JP5067594B2 - EMI prediction method and apparatus - Google Patents

Description

  The present invention relates to a method and apparatus for predicting EMI (Electromagnetic Interference) radiation generated from a printed circuit board.

  In recent years, with the increase in the speed of a microprocessor or the like mounted on a multilayer printed circuit board, unnecessary electromagnetic noise caused by resonance between a power supply layer and a ground (GND) layer cannot be ignored. In order to reduce such EMI radiation at the design stage, many EMI simulation techniques for accurately grasping the radiated electromagnetic wave intensity have been proposed.

  In Japanese Patent Laid-Open No. 10-301977 (Patent Document 1), the structure of a multilayer printed circuit board is meshed, a mutual impedance between these elements is obtained by a predetermined calculation, and an equation of the moment is solved to solve each element. A method for calculating the intensity of a radiated electromagnetic field by obtaining a current and a magnetic current is disclosed.

  In JP 2002-16391 A (Patent Document 2), in addition to electromagnetic noise due to the current of the signal line, new electromagnetic noise due to the influence of the electromagnetic field formed by the power supply layer, the ground layer, and the housing is also considered. An EMI simulation apparatus for calculating electromagnetic wave noise radiated from an electronic device is disclosed.

  Japanese Patent Laid-Open No. 2001-221822 (Patent Document 3) discloses a method for obtaining an EMI value based on actual circuit characteristics without depending on equation analysis. Specifically, S parameters, reflection coefficients, standing wave ratios, and EMI values are measured for a known circuit, and the relationship is stored. Then, the S parameter, reflection coefficient, and standing wave ratio of the unknown circuit are measured, and the EMI value is estimated using the stored relation (paragraph 0012).

Japanese Patent Laid-Open No. 10-301977 JP 2002-16391 A JP 2001-221822 A

  However, in the EMI simulation disclosed in Patent Documents 1 and 2, as a basic idea, the current component flowing in the circuit is analyzed, and the electromagnetic field thereby is obtained from the wave equation to try to predict the EMI. In particular, for the power supply layer and the ground layer having a two-dimensional extension, it is necessary to model them by dividing them into a grid pattern, and it is difficult to analyze the current component.

  Although the method disclosed in Patent Document 3 does not depend on equation analysis, it measures a real circuit and estimates an EMI value, and is basically different from EMI prediction at the design stage.

  The present invention is intended to solve the above-described conventional problems. When a printed circuit board composed of a power supply layer and a ground layer is taken as an example, the power supply / ground layer structure and the space are regarded as one transmission line. As a result, the EMI radiated from the power source / ground layer to the space can be predicted in the same manner as the transmission coefficient of signals when transmission lines having different impedances are connected.

  The method according to the present invention is a method for predicting EMI radiation from a substrate structure (for example, a printed circuit board) including a plurality of conductor layers serving as a radiation source of electromagnetic noise to a space, and the characteristic impedance of the substrate structure and the space The transmission coefficient from the substrate structure to the space is calculated using the impedance of the EMI, and the EMI radiation is predicted based on the transmission coefficient.

  Preferably, the transmission coefficient is corrected in consideration of propagation of the EMI radiation to the space in a spherical wave shape, and a spectrum analysis result of electromagnetic noise of a circuit mounted on the substrate structure and the corrected transmission coefficient. And predicting EMI radiation corresponding to the distance from the substrate structure.

According to an embodiment of the present invention, a method for predicting EMI radiation from a board structure including a plurality of conductor layers serving as a radiation source of electromagnetic noise into a space models the board structure and a mounting circuit, and the board A characteristic impedance of the structure is obtained, a transmission coefficient from the board structure to the space is calculated using the characteristic impedance and the impedance of the space, and the model of the mounting circuit incorporating the model of the board structure is obtained. Based on the above, the spectrum analysis of the electromagnetic noise of the mounting circuit is executed, and the EMI radiation from the substrate structure is predicted based on the spectrum analysis result of the electromagnetic noise of the mounting circuit and the transmission coefficient. .

  As described above, according to the present invention, the transmission coefficient can be calculated from the impedance of the substrate structure and the space by regarding the transmission path as the respective transmission paths. That is, since the impedance of the space is known, it is possible to predict the EMI radiation amount by a very simple calculation simply by obtaining the characteristic impedance of the substrate structure.

