JP4343145B2 - Wiring board - Google Patents

Description

  The present invention relates to a wiring board, and more particularly to a wiring board that reduces radiation noise.

  With the increase in the volume of data handled by digital electronic devices such as personal computers, the processing capacity has been improved. That is, the CPU clock frequency is increased, and the bus wiring, clock lines, data lines, and the like to peripheral ICs are increased in speed and density. For this reason, unnecessary radiation noise from electronic equipment, that is, EMI has become a problem.

When connecting a CPU and a plurality of memories, or driving a display from a personal computer or AV device, a large amount of data and image signals are handled, and the connection between the transmitting IC and the receiving IC is 1 In many cases, there are one-to-multiple connections instead of one-to-one.
In this case, a plurality of ICs are driven by one signal wiring, and a branch wiring in which one signal wiring branches into a plurality of signal wirings is used. At this time, the input impedance of each IC is usually different from the characteristic impedance of the branch wiring. For this reason, a high frequency signal is reflected in a branch location, and unnecessary radiation noise may occur.

Various techniques have been disclosed in connection with branch wiring.
A method is disclosed in which not only the signal waveform is disturbed but also EMI is reduced in the driving device of the flat display device by equalizing the impedances of the left branch wiring and the right branch wiring extending in the left-right direction from the branch point ( Patent Document 1).
A meander line technology that configures a delay circuit by folding a wiring several times is disclosed (see Non-Patent Document 1). There is a non-pass band at a frequency where the coupling length of the meander line is λ (wavelength) / 2, and radiation increases at this frequency.
JP 2002-169519 A Toshio Sudo, Jun-ichi Kudo, Hakushishiba, Kenji Ito, “Electrical characteristics of differential transmission lines and meander lines”, Journal of Japan Institute of Electronics Packaging, Vol. 4, No. 7, November 2001

Patent Document 1 discloses only the EMI reduction method when the signal input position to the branch wiring is in the center, and does not disclose the EMI reduction method when the input position is at the end. Non-Patent Document 1 does not disclose the EMI reduction method itself.
There are cases where the user can determine the input position of the signal to the branch wiring and cannot determine it. For example, when the transmission side IC is mounted on the wiring board, the input position can be changed by changing the arrangement of the ICs. On the other hand, if the position of the transmission side IC on the wiring board cannot be changed for some reason, or if the transmission side IC is on a separate board and the connector position for connecting them is determined, the input position is changed. Is difficult.
In view of the above, an object of the present invention is to provide a wiring board capable of reducing radiation noise (EMI) regardless of the input position of a signal to a branch wiring.

  A wiring board according to the present invention includes a flat plate electrode, an insulating substrate disposed on the flat plate electrode, a first wiring disposed on the insulating substrate, and the first wiring on the insulating substrate in proximity to the first wiring. And a second wiring connected to the first wiring, and a plurality of third wirings connected to different portions of the second wiring. And

  ADVANTAGE OF THE INVENTION According to this invention, the wiring board which can reduce radiation noise (EMI) can be provided irrespective of the input position of the signal to branch wiring.

(First embodiment)
FIG. 1 is a top view showing a wiring board 10 according to the first embodiment of the present invention. FIG. 2 is a cross-sectional view illustrating a state in which the wiring board 10 is cut along the line A1-A2 of FIG.
The wiring board 10 is a two-layer board, and the boards 11 and 12 and the protective layer 13 are laminated in this order. Note that FIG. 1 shows a state in which the substrates 11 and 12 and the protective layer 13 are excluded for easy understanding.

  As shown in FIG. 2, the wiring board 10 can be divided into members 16 to 18 by boundaries 14 and 15. The materials of the members 16 to 18 may be the same or different. In FIG. 2, the layers of the members 16 to 18 (substrates 11 and 12, protective layer 13) are arranged to have the same depth, but the thickness of each of the members 16 to 18 is changed, The layers of the members 16 to 18 can be discontinuous.

For example, the member 16 is an abbreviation of FR4 (Flame Retardant Type 4 used for printed circuit boards, a flame-retardant printed circuit board material made of a composite material of glass fiber and epoxy resin) or an FPC (Flexible Printed Circuit) circuit board. The polyimide material can be used. Corresponding to the former and the latter, the member 16 is constituted by a printed board or an FPC board.
The member 17 can be composed of an FPC board. In this case, when the member 16 is formed of an FPC board, the members 16 and 17 are formed of a continuous FPC board. Further, when the member 16 is constituted by a printed board, the members 16 and 17 are constituted by a printed board and an FPC board connected to each other.
The member 18 may be configured by any of a printed board, an FPC board, and a glass board.

On the wiring board 10, N (for example, six in FIG. 1) receiving-side ICs 21 (21 (1) to 21 (6)) are arranged. The receiving-side IC 21 includes signal pads 22 (22 (1) to 22 (6)) and ground pads 23 (23 (1) to 23 (6)), and a ground surface (a plate electrode for ground) 24. It arrange | positions facing (1) -24 (6).
Signal wirings 25 (1) to 25 (6) are connected to the signal pads 22 (1) to 22 (6), respectively. In addition, ground wirings 26 (1) to 26 (6) are connected to the ground pads 23 (1) to 23 (6), respectively.
The signal wiring 25 and the ground wiring 26 are connected to the receiving side IC 21 via the members 17 and 18. The ground plane 24 as a backplane is not essential and may be omitted.

The signal wirings 25 (1) to 25 (6) are connected to the signal wiring 31. That is, the signal wiring 31 is a branch wiring having a branch point branched to the signal wirings 25 (1) to 25 (6).
The signal wiring 31 is connected to the signal wiring 33 at the connection portion 32. The signal wiring 33 is disposed substantially parallel to the signal wiring 31.

