萝莉影视

In the previous chapter entitled "Noise propagation on printed circuit boards," we used a 3D electromagnetic field simulator to visualize the propagation of the noise current and magnetic field. We showed that the signal and the return current of noise are spread and propagated to the ground (GND), and that an induced current is generated when metal is exposed to a magnetic field, causing the magnetic field to be reradiated.
It is widely known that because the GND plays a major role in noise propagation in this manner, the shape of the GND has a significant effect on the noise. For example, the noise can be changed significantly merely by connecting the board GND to a metal chassis. In this chapter, we will continue the discussion by presenting simulation results that visualize how the GND design affects the noise.

First, we will show that the noise varies depending on the shape of the board GND.
To make it easier to see the shape of signal pattern and GND pattern at the same time, we used single-sided boards.
The model used in the simulation is shown in Figure 2-1.

A 100mm signal pattern is surrounded by a GND pattern. There are constant spacing (GND gap) between the signal pattern and the GND pattern on either side (Coplanar waveguide transmission line). We prepared different models with GND gaps of 5mm, 10mm, and 20mm. The other conditions were the same, with a signal pattern width of 3mm and an input/load port impedance of 50Ω. The input signal was at 1V (Peak), and scanned frequency range was 30MHz to 1GHz.

We simulated the effect of these GND gaps on the radiated emission. Here, we define the radiated emission as the maximum electric field strength at a distance of 3 meters away, when the board is rotated horizontally and vertically, as shown in Figure 2-2.

Figure 2-3 shows the simulation results of the radiated emission. As you can see, the radiated emission becomes stronger as the size of the GND gaps is increased.

Next, we simulated the magnetic field distribution. Figure 2-4 shows the analysis surfaces. The analysis surface (horizontal) was set at the dielectric side (bottom) of the signal pattern and GND pattern. This is because the magnetic field was stronger at the bottom than at the surface layer side (top) near the signal pattern. The analysis surface (vertical) was set at the center of the board. We set the analysis frequency to 1GHz, which had the strongest radiated emission.

The results of the magnetic field simulation are shown in Figure 2-5. When the size of the GND gaps is increased, the strength of the magnetic field also increases. This shows that even a simple change in the GND shape can change the noise.

It is widely known that, in addition to the shape of the GND pattern, the connection between the board GND and the metal chassis can affect noise. The implementation of such changes to the GND for the purpose of reducing noise is called GND improvement. Here, we installed a metal plate (GND plate) under the board and connected it to the GND pattern of the board (Grounding). The simulation model is shown in Figure 2-6. Metal spacers (diameter 6mm, length 6.6mm) were used for the connection.

Figure 2-7 shows the change to the radiated emission due to the GND improvement. As you can see, the radiated emission is improved by connecting the board GND to the GND plate.

Such GND improvement is effective at reducing noise for the following reasons:

To confirm the reason for this, we simulated the magnetic field distribution. The analysis surfaces are shown in Figure 2-8, and the simulation results are shown in Figure 2-9.

Noise can be reduced by implementing GND improvement in this way. To effectively reduce noise, it is necessary to connect the board GND and GND plate to achieve low impedance in the noise band.
As an example of poor design, we established the connection with metal wire (diameter 1 mm, length 31.6 mm) to increase the impedance. The simulation model is shown in Figure 2-10.

Figure 2-11 shows the results of the radiated emission simulation. As a result of changing the connection method, the radiated emission increased, and it was worse at 1GHz than before the change.

The magnetic field distribution in Figure 2-12 shows that the magnetic field becomes stronger due to the metal wire connection.
It is difficult to obtain the noise reduction effect by GND improvement if the board GND and GND plate are connected at a high impedance. Therefore, it is important to connect at a low impedance.
In addition, it is also effective to use a case with conductive plating instead of a metal plate.

Since single-sided boards are less suitable for achieving high-density mounting, double-sided or multilayer boards are used. Therefore, next we will use a double-sided board and discuss the effects that GND design has on noise.

The simulation model uses a microstrip line whose back surface is entirely grounded, as shown in Figure 2-13.
It is widely known that the noise can be changed simply by modifying the size or layout of the board, even when using the same circuit design. Therefore, we used three simulation models with board widths (GND widths) of 200mm, 100mm, and 50mm. The other conditions were the same, with the transmission line designed to have a characteristic impedance of 50Ω, and with the input signal at 1V and scanning at 30MHz to 1GHz.

Figure 2-14 shows the results of the radiated emission simulation. When the board width is narrowed down, the noise increases to around 430MHz.

Figure 2-15 shows the results of the magnetic field distribution simulation at 430MHz, where the radiated emission is high. These results reveal the following:

We can see that as the GND width is narrowed down, the magnetic field wraps more strongly around to the back surface of the board, resulting in a significant change in the magnetic field distribution. This indicates that it is advantageous to consolidate the circuits that have strong noise due to a high-speed signal at the center of a large board.