  By taking into account that the EMI radiated from the substrate structure propagates in the form of a spherical wave in the space, the amount of EMI radiation can be predicted as a function of the distance from the substrate structure and the frequency.

  FIG. 1A is a schematic perspective view of a printed circuit board for explaining an embodiment of the EMI prediction method according to the present invention, and FIG. 1B is a transmission line connection for explaining the principle of EMI radiation amount prediction. FIG.

  In this embodiment, power supply noise, which is one of the main causes of EMI, is taken as an example, and the conductor layers 101 and 102 of the printed circuit board, which is a conductor layer structure, are used as a power supply layer and a ground layer, respectively. Assuming that the characteristic impedance of the power supply-ground layer structure is Zo (f) as a function of the frequency f and the impedance of the space is Za, the transmission coefficient from the printed board to the space is expressed by the following equation.

Transmission coefficient = 2Zo (f) / (Za + Zo (f)) (1)
This can be understood from the transmission coefficient when two different impedances are connected as shown in FIG.

  As is well known, when two transmission lines having impedances Z1 and Z2 are connected as shown in FIG. 1B, the ratio (transmission coefficient) of the voltage transmitted from the transmission line Z1 to Z2 is 2Z1 / (Z2 + Z1). is there.

  The relationship between the power supply-ground layer structure Zo (f) and the space Za of the printed circuit board shown in FIG. 1A is considered to be equivalent to the relationship between the transmission line Z1 and the transmission line Z2 in FIG. Can do. That is, the structure between the conductor layers 101 and 102 of the printed circuit board is regarded as a transmission line for the characteristic impedance Zo (f), and similarly, the space is regarded as a transmission line for the characteristic impedance Za. In this way, the contact between the printed circuit board and the space can be considered as the connection of the transmission line, and the transmission coefficient expression in the transmission line is the transmission coefficient from the power source-ground layer structure to the space, that is, the radiation ratio of noise. It can be used as a formula. As will be described later, it is demonstrated that there is a correlation between this transmission coefficient and EMI.

It is known that the spatial impedance Za is (μ ₀ / ε ₀ ) ^1/2 and about 377Ω when the vacuum permeability μ ₀ and the vacuum dielectric constant ε ₀ are set. Therefore, if the characteristic impedance Zo (f) of the power source / ground layer structure is known, the transmission coefficient from the power source / ground layer structure to the space can be calculated.

  The electromagnetic wave transmitted from the printed circuit board to the space propagates while spreading three-dimensionally. Therefore, it can be considered that energy spreads on the spherical surface with the radius R at the position of the distance R, and the ratio of the radiation amount can be expressed as follows using the distance R.

Radiation amount = {2Zo (f) / (Za + Zo (f))} / 4πR ² (2).

Furthermore, using the correction term A for correcting the difference in unit system,
Radiation quantity = A {2Zo (f) / (Za + Zo (f))} / 4πR ² (3)
EMI is predicted by this equation.

  FIG. 2 is a flowchart showing a procedure of an EMI prediction method according to an embodiment of the present invention. Hereinafter, each step will be described in detail.

  Step 201: Model the power supply-GND layer structure of the printed circuit board. As described above, the power supply-GND layer can be regarded as a two-dimensional transmission line that propagates power supply noise. For the modeling, commercially available analysis software that generates a simulation model from a physical structure by an analysis method such as FDTD (Finite-Difference Time-Domain) or FEM (Finite Element Method) can be used.

  Step 202: Create a circuit diagram to be mounted on a printed circuit board and a model for circuit simulation based on the mounting diagram. In general, when performing circuit simulation, the power supply and GND are treated as ideal, but in this embodiment, the behavior of power supply noise due to circuit operation is incorporated by incorporating the model of the power supply-GND layer structure generated in step 201. Can be analyzed. In addition, a capacitor (bypass capacitor) inserted between the power supply GND is very important as an element for determining the impedance between the power supply layer and the GND layer.