The connection part 32 is a signal wiring itself and is connected to both ends of the signal wirings 31 and 33. The connection part 32 can be considered as a turning point of the wiring when the signal wirings 31 and 33 and the whole connection part 32 are considered as one wiring. Since the signal wirings 31 and 33 are folded back, radiation noise from the wiring board 10 can be reduced. Details of this will be described later.
The signal wiring 33 is connected to the connector-side signal pad 35 via the signal wiring 34. The signal wirings 31, 33, and 34 are disposed on the substrate 11.

  The ground wirings 26 (1) to 26 (6) are connected to the ground plane (ground plate electrode) 41 by vias 27 (1) to 27 (6). The ground surface 41 is electrically connected to the ground surface 42 and further to the ground surface 43 (ground plate electrode) via wirings 44 and 45. A connector-side ground pad 46 is connected to the ground surface 43. The ground surfaces 24 and 41 to 43 are disposed at the boundary between the substrates 11 and 12.

  Between the ground surfaces 24 and 41 to 43, gaps (slits) 47, 48, and 49 where no ground surface is formed are disposed. As the signal wirings 25 and 34 pass through the gaps 47, 48 and 49, radiation from the signal wirings 25 and 34 can occur.

  The signal wirings 31 and 33 are opposed to the ground surface 41 through the substrate 12 and function as so-called strip lines. The signal wiring 34 faces the ground planes 41 to 43 through the substrate 12 and functions as a strip line.

In the present embodiment, the wiring board 10 has the following characteristics.
(1) The wiring board 10 is an interface board composed of boards 11 and 12 stacked one above the other.
(2) The signal wirings 31, 33, and 34 are high-speed differential signal wiring groups that are arranged on the substrate 12 and have a microstrip line structure on the surface thereof. On the other hand, the ground planes 41 to 43 have a reference potential and are arranged between the substrates 11 and 12.
(3) The signal wiring 31 is branched to signal wirings 25 (1) to 25 (6) corresponding to the receiving side ICs 21 (1) to 21 (6), respectively.
(4) The signal wirings 31 and 33 are substantially parallel to each other and are arranged close to each other so that the distance S between the signal wirings satisfies S <2W with respect to the width W of the wirings. In general, W> S because of ease of manufacturing.

(Comparative example)
A wiring board according to a comparative example of the embodiment of the present invention will be described.
3 to 5 are top views showing wiring boards 50, 50a, and 50b according to comparative examples of the present invention, respectively, and correspond to FIG.

  In the wiring boards 50, 50 a and 50 b, the signal wiring 31 is directly connected to the signal wiring 34 at the connection portion 36. That is, it differs from the wiring board 10 in that it does not have the signal wiring 33 arranged substantially parallel to the signal wiring 31. In other words, the signal wirings 31 and 34 are not folded back.

  The connecting portion 36 can be said to be a signal input position from the signal wiring 34 to the signal wiring 31 or a branch position. In each of the wiring boards 50, 50 a, 50 b, the connection part 36 is between the receiving ICs 21 (1), 21 (2), between the receiving ICs 21 (2), 21 (3), and between the receiving ICs 21 (3), 21 (4 ). That is, the branching ratio in each of the wiring boards 50, 50a, and 50b is 5: 1, 4: 2, and 3: 3.

FIG. 6 is a graph showing an EMI simulation result in each of the wiring boards 50, 50a, and 50b. Each of the graphs G, H, and I in FIG. 6 corresponds to the result of EMI measurement at a branching ratio of 5: 1, 4: 2, and 3: 3. The horizontal axis in FIG. 6 corresponds to the frequency, and the vertical axis corresponds to the radiation intensity. The 3m method was used as a simulation method.
As a result of the EMI measurement, the maximum value of the radiation intensity increased in the following order.
Radiation intensity at a branching ratio of 5: 1> Radiation intensity at a branching ratio of 4: 2> Radiation intensity at a branching ratio of 3: 3 That is, the branching ratio of 5: 1 is better than that of the branching ratio of 3: 3. It is necessary to reduce unnecessary radiation noise.

  As a method of reducing unnecessary radiation noise on a substrate having a branch wiring, there is a method of reducing reflection due to discontinuity of the characteristic impedance of the wiring. The characteristic impedance discontinuity occurs in the connector portion, the gaps 47 to 49, the input pins of the receiving side IC 21, and the like. When a signal is reflected by this discontinuity, the reflected signal waveform is added to the original signal waveform, resulting in signal wiring noise. Since this noise flows as a return current not only in the signal wirings 31, 33 and 34 but also in the ground surfaces 41 to 43, radiation increases.

In the case of the branching ratio (5: 1), the reason why the radiation increases due to the reflection due to the above characteristic impedance discontinuity will be described.
When reflection occurs at the input portion to the ICs 21 (1) to 21 (6) in FIG. 3, reflection occurs not only in the signal wiring 25 but also in the ground wiring 26. If there is a gap 47 between the ground surface 24 of the receiving-side IC 21 (which may not be present) and the ground surface 41 with the branch wiring 31, a loop is formed between the signal wiring 25 and the ground wiring 26. Current is generated and radiation increases.
Reflection occurs at each of the branch portions (connection portions of the signal wirings 25 and 31). For this reason, when the connection part 36 is at the end of the signal wiring 31, the input signals to the ICs 21 (1) to 21 (6) are added, and different waveforms are obtained at the branch parts (connection points of the signal wirings 25 and 31). Expected to have.

FIG. 7 is a graph showing a signal waveform at the signal pad 22 of the receiving side IC 21. FIG. 8 is a graph showing a ground waveform at the ground pad 23 of the receiving side IC 21. These represent SPICE simulation results.
In the signal waveform at each of the input terminals 1, 4, 6 (connection portions between the signal wirings 25 (1), 25 (4), and 25 (6) and the signal wiring 31), the signal waveform from the branch point (connection portion 36) The distant input end 1 has small overshoot and undershoot, and the input end 6 closest to the branch point (connection portion 36) has large overshoot and undershoot.
From this point, the closer to the one-to-a plurality of branch points (connections 36) as in the input end 1, the greater the disturbance of the waveform by adding the reflected waveforms at all the input ends 1 to 6. I understand.

