In general, PCB stack-up design mainly follows two rules:
1. Each trace layer must have an adjacent reference layer (power or ground layer);
2. Maintain a minimum distance between the adjacent main power layer and the ground layer to provide a large coupling capacitance;
The following is an example to explain the stack-up from two-layer board to eight-layer board:
I. Stack-up of single-sided PCB boards and double-sided PCB boards
For two-layer boards, the problem of lamination is no longer due to the small number of layers. Controlling EMI radiation is mainly considered from wiring and layout.
The electromagnetic compatibility problem of single-layer board and double-layer board is more and more prominent. The main reason for this phenomenon is that the area of ??the signal loop is too large, which not only generates strong electromagnetic radiation, but also makes the circuit sensitive to external interference. The easiest way to improve the electromagnetic compatibility of a line is to reduce the loop area of ??critical signals.
Key signals: From the perspective of electromagnetic compatibility, key signals mainly refer to signals that generate strong radiation and signals that are sensitive to the outside world. Signals capable of generating strong radiation are generally periodic signals, such as the low-order signal of a clock or address. Signals that are sensitive to interference are those analog signals that are low in level.
Single and double-layer boards are usually used in low-frequency analog designs below 10KHz:
1) The power cables on the same layer are routed in a radial pattern, and the total length of the cables is minimized;
2) When walking the power and ground wires, keep close to each other; lay a ground wire next to the key signal wire, and this ground wire should be as close to the signal wire as possible. This creates a smaller loop area and reduces the sensitivity of differential mode radiation to external interference. When a ground line is added next to the signal line, a loop with the smallest area is formed, and the signal current will definitely take this loop instead of other ground paths.
3) If it is a double-layer circuit board, you can lay a ground wire along the signal line on the other side of the circuit board, close to the signal line, and make the line as wide as possible. The area of ??the loop formed in this way is equal to the thickness of the circuit board times the length of the signal line.
Stack-up of two-layer and four-layer
1. SIG-GND (PWR) -PWR (GND) -SIG; 2. GND-SIG (PWR) -SIG (PWR) -GND;
For the above two stack-up designs, the potential problem is that for the traditional 1.6mm (62mil) board thickness, the layer spacing will become very large, which is not only detrimental to controlling impedance, interlayer coupling and shielding, but also makes the distance between the power supply ground planes larger, thus reduces the board capacitance and is not conducive to filtering noise.
For the first solution, it is usually applied to the case where there are many chips on the board. This solution can get better SI performance, which is not very good for EMI performance. It is mainly controlled by routing and other details. Main attention: the ground layer is placed in the connected layer of the signal layer with the densest signal, which is beneficial to absorb and suppress the radiation, increase the area of ??the board and reflect the 20H rule.
For the second solution, it is usually applied to the situation where the chip density on the board is low enough and there is a sufficient area around the chip (the required copper layer for the power supply is placed). In this solution, the outer layers of the PCB are ground layers, and the middle two layers are signal / power layers. The power supply on the signal layer is routed with a wide wire, which can make the path impedance of the power supply current low, and the impedance of the signal microstrip path also low. It can also shield the inner layer signal radiation through the outer layer. From an EMI control perspective, this is the best 4-layer PCB structure available.
Note: The distance between the middle two layers of signal and power layers should be widened, and the direction of the trace should be vertical to avoid crosstalk. The area of ??the board should be properly controlled to reflect the 20H rule. If you want to control the impedance of the trace, the above scheme should be carefully arranged under the power and ground copper. In addition, the power or ground copper should be interconnected as much as possible to ensure DC and low frequency connectivity.
Stack-up of three-layer and six-layer
For designs with higher chip density and higher clock frequency, the design of 6-layer boards should be considered. The recommended stacking method is:
1.SIG-GND-SIG-PWR-GND-SIG; For this scheme, this stacking scheme can get better signal integrity, the signal layer is adjacent to the ground layer, the power layer and the ground layer are paired, each impedance of the trace layer can be well controlled, and both ground layers can absorb magnetic lines of force well. And in the case of complete power and ground layers, it can provide a better return path for each signal layer.
2. GND-SIG-GND-PWR-SIG-GND; For this scheme, this scheme is only suitable for situations where the device density is not very high. This stack-up has all the advantages of the above stack-up, and such top and bottom ground layer is relatively complete and can be used as a better shielding layer. It should be noted that the power layer should be close to the non-main component layer, because the layer of the bottom layer will be more complete. Therefore, EMI performance is better than the first solution.
Summary: For the six-layer board solution, the distance between the power layer and the ground layer should be minimized to obtain good power and ground coupling. However, although the board thickness of 62 mils and the layer spacing are reduced, it is still not easy to control the distance between the main power supply and the ground layer to be very small. Compared with the first scheme and the second scheme, the cost of the second scheme is greatly increased. Therefore, we usually choose the first option when stacking. When designing, follow the 20H rule and mirror layer rule design.
Stack-up of four-layer or eight-layer
1. This is not a good stacking method due to poor electromagnetic absorption capacity and large power supply impedance. Its structure is as follows:
1.Signal 1 component surface, microstrip routing layer
2.Signal 2 internal microstrip routing layer, better routing layer (X direction)
3.Ground
4.Signal 3 strip line routing layer, better routing layer (Y direction)
5.Signal 4 stripline routing layer
6.Power
7.Signal 5 internal microstrip routing layer
8.Signal 6 microstrip trace layer
2. It is a variant of the third stacking method. Because the reference layer is added, it has better EMI performance, and the characteristic impedance of each signal layer can be well controlled.
1.Signal 1 component surface, microstrip routing layer, good routing layer
2.Ground layer, better electromagnetic wave absorption ability
3.Signal 2 strip line routing layer, good routing layer
4.Power layer, forming excellent electromagnetic absorption with the underlying ground layer
5.Ground layer
6.Signal 3 strip line routing layer, good routing layer
7.Ground layer with large power impedance
8.Signal 4 microstrip trace layer, good trace layer
3. The best stack-up method, because of the use of multiple ground reference layers, it has very good geomagnetic absorption capacity.
1.Signal 1 component surface, microstrip routing layer, good routing layer
2.Ground layer, better electromagnetic wave absorption ability
3.Signal 2 strip line routing layer, good routing layer
4.Power layer, forming excellent electromagnetic absorption with the underlying ground layer
5.Ground layer
6.Signal 3 strip line routing layer, good routing layer
7.Ground layer, better electromagnetic wave absorption capacity
8.Signal 4 microstrip trace layer, good trace layer
For how to choose and how to design several layers of boards and how to stack them depends on many factors such as the number of signal networks on the board, device density, PIN density, signal frequency, board size, and so on. For these factors, we need to consider them comprehensively. For the number of signal networks, the higher the device density, the higher the PIN density, and the higher the frequency of the signal, the multilayer design should be used. In order to obtain good EMI performance, it is best to ensure that each signal layer has its own reference layer.