Step 203: The characteristic impedance of the power source-GND layer structure is obtained from the reflection characteristic S11 of the S parameter. For example, a calculation is performed by applying a signal to a certain point on the printed circuit board and observing the response at the same point. This is the same as applying the voltage Vi from the known characteristic impedance Zi and observing the reflected voltage Vr. If the characteristic impedance of the object to be measured is Z, the reflection coefficient is Vr / Vi = (Z−Zi) / (Z + Zi).
Z = Zi {1+ (Vr / Vi)} / {1- (Vr / Vi)}. A characteristic impedance as shown in FIG. 3 can be obtained by calculating Z with Zo (f) and sweeping the frequency f. FIG. 3 is a graph showing an example of the characteristic impedance Zo (f).

  Step 204: Calculate the amount of spatial transmission using equation (1). As described above, the spatial impedance Za is about 377Ω, and the characteristic impedance Zo (f) of the power source-ground layer structure is obtained in step 203 above. Therefore, the transmission characteristic shown in FIG. 4 can be obtained by calculating the transmission amount from the characteristic impedance Zo (f) shown in FIG. FIG. 4 is a graph showing an example of the transmission amount characteristic.

Step 205: The transmission amount is corrected to obtain the above equation (3). As described above, since the signal transmitted from the printed circuit board to the space propagates while spreading in a spherical wave shape, the amplitude is a reciprocal of the surface area of the sphere having the radius R by 1 / 4πR ^{2 at a} distance R from the printed circuit board. (See equation (2)). Further, since the unit of signal amplitude is volt [V] and the unit of electric field strength is [V / m], the final term (3) is generated by introducing a correction term A that absorbs the difference of this unit system. To do. The corrected radiation coefficient shown in Equation (3) represents the ease of EMI radiation from the printed circuit board to the space.

  Step 206: When a model for circuit simulation is created in the above step 202, the power supply terminal of each semiconductor element is selected as a power noise observation point when the simulation is executed, and the power noise waveform is calculated. Subsequently, spectrum analysis of power supply noise is performed from the calculated power supply noise waveform by fast Fourier transform FFT or the like.

  Step 207: The predicted value of EMI is calculated by multiplying the power generation noise spectrum data obtained in Step 206 by the radiation coefficient generated in Step 205 (addition in logarithmic notation).

  FIG. 5 is a graph showing a result of measuring EMI by applying noise having a frequency swept to the power supply-GND layer structure of the present example. It can be confirmed that the shape of the EMI measurement result is in good agreement with the shape of the transmission coefficient shown in FIG. That is, there is a correlation between the transmission amount obtained from the characteristic impedance and the EMI, and by absorbing the difference between the two using the correction term A, an accurate EMI can be predicted.

  FIG. 6 is a graph showing the result of measuring the spectrum of a 100 MHz rectangular wave (clock waveform). It shows that a harmonic component exists with respect to a fundamental frequency of 100 MHz. In general, a current for driving a signal flows between the power source and the GND layer, and the frequency component thereof is the same as the signal to be driven. For this reason, it is necessary to obtain a power supply noise spectrum by circuit simulation in consideration of harmonic components of the clock signal.

  FIG. 7 is a block diagram showing the configuration of the EMI prediction apparatus that executes this embodiment. The EMI prediction apparatus includes a power supply-GND simulation model generation unit 301 that inputs power supply-GND layer structure data of a printed circuit board and generates a simulation model of a power supply-GND layer, and inputs circuit data and mounting data. A circuit simulation model generation unit 302 that generates a circuit simulation model incorporating the simulation model. Further, the EMI prediction apparatus analyzes a power source noise spectrum from a characteristic impedance analysis unit 303 that analyzes the characteristic impedance of the printed circuit board, a radiation coefficient calculation unit 304 that calculates a radiation coefficient by correcting the transmission coefficient, and a circuit simulation model. A power supply noise spectrum analysis unit 305 is included.

  The operation of the entire EMI prediction apparatus is controlled by the control unit 306. The control unit 306 is a program control processor such as a CPU, and may execute the functions described in steps 201 to 207 by executing an EMI prediction program stored in a memory (not shown). Alternatively, similar EMI prediction can be performed by controlling the functions of the blocks 301 to 305. The predicted EMI radiation may be output by an output unit 307 such as a display unit or a printer, or may be transmitted through a communication line.