In the ground waveform, as in the signal waveform, the input 1 farthest from the branch point (connector 36) has a small overshoot and undershoot, and the input 6 closest to the branch point (connector 36) has a large overshoot and undershoot.
Since ground lines having different voltages from the input terminals 1 to 6 are connected to the ground surface 41 having a long horizontal length, a voltage difference is generated between them, and a high-frequency current is generated on the ground surface. Since there is no further ground plane to cancel out these high-frequency currents, they become common mode noise and cause large radiation in the far field.

FIG. 9 and FIG. 10 are graphs showing FFT (Fast Fourier Transforms) waveforms of signals at the signal pad 22 and the ground pad 23 of the receiving side IC 21 at a branching ratio of 5: 1, respectively. This is for examining the frequency characteristics of the waveform at the input part and the ground part in the receiving-side IC 21.
The vertical axis in FIG. 9 is represented by voltage for convenience, but the value obtained by dividing this by the period of the repetitive pulse is the intensity of the original FFT waveform, that is, the amplitude for each frequency. The value obtained by dividing the voltage value on the vertical axis in FIG. 9 by the period of the repetitive pulse of 30 ns is the amplitude value for each frequency.
This situation is the same in the other FFT waveforms shown in FIGS. 10 to 12, 17, 18, 32, and 33, and the value obtained by dividing the voltage value on the vertical axis by the repetition pulse period of 30 ns is the amplitude for each frequency. Value.

  As shown in FIG. 9, the first harmonic and the third harmonic are large in the signal pad 22. As shown in FIG. 10, the ground pad 23 has a large third-order harmonic and a large distribution difference depending on the position of the input end. In the case of branching (5: 1), it is expected that radiation is reduced by reducing the potential difference at each branching point of the ground surface 41.

FIGS. 11 and 12 are graphs showing FFT (Fast Fourier Transforms) waveforms of signals at the signal pad 22 and the ground pad 23 of the receiving side IC 21 at a branching ratio of 3: 3 (FIG. 5), respectively. This is for examining the current gradient on the ground plane 41 when the branching ratio is 3: 3.
11 and 12, when the branching ratio is 3: 3, the signal pad 22 and the ground pad 23 are both at the input end 4 near the branch point (connection portion 36) and at the input end 6 far from the branch point. The frequency dependence of these signals is almost the same.

In the EMI simulation result shown in FIG. 6, the branch (3: 3) emits less radiation than the branch (5: 1). Since the branch (3: 3) does not generate a current distribution on the ground surface 41 than the branch (5: 1), it is considered that unnecessary radiation noise is reduced. By making the ground potential flowing into the ground planes 41 to 43 substantially equal, it is predicted that no current flows through the ground planes 41 to 43 and radiation is reduced.
Further, when the rising and falling edges of the signal become large, noise increases due to harmonic components.

Hereinafter, radiation characteristics of the wiring board 10 according to the present embodiment will be described.
A. Effect of Folding Consider why the EMI is reduced by folding the wirings 31 and 33.
As shown in FIG. 2, by folding the wirings 31 and 33, signal wirings having substantially the same signal waveform and reverse current directions come close to each other.
FIG. 13 is a schematic diagram showing a distribution of electromagnetic fields formed around the currents flowing in the wirings 31 and 33.
Since the directions of the electromagnetic fields 61 and 62 in the wirings 31 and 33 are reversed, the electromagnetic fields cancel each other in the far field, and unnecessary radiation noise can be reduced.

  As an effect of the folded wiring, there is an effect that the potentials on the ground planes at the respective branch portions are substantially equal, high frequency currents do not flow on the ground planes, and unnecessary radiation noise can be reduced. Details of this are described in C.I. Will be described later.

A reduction effect by placing the folded wirings 31 and 33 close to each other will be described. In a structure in which the signal wirings 31 and 33 are brought close to each other with the wiring width W being 100 μm and the inter-wiring distance S being 100 μm, the coupled wiring length (the distance that the wiring 31 and the wiring 33 run in parallel, also referred to as the coupling length) L _coup Is compared with the case where the wiring is not folded. The distance L _all from the branching portion 1 to the branching portion 6 is the same in both cases where the wiring is folded back. The distance L _all corresponds to the line length (wiring length) of the longer one of the wirings 31 and 33 (wiring 31).

  FIG. 14 is a graph showing an EMI simulation result when the wiring is not folded (graph G) and when the wiring is folded (graph K) at the branch 5: 1. By turning back the wiring, 3 dB at 100 MHz and 8 dB at 200 MHz are reduced.

  FIGS. 15 and 16 are graphs showing signal waveforms at the signal pad 22 and the ground pad 23, respectively, and correspond to FIGS. When FIG. 7 is compared with FIG. 15, the difference in amplitude due to the branching portion is smaller in FIG. Similarly, comparing FIG. 8 and FIG. 16, which are ground waveforms, FIG. 16 shows a smaller difference in amplitude due to the branching portion.

FIGS. 17 and 18 are graphs showing the FFT results of the signal waveforms at the signal pad 22 and the ground pad 23, and correspond to FIGS. 9 and 10, respectively. When FIG. 9 and FIG. 17 are compared, the amplitude in FIG. 17 is smaller at 200 MHz or more. Also in the ground waveform, when FIG. 10 is compared with FIG. 18, the amplitude of 200 MHz or more is smaller in FIG.
As described above, it has been found that radiation noise can be reduced by folding signal wiring.

