26th International Spring Seminar on Electronics Technology May 8 - 1I , 2003, Sara Lesnb, Slovak Republic
Signal Integrity Face to Face with EMC in PCB Design
Dan Pitica, Serban Lungy Ovidiu Pop
Applied Electronics Department, Technical University of Cluj-Napoca, Cluj-Napoca, 3400, Romania Phone: +402 64 194806, e-mail: Dan.Pitica@ael.utcluj.ro
The paper presents a review of the most important phenomena thot determine the Signal Integrity and EMC
on the signal traces for the PCB equipped with digital circuits. The aim o thispaper is ta highlight the effects f
o variation of design parameters. signal integrity and EMCfield For all o the mentionedphenomena the f f mathematical models are specified. These models can be used in order to characterize the PCB signal traces and the way in which these models can be implemented in Matlab program. The Matlab simulated results ond other elechonics circuit simulator results (Pspice. Omego+) are compared Based on this comparisan, some observations about modelpe$onnances and limitations can be done. Using these results. the paper analyses the efect tho1 appears in Signal Integrity assurance and EMC. based on variation of some signal baces parameters. The conclusions of this paper are recommendation for signal traces design when digital circuits ore used.
In the educational process it is essential to use alternative tools in order to analyse complex phenomena that appear in electronic circuits operation. Thus, graphical methods allow obtaining fast results, without affecting the results pnrport. Using analytical methods, the phenomena description becomes more complex but the results describe reality more truthful. If the simulation programs are used, the effort is focused on adequate model choosing and interpretation and validation of results. One of the criteria that must he considered in investigation method selection is of parasitic phenomena analysing. Such broach of the problem is proposed in this paper regarding to the signal transmission on the interconnecting traces between digital circuits. In the context in which the signals traosmitted frequencies on these connections are increased it is essential to include in their analysis all effects which can modify the system behavior and remain focused
on the two major aspects which can affect the system integrity: Signal Integrity and EMC. The starting point of this analysis consists of the determination of electric parameters of different connection elements: traces, via, pins, etc. The higher frequencies of the signals processed by the digital systems on these connections imply the description of their behavior as systems with distributed constants. The complexity of the analysis is demonstrated through the description of such system
0-7803-8002-9/03/%10.00 02003 IEEE
using Matlab program. The Matlab simulated results and the other electronic circuit simulator results (Pspicc and Omega+) are compared. Using the conclusions, Signal Integrity and EMC are evaluated for various situations.
~ THINKING E OBSTACLES
When students broach the study of connections, they already have skills in electronic circuit analysis that determines sometimes retention in accepting two realities: non-zero value of electric parameters of connections and assimilation of these connections as distributed systems. In addition, in digital system analysis, even if they are used to represent signal as voltage level they neglect the effects of current that flows through the circuits.
2.1. AN connections are electrical elemenis
All the interconnection lines, from electric schemes, are implemented as cables or PCB traces or planes, etc. Each element from schemes has electrical parameters S L, C, and G different from zero. The evaluation of these parameters can be realised using sets of formula 191, or using specialised programs.
26th ISSE 2003
26th International Spring Seminar on Electronics Technology May 8 - 11,2003, Stari Lesni, Slovak Republic
2.2. AII connections are transmission tines
The signal propagation represents a phenomenon, which deploys with finite speed. That means that electrical signals are electrical waves. In order to detect these effects it is necessary to consider systems as distributed constants. The main effects: time delay and signal reflection are permanently present, but cannot be noticed only if the periods of signals are comparable with these phenomena times. This is the consideration why the classical analysis of electrical circuits, where the frequency of processed signals is lower, can make abstraction of these phenomena.
2.3. Current effects in DigitaI Circuit
Fig. 2 Structure of analysed circuit in Matlab
The level of input voltages mainly assures the correct operation of digital circuits. In order to assure these levels, all the specific intervention tools are concentrated in Signal Integrity area. But, in parallel, the aspect regarding the fact that digital circuit operation presumes the current flow through traces cannot be neglected. These current flows through signal traces and than are closed either through ground traces (plane) or through supply traces (plane). These current loops determine the generation of fieldcoupled perturbation. Digital trace currents determine EMC.
Using this method, a connection characterised by parameters L and C measured by unit length can be divided in a number of n circuits with Wn and Cfn parameters. This structure interconnects a generator with programmable output impedance and also with aprogrammable load. An ideal rectangle pulse is applied to the circuit, and a modification of output signal waveform can be shown, with increasing the cells number n. The description of the circuit is based on the following equations system:
di, -_ dt
- E - uc, - R,iL,
3 SOME STEPS IN DIGITAL CONNECTIONS ANALYSIS .
Starting from the known general model for a connection with distributed parameters, shown in figure 1, at a first approximation, a neglect of dissipating elements @dx, Gdx) can be made. Because the substance of phenomena is determined by the echelon time transfer, sequential, of a energy between distributed reactive elements (Ldx, Cdx).
Fig. 1 General model for a distributed system.
The method implies io Matlab solving of this equation system constjtute a helpful intellectual exercise for students (the solving of differential equation system are based on ode45 Matlab function).
3.2. Wave chamcter manifestation o eIectrica1 f signals
3.1 Main phenomena which determine Signal Integriq
In order to create a perception of the way in which the supposed model is more or less close to the reality of presented phenomena, the next Matlab application is proposed.