  In the above-described embodiment, the description of the case where the power supply layer-GND layer of the printed circuit board is regularly required, the present invention is not limited to this embodiment. By analyzing the impedance of the housing with respect to the reference ground, it can be applied to a metal housing surrounding the printed circuit board.

(A) is a perspective view of a schematic printed circuit board for explaining an embodiment of the EMI prediction method according to the present invention, and (B) is a transmission line connection diagram for explaining the principle of EMI radiation amount prediction. is there. It is a flowchart which shows the procedure of the EMI prediction method by one Example of this invention. It is a graph which shows an example of characteristic impedance Zo (f). It is a graph which shows an example of the transmission amount characteristic. It is a graph which shows the result of having applied the noise which swept the frequency to the power supply-GND layer structure of a present Example, and having measured EMI. It is a graph which shows the result of having measured the spectrum of the rectangular wave (clock waveform) of 100 MHz. It is a block diagram which shows the structure of the EMI prediction apparatus which performs a present Example.

Explanation of symbols

101,102 Conductor layer (power supply layer, ground layer)

Claims (6)

In a method for predicting EMI radiation from a substrate structure including a plurality of conductor layers serving as a source of electromagnetic noise into space,
Calculate the transmission coefficient from the substrate structure to the space using the characteristic impedance of the substrate structure and the impedance of the space,
Predicting the EMI radiation based on the transmission coefficient;
EMI prediction method characterized by this.   The transmission coefficient is corrected in consideration of propagation of the EMI radiation to the space in a spherical wave shape, and is based on a spectrum analysis result of electromagnetic noise of a circuit mounted on the substrate structure and the corrected transmission coefficient. The EMI prediction method according to claim 1, wherein EMI radiation is predicted according to a distance from the substrate structure. In a method for predicting EMI radiation from a substrate structure including a plurality of conductor layers serving as a source of electromagnetic noise into space,
Modeling the substrate structure and the mounting circuit;
Determining the characteristic impedance of the substrate structure;
Calculate the transmission coefficient from the substrate structure to the space using the characteristic impedance and the impedance of the space,
Based on the model of the mounting circuit incorporating the model of the substrate structure, a spectrum analysis of electromagnetic noise of the mounting circuit is performed,
Predicting EMI radiation from the substrate structure based on a spectrum analysis result of the electromagnetic noise of the mounting circuit and the transmission coefficient;
EMI prediction method characterized by this.   The transmission coefficient is corrected in consideration of propagation of the EMI radiation in a spherical shape whose radius is a distance from the surface of the substrate structure, and the EMI radiation is predicted as a function of the distance and frequency. The EMI prediction method according to claim 3. In an apparatus for predicting EMI radiation from a substrate structure including a plurality of conductor layers serving as a source of electromagnetic noise into a space,
Simulation model generating means for modeling the substrate structure and the mounting circuit;
Characteristic impedance generating means for generating characteristic impedance of the substrate structure;
Transmission coefficient generation means for calculating a transmission coefficient from the substrate structure to the space using the characteristic impedance and the impedance of the space;
Analyzing means for analyzing a spectrum of electromagnetic noise of the mounting circuit based on a model of the mounting circuit incorporating the model of the substrate structure;
EMI radiation amount calculating means for calculating the EMI radiation amount from the substrate structure based on the electromagnetic noise spectrum analysis result of the mounting circuit and the transmission coefficient;
The EMI prediction apparatus characterized by having . In a program for implementing a function of causing a computer to predict EMI radiation from a substrate structure including a plurality of conductor layers serving as a radiation source of electromagnetic noise to a space,
Modeling the substrate structure and mounting circuit;
Determining a characteristic impedance of the substrate structure;
Calculating a transmission coefficient from the substrate structure to the space using the characteristic impedance and the impedance of the space;
Performing a spectrum analysis of electromagnetic noise of the mounting circuit based on the model of the mounting circuit incorporating the model of the substrate structure;
Predicting EMI radiation from the substrate structure based on a spectrum analysis result of the electromagnetic noise of the mounting circuit and the transmission coefficient;
The program characterized by having.

Images