B. Influence of distance between wirings It is desirable that the wirings 31 and 33 interfere with each other. According to Non-Patent Document 1, the even-mode characteristic impedance and the odd-mode characteristic impedance differ depending on the distance between the wires. Here, the dielectric width, the dielectric constant of the dielectric, and the thickness of the wiring are all the same. As a condition that the two types of characteristic impedances have different values, in the case of a microstrip line, the relationship between the wiring width W and the inter-wiring distance S (the distance from the edge of the wiring 31 to the edge of the wiring 33) is S <2W. It ’s time.

FIG. 19 is a graph showing a simulation result of the EMI horizontal component when the inter-wiring distance S between the wirings 31 and 33 is changed to 100, 180, 200, and 900 μm on the wiring board 10.
When the inter-wiring distance S is 200 μm or more, which is twice the wiring width W, the frequency characteristics change greatly. Further, by folding, the EMI of 200 MHz or less is reduced compared to the case of not folding.
However, the EMI with an inter-wiring distance of 180 μm or less is greatly reduced at 300 MHz to 500 MHz and 700 to 900 MHz, and the EMI with an inter-wiring distance of 200 μm is reduced at 500 MHz to 700 MHz.

It is conceivable to adopt a structure with an inter-wiring distance of 200 μm or less that reduces EMI from 300 MHz to 500 MHz. This is because when the clock frequency is several tens of MHz or more, it is more effective to reduce the component with the lower harmonic component, and considering the mounting area, if the distance between the folded wires is long, the folded wires This is because different wirings must be placed between them, and desired characteristics may not be obtained.
That is, the condition (S <2W) can be adopted in consideration of the coupling between the signal wirings 31 and 33.

C. Influence of coupling length L _coup As described above, by folding back the wirings 31 and 33, the potential at the ground surface 41 at each branch portion is made substantially equal, and high-frequency current does not flow through the ground surface 41, and unnecessary radiation noise is reduced. Reduction is possible. This effect is closely related to the wiring length of the portion where the two signal wirings 31 and 33 run side by side (hereinafter referred to as the coupling length).

First, a change in transmission characteristics due to the wirings 31 and 33 being folded back will be described with reference to Non-Patent Document 1. In Non-Patent Document 1, it is assumed that there is a non-pass band at a frequency whose coupling length is equivalent to an integral multiple of λ / 2.
By extending the turning point (connection portion 32), the coupling length _Lcoup is changed, and the change in the characteristic is obtained.
FIG. 20 is a graph showing a simulation result of radiation intensity by the 3 m method when the coupling length in the case of branching (5: 1) is 115 mm and 132 mm.
When the coupling length is 115 mm, the wirings 31 and 33 are folded back at the branch portion 6. In the case of a coupling length of 132 mm, the wiring 33 is extended to a position far from the position of the branching section 6 when viewed from the branching section 1 and then turned back.

  From FIG. 20, it can be seen that the radiation with the aliasing is increased by 10 dB at 500 to 600 MHz as compared with that without the aliasing. On the other hand, the radiation is reduced by 10 dB or more compared to the one without 200-400 MHz. It can also be seen that the shorter the coupling length, the higher the frequency at which the radiation increases.

Therefore, the frequency at which the coupling length is λ / 2 is obtained. The speed of the high-frequency current is v (m / s), the frequency of the signal that becomes the non-pass band is f _{λ / 2} (Hz), the wavelength of the signal is λ (m), c is the speed of light (m / s), When the dielectric constant and epsilon _r following relation (1) is satisfied.
v = c / √ε _r = f _{λ / 2} · λ (1)
When the coupling length (also referred to as coupling wiring length) is L _coup (m), λ = 2 · L _coup is obtained, and therefore the frequency f _{λ / 2} that becomes the non-pass band can be expressed by the following equation (2).
_{f λ / 2 = c / (} √ε r · λ) = c / (√ε r · 2 · L coup) ... (2)
The above-mentioned frequency f _{λ / 2} can be expressed by the following formula when the coupling length L _coup is 115 mm (non-dielectric constant ε _r is 3.3).
f _{λ / 2} = 3 · 10 ⁸ /(3.3 ^1/2 · 115 · 2) = 716 MHz
Further, when the coupling length L _coup is 132 mm, f _{λ / 2} = 624 MHz.

Considering the reduction of the resonance frequency due to the coupling of the signal wirings 31 and 33, 500 to 600 MHz at which the radiation increases corresponds to a frequency at which the coupling length L _coup becomes λ / 2, and 200 to 200 at which the radiation is reduced. It can be said that 400 MHz corresponds to a frequency with a coupling length of λ / 4.

(1) Mechanism by which radiation increases at a coupling length λ / 2 A mechanism by which radiation increases at a frequency at which the coupling length is λ / 2 will be described.
FIG. 21 is a graph showing the current distribution in the horizontal direction of the signal wirings 31 and 33. (A) to (C) respectively correspond to the case where there is a turn, the case where there is a branch wiring, and the case where there is a turn and a branch. The horizontal direction of the graph corresponds to the position of the input end, and the right end and the left end of the graph correspond to the input ends 1 and 6, respectively. The branch part 1 is closer to the signal source that is the source of the branch.

In FIG. 21A, the wirings 31 and 33 are running side by side so that resonance occurs and a current distribution is generated. At the resonance frequency at which the coupling length is λ / 2, the current value is small because the wiring is bent in the horizontal direction (connection portion of the wirings 33 and 34) and the bending portion (connection portion 32) are nodes of current. . On the other hand, the current becomes maximum at the center of the signal wiring 31.
In FIG. 21B, the influence of reflection at the connection portion between the signal wiring 31 and the signal wirings 25 (1) to 25 (6) is considered. At the branch source (connection portion of the wirings 33 and 34), the current increases due to the addition of reflection, and the current decreases near the branch end (connection portion 32).
The graph in FIG. 21C is obtained by adding the current distributions in FIGS. 21A and 21B. Since the current imbalance of the ground plane 41 due to the current distribution of the branch wiring 31 is large and a resonance phenomenon occurs, radiation increases.