Next, we changed the board thickness and simulated the effect on the noise.
The simulation model is shown in Figure 2-16. We prepared three different models with board thicknesses of 1.6mm, 0.8mm, and 0.4mm. The signal pattern width was adjusted to provide a characteristic impedance of 50Ω.　This ensures that the current value remains the same even when the board thickness is changed.

Figure 2-17 shows the results of the radiated emission simulation. We can see that the radiated emission decreases as the board thickness is reduced.

From the results of the magnetic field simulation (Figure 2-18), we can see that a thinner board decreases the strength of the magnetic field.

This phenomenon can be explained by the following (as shown in Figure 2-19):

Thus, there is an increase in the strength of the magnetic flux cancellation due to the signal current and return current, and a decrease in the radiated emission.

As described earlier, the noise reduction effect can be achieved by making the board thinner and decreasing the distance between the signal pattern and GND pattern. However, because thinner boards tend to bend more easily and increase the risk of failure, it is necessary to use multilayer boards to ensure the appropriate thickness.

When a cable is connected to a board, the noise that propagates from the connection is often a problem.
Here, we will discuss how connecting a shielded cable (coaxial cable) affects the noise.

The simulation model is shown in Figure 3-1. We cut the board to align its edge with the end of the signal pattern, and connected a shielded cable. We set the characteristic impedance of the shielded cable to 50Ω, and used a 50Ω termination in the cable. We prepared four different models. One model did not have a cable, and the other three had cable lengths of 110mm, 210mm, and 410mm.

Figure 3-2 shows the results of the radiated emission simulation. As you can see, the radiated emission increases when a cable is connected. It is also clear that the resonance frequency shifts to a lower frequency as the cable length increases. This shows that the cable affects the noise radiation.

The results of the magnetic field distribution simulation (Figure 3-3) reveal the following:

These results show that the noise generated by the board propagates to the surface of the cable shield and radiates from there to generate strong radiated emission. In this case, it is necessary to shield the board, or reduce the noise by redesigning the board or implementing GND improvement.

We tested whether it is possible to reduce noise even when a shielded cable is connected, by reducing the thickness of the board whose effects we introduced earlier.
We prepared three different simulation models with board thicknesses of 1.6mm, 0.8mm, and 0.4mm, as shown in Figure 3-4.

Figure 3-5 shows the results of the radiated emission simulation, and Figure 3-6 shows the results of the magnetic field distribution simulation.
As you can see, the radiated emission decreased. The magnetic field distribution also shows a decrease in the noise generated by the board, as well as the noise propagated to the shielded cable. In other words, reducing the thickness of the board is also effective at suppressing the noise that propagates to and radiates from the cable shield.

We installed a filter to achieve further noise reduction.
The filter was a ferrite bead (BLM 600Ω at 100MHz) installed 5mm from the input, as shown in Figure 3-7.

Figure 3-8 shows the results of the radiated emission simulation, and Figure 3-9 shows the results of the magnetic field simulation.
As you can see, installing a ferrite bead decreases the strength of the magnetic field and radiated emission.
By filtering the signal line in this way, we can suppress the noise from the board, as well as the noise radiation from the cable shield.

The radiated emission can vary significantly depending on the connected cable. The method of connection between the cable shield and metal connector has a particularly significant impact.

The proper shielding effect is achieved when the entire perimeter of the cable shield is in full termination with the metal connector. However, when using a pigtail connection with metal wire, the shielding effect is reduced because a slit is formed. Noise leaks from the slit and propagates to the shield and board, thereby amplifying the radiated emission. Here, we will present the results of the simulation conducted to visualize this phenomenon.

The simulation model is shown in Figure 3-10. A 0.3mm diameter metal wire was used for the pigtail connection, with the cable shield positioned 9.5mm away from the metal connector.

Figure 3-11 shows the results of the radiated emission simulation. As you can see, the radiated emission increases when using the cable with a pigtail connection.

The results of the magnetic field distribution simulation in Figure 3-12 also show that the magnetic field leaking from the slit propagates to the shield and board, causing the magnetic field to be reradiated. In contrast, if there is no opening, this phenomenon does not occur and the magnetic field is confined.

These results demonstrate that using a pigtail connection increases the radiated emission due to noise leaking from the slit. It is necessary to use a cable that its shield is fully terminated to the metal connector.

In this chapter, we discussed how GND design can significantly influence the noise, using a 3D electromagnetic field simulator to visualize the effects.
Noise is affected by various factors such as the GND width, board thickness, and cable shielding. Printed circuit boards require careful design that involves measures such as implementing GND improvement, reducing board thickness, and installing filters. It is also important to note that when implementing GND improvement, the chassis design needs to be considered from the initial stages.