick plate
Besides connectors, we usually need some kind of housekeeping circuit. These components will need to be surface mounted, especially when the backplane is on the thicker side. The standard pins on the through-hole components do not reach the required length of the far end of the backplane, so they are not solderable or reworkable. Then the aforementioned pound of copper used for the ground plane to increase the reflow temperature. Not pretty.
Another consideration related to the thickness of the circuit board is the through hole in the Z axis. If you want to wire from the top to the 4th floor of the 24-floor board, the other 20 floors will form a suspension line. If we go back after all electroplating and drill a new slightly larger hole where the 20-story short pile is located, this discontinuity can be eliminated. It takes extra effort to drill holes in the reverse direction, but the results of signal integrity eye diagram can be measured.
This brings us to all the other PCBs in the system. In the data center, server blades will occupy most slots, so you can expect some uniformity. The system is scalable, and each new board can improve the overall performance.
Other network devices may be more diversified. Each card slot has a different purpose, and each card slot must be filled to make the whole group work properly. Because of all the huge parts and the accompanying extreme voltage swing, the power supply is usually an independent board. The most power-hungry card finds a place next to the power supply, while the most sensitive victim hides in his own small box away from heat and noise. Think of all the gadgets that flow down from wall warts. The power supply is put on the power socket, and the juice is transported to the rest of the system through ferrite beads and/or shielded wires. The same principle.
good-neighbour policy
Subdividing the rest of the schematic diagram into circuit boards is usually based on block diagrams. The combination of these modules should depend on the technology involved. Some circuits will require a higher number of layers or active HDI technology. Others can live with the vanilla-flavored four-layer board. Fine adjustment in function and technology depends on the power supply range, coexistence, thermal path, weight distribution and final aesthetics. I remember a board in a system called “tROtS Board”. Stands for “the rest of the story”. Sometimes, another word will take the place of “story”.
Not all systems need a 7-foot rack with electronic equipment to be qualified. Only a few PCIe or SATA connectors are needed. The example I think of is the prototype version of Google’s “home” mesh router. Use another Qualcomm chipset, which is a larger size. Three additional PCIe card slots enable it to test the functions of multiple router configurations. The air interface is another thing. The only real way to test the wireless part of the link is to set up a test room. We used the four-board system to enable the baseband level test in the laboratory. Prototype is one of the cool things you’ve ever heard about 20% of the time.
These microsystems may bring you more freedom. Usually, the backplane is more of a motherboard that carries some computing load. In this case, the suspect you usually need must attend the meeting. Memory and various drivers may find a house next to the microprocessor. Of course, the memory usually resides on the SIM module, which comes down to a circuit board with a connector and a row of memory chips. In this respect, it can be regarded as another board of the system. Any indicators, buttons, monitors or other controls that have been used in the product will eventually appear on this layout. Then, when we want to add more functions, the plug-in will act as an expansion slot. That’s the model we used to test the vehicle at Google Home.
This type of healthy multi-board system will require a certain degree of standardization. In particular, daughter cards will want to comply with the protocol, and may even be ready-made sub-components. In this case, the content of the line card (or any other content) is predetermined.
With the development of the world towards 5G mesh network, the subsystem will have to have replaceable and upgradeable equipment of cellular level, as well as some serious bandwidth of substation. Radio must be cheap, durable and ubiquitous. Centralized blocks will require unlimited throughput to ensure bulletproof reliability. The aggregation of all these data will not be done on a single board. The big network is coming soon. Get ready.