(2) Mechanism for reducing radiation at coupling length λ / 4 A mechanism for reducing radiation at a frequency at which the coupling length is λ / 4 will be described.
FIG. 22 is a graph showing the current distribution of the wiring portions in the horizontal direction of the wirings 31 and 33. (A) to (C) respectively correspond to the case where there is a turn, the case where there is a branch wiring, and the case where there is a turn and a branch. The horizontal direction of the graph corresponds to the position of the input end, and the right end and the left end of the graph correspond to the input ends 1 and 6, respectively.

In FIG. 22A, resonance occurs because the wirings 31 and 33 are running side by side, and current distribution is generated. At a frequency where the coupling length is λ / 4, the current is the smallest at the portion that is bent in the horizontal direction (connection portion of the wirings 33 and 34), and the current is maximized at the bent portion (connection portion 32) that is the substrate end.
In FIG. 22B, the influence of reflection at the connection portion between the signal wiring 31 and the signal wirings 25 (1) to 25 (6) is considered. At the branch source (connection portion of the wirings 33 and 34), the current increases due to the addition of reflection, and the current decreases near the branch end (connection portion 32).
The graph in FIG. 22C is obtained by adding the current distributions in FIGS. 22A and 22B. The current imbalance in the ground plane 41 due to the current distribution in the branch wiring 31 is reduced, the current distribution is not biased, and radiation is reduced.

(3) Prevention of radiation increase when the coupling length is λ / 2 Next, a method for preventing radiation from increasing when the coupling length is λ / 2 will be described.
In particular, high frequency noise is caused by harmonic components due to the steep rise and fall times of the clock signal (reference: Clayton R. Paul, “Introduction to Electromagnetic radiation”, pp. 357-402, mimatsu). -data-system, JAPAN, 1994).
Therefore, in the harmonic component of the clock frequency, if the frequency component having a coupling length of λ / 2 wavelength is sufficiently small, the noise source is eliminated, so that radiation can be reduced.

In general, when the harmonic order n increases, the intensity of the harmonic component, and hence radiation due to the harmonic component, tends to decrease. For this reason, it becomes possible to set an upper limit to the order n of the harmonics that require measures to reduce radiation. For example, radiation reduction measures up to harmonic components of order n (n <15, n ≠ 20) are considered. At this time, even when resonance occurs at a coupling length L _coup of λ / 2 at the frequency of the n = 20th harmonic, if the harmonic component of the clock that is the signal source is sufficiently small, the increase in radiation will occur. It is suppressed.

(4) Appropriate range of bond length If the bond length L _coup deviates from an appropriate range, that is, if the bond length L _coup is too long or too short, it becomes difficult to suppress radiation.
If the coupling length L _coup is too long, the frequency at which the radiation increases due to the λ / 2 resonance is lowered, and there is a possibility that the radiation increases in a frequency band having a large harmonic component. That is, it is assumed that the λ / 2 resonance occurs at a frequency equal to or higher than the nth harmonic (= n · f _ck ) of the clock signal (frequency f _ck ). At this time, if the signal intensity of the nth-order harmonic component is sufficiently small, radiation by the nth-order harmonic component is suppressed.

The frequency relationship at this time is expressed by the following expression based on the frequency f _{λ / 2} of the λ / 2 resonance, the speed of light c, the relative permittivity ε _r , the harmonic order n, and the frequency f _{ck of the} clock signal. The
_{f λ / 2 = c / (} √ε r · 2 · L coup)> n · f ck
When this equation is transformed with the coupling length L _coup , the following equation (3) is derived.
L _coup <c / (2 · n · f _ck √ε _r ) (3)
When the frequency f _ck = 33 MHz of the clock signal, the order n = 20, and the relative dielectric constant ε _r = 3.3 in the expression (3), the following expression is derived.
L _coup <124mm
In this embodiment, the coupling length L _coup is 115 mm, and the above formula is satisfied.

As described above, the (n / 2) -th order and the n-th order of the clock frequency are the frequency at which radiation is desired to be reduced and the frequency at which radiation is increased. When the n-th order harmonic is suppressed and the signal intensity is sufficiently small, the upper limit of the coupling length L _coup can be set by the above-described equation (3). That is, when the signal strength of the n-th order harmonic component is somewhat small (condition 1), on the other hand, when the signal strength of the (n / 2) -order harmonic component is large (for example, twice) (condition 2) ), Radiation due to harmonics can be effectively suppressed.

  According to the experiment, the results satisfying these conditions 1 and 2 were obtained for the harmonic components of the 20th order and higher. As an example, this is shown by input waveforms 1 and 2 shown in FIGS. FIG. 32 and FIG. 33 described later are graphs showing the results of FFT transform (Fast Fourier Transform) of the input waveforms 1 and 2, respectively. At this time, if the clock frequency is 33 MHz, the 10th harmonic is 330 MHz, and the 20th harmonic that is twice that is 660 MHz.

  Details of the condition 1 will be described later as “conditions for preventing the radiation intensity from exceeding the regulation value”. From FIG. 32 and FIG. 33, it is recognized that the signal intensity is clearly reduced after 500 MHz.

