Q: What are the important PCB layout rules when using high-speed converters?
A: In order to ensure that the design performance to achieve the technical specifications of the manual, must comply with some guiding principles. First, there is a common question: "Can the AGND and DGND ground planes be separated?" The simple answer is: as the case may be.
The detailed answer is: usually not separated. Because in most cases, separating the ground layer will only increase the return current of the inductor, and the disadvantages it brings are greater than the benefits. From the formula V = L (di / dt) can be seen, as the inductance increases, the voltage noise will increase. As the switching current increases (because the converter sampling rate increases), the voltage noise will also increase. Therefore, the ground plane should be connected together.
One example is that in some applications, in order to meet traditional design requirements, dirty bus power or digital circuits must be placed in certain areas, while also subject to size constraints, making the circuit board can not achieve a good layout of the In this case, the separation of the ground layer is the key to achieving good performance. However, in order to make the overall design effective, it must be somewhere in the circuit board through a bridge or connection point to connect these ground layer together. Therefore, the connection points should be evenly distributed on the separate ground layer. Ultimately, PCBs often have a connection point that becomes the best place to return the current without causing performance degradation. This connection point is usually located near or below the converter.
When designing the power supply layer, you should use all the copper wires that these layers can use. If possible, do not allow these layers to be trapped because additional traces and vias split the power plane into smaller pieces, which can quickly damage the power plane. The resulting sparse power supply layer can squeeze the current path to where the path is most needed, the power supply pin of the converter. The current between the extruded vias and the traces increases the resistance, resulting in a slight voltage drop across the power supply pins of the converter.
Finally, the placement of the power supply layer is essential, do not stack high noise digital power on the analog power layer, otherwise the two are located in different layers, but still possible coupling. In order to minimize the risk of system performance degradation, these types of layers should be separated from each other as much as possible.
The first part of this RAQ discusses why AGND and DGND ground planes are not necessarily separated, unless the specific requirements of the design require you to do so. The second part discusses the design of a printed circuit board (PCB) transmission system (PDS), which is often overlooked, but is critical for system-level analog and digital designers.
The design goal of the PDS is to minimize the voltage ripple generated in response to supply current requirements. All circuits require current, some circuit demand is large, some circuits need to provide a faster rate of current. Using a fully decoupled low-impedance power plane or ground plane and a good PCB stack, the voltage ripple due to the current demand of the circuit can be minimized. For example, if the designed switching current is 1A and the impedance of the PDS is 10mΩ, the maximum voltage ripple is 10mV.
First, you should design a PCB stack structure that supports larger layers of capacitance. For example, a six-layer stack may contain a top signal layer, a first ground layer, a first power supply layer, a second power supply layer, a second ground layer, and a bottom signal layer. It is prescribed that the first ground layer and the first power supply layer are close to each other in the laminated structure, and the two layers are spaced from 2 to 3 mils to form an intrinsic layer capacitance. The biggest advantage of this capacitor is that it is free, just in the PCB manufacturing notes. If you have to split the power plane and have multiple VDD rails on the same layer, use the power supply layer as large as possible. Do not leave empty, but also should pay attention to sensitive circuits. This will maximize the capacitance of the VDD layer. If the design allows the presence of additional layers (in this example, from six to eight), two additional ground planes should be placed between the first and second power planes. In the case of a core pitch of 2 to 3 mils, the inherent capacitance of the laminated structure will be doubled.
For ideal PCB stacking, decoupling capacitors should be used around the power source start-up point and around the DUT, which ensures that the PDS impedance is low over the entire frequency range. The use of several 0.001μF to 100μF capacitors helps to cover this range. It is not necessary to configure the capacitance everywhere; the capacitor is facing the DUT docking will destroy all the manufacturing rules. If such a stringent measure is needed, there are other problems with the circuit.
The first part discusses why the AGND and DGND ground planes are not necessarily separated unless the specific requirements of the design require you to do so. The second part discusses how the transmission system (PDS), and how the power and ground layers are squeezed together can provide additional capacitance. The third part will discuss the exposed pad (E-Pad), which is an easy to overlook the aspects, but it is to achieve the best PCB design performance and heat is essential.
The exposed pad (pin 0) refers to a pad near most modern high-speed ICs, which is an important connection to all the internal ground of the chip through which it is connected to the center of the device. The presence of exposed pads allows many converters and amplifiers to omit the ground pins. The key is to solder the pad to the PCB, to form a stable and reliable electrical connection and cooling connection, otherwise the system may be severely damaged.
The best possible electrical and thermal connections for exposed pads can be achieved in three steps. First, where possible, the exposed pad should be replicated on each PCB layer, which will provide a thicker thermal connection for all groundings, resulting in rapid heat dissipation, which is particularly important for high power devices. In electrical terms, this will provide a good equipotential bonding for all ground planes. When copying an exposed pad on the ground floor, it can be used as a place to decouple and place a radiator.
Next, the exposed pad is divided into a plurality of identical portions. To the best dish, can be crossed through the screen grid or welding to achieve. During reflow assembly, it is not possible to determine how the solder paste flows to establish a device-to-PCB connection, so the connection may be present, but unevenly distributed, and worse, the connection is small and located at the corner. Splitting the exposed pad into a number of smaller parts allows each area to have a connection point to ensure a reliable and uniform connection between the device and the PCB.
Finally, it should be ensured that the sections have vias connected to the ground. Each area is usually large enough to place multiple vias. Before assembling, be sure to fill each vias with solder paste or epoxy. This step is important to ensure that the exposed solder paste does not flow back into the holes, otherwise it will reduce the chance of proper connection.