Next, condition 2 will be described. The signal intensity (voltage) of the component in the vicinity of the 9th and 18th harmonic frequencies is obtained from FIGS. Here, the value of the actual component is shown by dividing the directly read value in the graph by 30 ns. Since the 9th harmonic component was larger than the 10th harmonic component and the 18th harmonic component was larger than the 20th harmonic component, the 9th and 18th orders were adopted instead of the 10th and 20th orders. It was decided to.
Input waveform 1 Signal strength at 9th harmonic (300 MHz): 0.0260 V
Signal strength at the 18th harmonic (600 MHz): 0.0071V
Input waveform 2 Signal strength at 9th harmonic (300 MHz): 0.0065V
Signal strength at the 18th harmonic (600 MHz): 0.00078 V

From these values, the signal intensity ratio R of the (n / 2) th harmonic component to the nth harmonic component (voltage at the (n / 2) th harmonic component / voltage at the nth harmonic component) is obtained. And the following.
Signal intensity ratio R in input waveform 1 3.66
Signal strength ratio R at input waveform 2 R 8.33
The obtained signal intensity ratio R is 2 or more and satisfies the condition 2. That is, by setting the harmonic order n in the input waveforms 1 and 2 to around 20, the suppression of radiation at the harmonic component of the order (n / 2 = 10) becomes effective.
If the condition 2 is satisfied, even if the condition 1 is not satisfied, the radiation suppression with the (n / 2) -order harmonic component is still effective.

As described above, when the order n is 20 or more, suppression of radiation by harmonic components in the vicinity of the (n / 2) order is effective. When the order n is slightly smaller than 20, for example, even in the range of 15 or more, it is effective to reduce the radiation with the harmonic component near the (n / 2) order.
Here, when the order n is increased, the signal intensity itself at the order (n / 2) is reduced, and there is a possibility that the suppression of the (n / 2) order radiation itself is not necessary.
Considering this, the order n in a practical range can be, for example, 15 or more and 25 or less.

・ If the coupling length L _coup is too short, the frequency band in which the radiation is reduced by λ / 4 resonance shifts to a high frequency, the frequency becomes higher than the frequency at which the radiation is desired to be reduced, and the radiation reduction effect may be reduced. is there.
If the coupling length L _coup is too short compared to the longer wiring length L _all of the branch wirings 31 and 33, the region where the electromagnetic field distribution is closed by the coupling is short. For this reason, the effect of unbalance of the return current appears, and the effect of reducing unnecessary radiation noise is small. Therefore, when the coupling length L _coup is longer than ½ of the total wiring length L _all, an effect of reducing unnecessary radiation noise appears. That is, it is desirable that the relationship is expressed by the formula (4).
L _coup > L _all / 2 (4)

To confirm the above, the coupling length L _coup is 9/10 of the wiring length L _all (FIG. 1), 7/10 (FIG. 23), and 5/10 (FIG. 24). The radiation intensity in the 3m method is compared with the case (FIGS. 3 to 5).
As a result, when the coupling length L _coup is 9/10 of the wiring length L _all (FIG. 14) and 7/10 (FIG. 25), the radiation intensity on the low frequency side is reduced when there is a turn. . However, in the case of 5/10 (FIG. 26), the radiation intensity on the low frequency side is increased. From the above, the formula (4) was confirmed.

(5) Prediction of resonance frequency It is effective to predict the frequency at which λ / 2 resonance or λ / 4 resonance occurs. In brief, resonance satisfying λ / 2 occurs when the expression (2) is satisfied. However, in reality, resonance occurs at a lower frequency due to the coupling effect of the signal wirings 31 and 33. Therefore, the relationship between the coupling length and the characteristic impedance is obtained by simulation. A characteristic impedance Z between the signal pad 35 before branching and the ground pad 46 in the wiring board 10 (FIG. 1) is obtained by simulation by SPICE. The reflection characteristics Z11 are obtained, and their phases are shown in FIG.

From FIG. 27, when the frequency is sufficiently low, the phase is about −100 degrees or less. When the coupling length L _coup is 115 mm, λ / 4 resonance (phase is about 80 degrees) occurs at a frequency of 320 MHz, and λ / 2 resonance (second time of −100 degrees) occurs at 590 MHz. When the coupling length is 132 mm, λ / 4 resonance (80 degrees) occurs at a frequency of 260 MHz, and λ / 2 resonance (second time of −100 degrees) occurs at a frequency of 550 MHz.
In order to suppress the low frequency of λ / 2 resonance in which EMI increases, it is preferable to select 115 mm having a short coupling length. As described above, since the frequency characteristics vary depending on the coupling length, when actually used, the characteristic impedance is simulated, the frequency at which the EMI reduction effect occurs, or the frequency at which the EMI increases is predicted, and at the frequency at which the EMI increases. It is effective to eliminate the noise source.

(Second Embodiment)
FIG. 28 is a schematic diagram showing a wiring board 10d according to the second embodiment of the present patent. In this figure, a chip component (high frequency suppression component) 38 that suppresses harmonic components is inserted in the input section. As this chip component, a resistor, an inductor (for example, a ferrite bead inductor), an LC filter, or the like can be considered. The high frequency suppression component 38 may suppress the harmonic component in which λ / 2 resonance occurs. The reference frequency f _{λ / 2 at} this time is expressed by the following equation from Equation (3).
f _{λ / 2} = c / (√ε _r · 2 · L _coup )

As described above, in order to reduce the harmonic component, for example, a chip component having a characteristic that the characteristic impedance Z increases at least in the vicinity of the reference frequency _{fλ / 2} as shown in FIG. 29 is desirable.
FIG. 30 shows an input waveform (input waveform 1) before insertion of the high-frequency suppressing component 38. Next, FIG. 31 shows an input waveform (input waveform 2) after the harmonic suppression chip component is inserted. 32 and 33 are graphs showing the results of FFT transform (Fast Fourier Transform) on the input waveforms 1 and 2, respectively.
The signal intensity of the harmonic component at a frequency of 400 MHz is 30 times larger in the case of FIG. 32 than in the case of FIG. From FIG. 34, the radiation intensity at 400 MHz is 55 [dBμV / m] in the input waveform 1 and 35 [dBμV / m] in the input waveform 2.

-Conditions for the radiation intensity not exceeding the regulation value Conditions for the radiation intensity in the far field from the signal wirings 31 and 33 of the wiring board 10 not exceeding the regulation value will be described.
In general, a case is considered in which a pair of conducting wires having a length L are arranged in parallel at an interval s0, and currents _ID flowing in opposite directions are flowing at a frequency f. The maximum value E _{d, max} of the radiation intensity (electric field strength) at a point (measurement point) at a distance d from the pair of conductors is expressed by the following equation (5) (reference: Clayton R. Paul, “EMC”). "Introduction", Mimatsu Data System, February 1996, p435 to p437).
| E _{d, max} | = 6.316 × 10 ⁻¹⁴ · I _D · f ² · L · s0 / d (5)

It is considered that this equation (5) is applied to the radiation intensity from one of the signal wirings 31 and 33 of the wiring board 10 in FIG.
In the wiring board 10, since the signal wirings 31 and 33 and the ground 41 are opposed to each other, a mirror effect (mirror image effect) is generated by the ground 41. Since a mirror image of the signal wirings 31 and 33 appears across the ground 41, this mirror image also affects the maximum value E _{d, max} of the radiation intensity. That is, Expression (5) can be used to calculate the electric field intensity generated by any one of the signal wirings 31 and 33 and its mirror image.

The wiring board 10 has signal wirings 31 and 33 and a total of four signal wirings of these mirror images. On the other hand, Expression (5) represents the intensity of the electric field generated by the two conductive wires (wirings), and is not necessarily an accurate model of the signal wirings 31 and 33 and their mirror images.
However, equation (5) can be used as a measure of the radiation intensity from the wiring board 10. As described above, in the wiring board 10, the signal lines 31 and 33 are arranged in parallel to reduce the radiation intensity (electric field intensity). For this reason, the electric field strength from the wiring board 10 is smaller than the electric field strength calculated by the equation (5). That is, the electric field strength value of the formula (5) can be used as a standard of the electric field strength from the wiring board 10 when the conditions are bad.

From the distance (thickness of the substrate 12) S _k between the signal wiring 31 and the ground 41 and the relative dielectric constant ε _r of the substrate 12, the distance s0 between the signal wiring 31 and its mirror image is expressed by the following equation.
s = s _k × √ε × 2
Substituting this equation into equation (6) leads to the following equation (6).
| E _{d, max} | = 2.632 × 10 ⁻¹⁴ · I _D · f ² · L · √ε _r · s _k / d (6)
At this time, the length L of the conducting wire in Expression (7) corresponds to the wiring length L _all (the length in the horizontal direction from end to end of the branch wiring 31 in FIG. 1).

In order to set the electric field intensity | E _{d, max} | to 1/10 of the electric field intensity regulation value E _reg , it is necessary to satisfy the following formula (7).
I _D <(d · E _reg /10)/(2.632×10 ⁻¹⁴ · f ² · L · √ε _r · s _k )
...... (7)

The electric field strength regulation value E _reg of 40 dB μV / m corresponds to 100 μV / m by its definition. It is considered that the electric field intensity from the wiring board 10 is suppressed to 20 dB μV / m (= 10 μV / m) which is 1/10 of the electric field intensity regulation value E _reg . 0.12m length L of the signal line 31, 180 [mu] m thickness S _k of the substrate 12, 4.8 the relative dielectric constant epsilon _r of the substrate 12, the frequency f of the clock signal 400MHz, the distance d to the measurement point 3m. At this time, if the current value _ID satisfies the following equation (8), the electric field intensity | E _{d, max} | is sufficiently smaller than the electric field intensity regulation value E _reg of 40 dBμV / m.
I _D ≦ 0.015 mA (8)

Next, the FFT amplitude of the nth-order harmonic component is obtained from an analytical expression. The amplitude C _n of the nth-order harmonic component of the clock signal is expressed by the following equation (9) (see page 385 of the aforementioned reference).
C _n = A · (τ / T) · sin (n · ω0 · τ / 2) · sin (n · ω0 · τ _r / 2) / (n ² · ω0 ² · τ · τ _rf / 4)
...... (9)
Here, the rising time τ _r , the falling time τ _f , the average value τ _{rf of} the falling time τ _r and the rising time τ _f , the period T, and the pulse width (interval of 50% of the waveform amplitude) of this clock signal Let τ.

In the input waveform 1, when τ _r = τ _f = 6 ns, period T = 30 ns, pulse width τ = 16 ns, and input amplitude 3 V, 400 MHz is a 12th harmonic, and the frequency component cn is expressed as follows.
c ₁₂ (Waveform 1) = 0.409V (11)
In the input waveform 1, when τ _r = τ _f = 14 ns, period T = 30 ns, pulse width τ = 15 ns, and input amplitude 3 V, 400 MHz is a 12th harmonic, and the frequency component cn is expressed as follows. The
c ₁₂ (Waveform 1) = 0.00026 V (12)

The relationship between the current I _D , voltage V _D , and characteristic impedance Z in the wiring 31 is expressed by the following equation.
I _D = V _D / z
Using this formula, when formulas (11) and (12) are expressed by current values and the characteristic impedance Z of the wiring is 50Ω, the following formulas (11) and (12) are derived.
I ₁₂ (waveform 1) = 0.18 mA (13)
I ₁₂ (Waveform 2) = 0.005 mA (14)

From these formulas (13) and (14), the condition of 20 dBμV / m, which is 1/10 of the electric field strength regulation value E _reg of 40 dBμV / m, is not satisfied by waveform 1 but is satisfied by waveform 2 from formula (9). Has been. In the waveform 2, it is effective in reducing the electric field strength to make the waveform blunt by adding the high frequency suppressing component 38 to the IC input section.

In order to reduce the radiated electric field from the wirings 31 and 33 to 1/10 or less of the electric field intensity regulation value E _reg at the frequency n · f _ck of the n-th harmonic, the current _ID in the wirings 31 and 33 is expressed by the equation (7). It is necessary to satisfy. At this time, if the coupling length L _coup of the folded wirings 31 and 33 is equal to or less than the frequency at which λ / 2, the radiation does not increase. For that purpose, the above-described equation (3) may be satisfied.

  As described above, in the wiring board having the branch wiring, the signal wiring is folded at the branch end. As a result, the direction of the high-frequency current in the adjacent wiring is reversed, and the radiation can be reduced by acting so that the mutual electromagnetic fields cancel each other in the far field. When the coupling length is λ / 4 resonance, unnecessary radiation noise can be further reduced by reducing the bias of the current distribution on the ground plane due to the branch wiring. It is preferable that the harmonic component of the signal wiring does not reach the frequency at which the coupling length causes λ / 2 resonance.

(Other embodiments)
Embodiments of the present invention are not limited to the above-described embodiments, and can be expanded and modified. The expanded and modified embodiments are also included in the technical scope of the present invention.
For example, in the above embodiment, the number of layers of the wiring board is 2, but the present invention can be applied even if the number of layers is 3 or more.

1 is a top view of a wiring board according to a first embodiment of the present invention. It is sectional drawing of the wiring board which concerns on the 1st Embodiment of this invention. It is a top view of the wiring board which concerns on the comparative example of this invention. It is a top view of the wiring board which concerns on the comparative example of this invention. It is a top view of the wiring board which concerns on the comparative example of this invention. It is a graph showing the frequency dependence of the radiation intensity of the wiring board which concerns on the comparative example of this invention. It is a graph showing the signal waveform in the signal pad of receiving side IC. It is a graph showing the ground waveform in the pad for ground of receiving side IC. It is a graph showing the spectrum of the signal waveform in the signal pad of receiving side IC. It is a graph showing the spectrum of the ground waveform in the pad for ground of receiving side IC. It is a graph showing the spectrum of the signal waveform in the signal pad of receiving side IC. It is a graph showing the spectrum of the ground waveform in the pad for ground of receiving side IC. It is a schematic diagram which shows distribution of the electromagnetic field which the electric current which flows into two wiring forms in the circumference | surroundings. It is a graph which shows the EMI simulation result in the case where the wiring is not folded and the wiring is folded. It is a graph showing the signal waveform in the signal pad of receiving side IC. It is a graph showing the ground waveform in the pad for ground of receiving side IC. It is a graph showing the spectrum of the signal waveform in the signal pad of receiving side IC. It is a graph showing the spectrum of the ground waveform in the pad for ground of receiving side IC. It is a graph showing the result of EMI simulation when changing the distance between wiring. It is a graph showing the result of EMI simulation when changing the bond length. It is a graph showing the current distribution of signal wiring. It is a graph showing the current distribution of signal wiring. It is a top view of the wiring board which concerns on the modification of the 1st Embodiment of this invention. It is a top view of the wiring board which concerns on the modification of the 1st Embodiment of this invention. It is a graph showing the voltage distribution of the signal wiring which concerns on the modification of the 1st Embodiment of this invention. It is a graph showing the voltage distribution of the signal wiring which concerns on the modification of the 1st Embodiment of this invention. Comparison of phase of characteristic impedance by difference of coupling length It is a top view of a wiring board according to a second embodiment of the present invention. It is a graph showing an example of the frequency characteristic of the impedance of a high frequency suppression chip component. It is a graph showing an example of the waveform before high frequency suppression chip components mounting. It is a graph showing an example of the waveform after high frequency suppression chip components mounting It is a graph showing an example of the spectrum of the waveform before high frequency suppression chip components mounting. It is a graph showing an example of the spectrum of the waveform after high frequency suppression chip components mounting. It is a graph showing the radiation intensity of a wiring board.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Wiring board | substrate, 11, 12 ... Board | substrate, 13 ... Protection layer, 21 ... Reception side IC, 22 ... Signal pad, 23 ... Ground pad, 24 ... Ground surface, 25, 31, 33, 34 ... Signal wiring, 26 ... ground wiring, 27 ... via, 32 ... connector, 35 ... connector side signal pad, 41-43 ... ground surface, 44,45 ... wiring, 46 ... connector side ground pad, 47-49 ... gap

Claims (4)

A plate electrode;
An insulating substrate disposed on the plate electrode;
A first wiring disposed on the insulating substrate;
A second wiring which is disposed the first and substantially parallel lines on the insulating substrate,
A plurality of third wirings connected to different portions of the second wiring ;
At the ends of a plurality of connection locations of the second wiring and the plurality of third wirings, the first and second wirings are folded and connected,
The distance S between the first and second wirings is smaller than twice the width W of the first and second wirings (2 · W) (S <2 · W);
The coupling length L _coup of the first and second wirings is derived from the clock frequency f _{ck of the} signal transmitted by the first and second wirings, the relative dielectric constant ε _{r of the} insulating substrate , the natural number n, and the speed of light c. Smaller than the calculated value [c / (2 · n · f _ck · √ε _r )] (L _coup <[c / (2 · n · f _ck · √ε _r )]),
The coupling length L _coup is larger than 1/2 (L _all / 2) of the longer line length L _all of the first and second wirings (L _coup > L _all / 2).
A wiring board characterized by that. Wiring board according to claim 1 Symbol placing the natural number n, characterized in that 15 or more. The wiring board according to claim 1, further comprising: a harmonic suppression element that suppresses a predetermined harmonic component of a signal that is connected to the first wiring and transmitted on the first wiring. . The harmonic suppression element includes a reference frequency [c / (√ε _r · 2 · L] calculated from the coupling length L _coup of the first and second wirings, the relative dielectric constant ε _r of the insulating substrate, and the speed of light c. wiring board according to claim 3 Symbol mounting, characterized in that impedance increases in the vicinity of the _coup)].

Images