In fig.3 the modification of signal waveform can be intimate when n=l, n=3 and n=S. Sucb experiment can convince @e shldent that the real situation is most truthfully detected by.the model with a higher value of n. In this way, he is "naturally" guided towards the
26th ISSE 2003
26th International Spring Seminar on Electronics Technology May 8 - 11,2003, Stari Lesni, SlovakRepublic transmission line concept, characterised by characteristic impedance and delay time. the
As can be shown, this application allows any parameter setting in the analysed circuit, in order that the student have the possibility to detect the influence of parameters modification upon the circuit operation.
3.3. Un-matching line eflecis
The modification of the pulse period allows the studcnt to ohserve that the phenomenon is bound with the presence of signal edge (existence of high frequency signals) and not from rectangular pulse frequency. Using the same Matlab application or Pspice simulation with Tline models, the student can change the values of R, and R, resistance, detecting in this way the effect of different reflection coefficients obtained at the extremity of transmission line. Deviations in the signal waveforms from ideal waveform allow understanding the Signal integrity concept. More than that, the student will be familiarised with typical voltage waveforms, based on which they will be able to estimate the reflection coefficients which appear at the extremities of the line. Figure 4 presents the waveforms, which appear at the extremities. of the transmission lines for practical situations. At both edges of the signal can appear combinations of presented situations. The other two combinations of reflection coefficients can be generated in practice only in case of fault circuits (sec figure 5), this fact being demonstrated by the levels of signals.
.. . .
. . ,
Fig. 4 Signal shapes for the real situations
Fig. 3 Matlab simulation results.
26th ISSE 2003
26th International Spring Seminar on Electronics Technology May 8 11,2003, Star6 Lesnl, Slovak Republic
Fig. 6 Model for width trace variation
Fig. 5 Possible shapes only for defect circuits.
3.4. R e d situation compleriry
The model of the transmission line presumes that line parameters are constant. Real situations imply the existence of two types of non-homogeneity:
Topology modification (strip line, microstrip, etc.) and geometric dimensions (trace width) of connections lead to characteristic impedance modification; Presence of some circuit elements that are equivalent either with series inductance (gap in the ground plane, connecton), or with supplementary capacitance in some points (via, 90' bends) . In other situations, these elements can be equivalent with complex LC circuits.
Fig. 7 Voltages in intermediatenodes
The effects of characteristic impedance variation can be easily demonstrated using combinations of Tline models with different parameters. Figure 6 shows the modeling of connection using three segments of Tline, and can also be detected the effect of modification of trace width of second T h e which determine the modification of characteristic impedance and propagation speed on this segment.
Using structure shown in figure 8, can he study the effects of a T topology for a trace which drive two inputs.
Fig. 8 Topology for trace driving hvo inputs
26th ISSE 2003
26th international Spring Seminar on Electronics Technology Lema, Slovak Republic May 8 - 11,2003, Star&
???? ???? .. .. .
Fig. 11 Lp and C, effects for intermediate nodes
Fig. 9 Results of T smchlre analysis.
3.4. Digifal frace currents effects
The effect of connection of a Lp inductance andor Cp capacitance at the structure of a homogeneous line can be equivalent from the input
and output point of view with the modification of characteristic impedance 6. and speed propagation [ ]
This equivalent model dose not characterise the effect of circuit elements Lp and Cp in the points where are connected. This thing is possible using the n i n e model again.
In most situations, the current waves that flow to the digital signal traces are quickly attenuated due to the high impedance which is coupled on the ends of the lines, this high impedance determines the reflection coefficients near to 1. Due to this effect, the closing of the high frequency currents can be considered not realised .on the whole length of trace. This effect leads to decreasing of a magnetic field generated around the trace. Unfortunately, some intervention tool that regards the Signal integrity (for example matching line) leads to distribution of high frequency current on the whole length of trace. This aspect determines the spread of the magnetic field on all length of trace. In this way, the influence bemeen the analysed trace and the near traces can aggravate.
In order to study this phenomenon, a complex simulator can be used (for example Omega+). In figure 12 the modification of the current waveform in a digital trace can be observed, when a matching procedure on the end of the line is applied. This effect is accompanied by the modification of distribution of the magnetic field, as can be see in figure 13.
Fig. 10 Models for Lp and C, effects.
All the above-mentioned effects are automatically included in Signal Integrity analysis if simulators which modeling the structure of printed circuit board are used (for example Omega+). From the point of view of the one who studies elechical connections the above broach is also essential. This makes him model the effect of non-homogeneity using T h e elements.
26th ISSE 2003
26th International Spring Seminar on Electronics Technology May 8 - 11, 2003, Stad Lesni, Slovak Republic From the educational point of view it is important to use models which highlight the essence of phenomena and which avoid collateral manifestations. This type of broach creates the possibility of identifying the cause-effect relationship that appears in complex phenomena. The using of complex models must be encouraged in the view of detecting the ensemble of phenomena. The presentation from this paper can obviously be completed with other elements that are involved in the complexity of a real situation. In this way the effect of non-linearity of digital circuits can be mentioned. Also, in such broach it is important to become aware of the significance of specifications and precision of input data.
Theregression fiom the wmplex model to the simple one must be encouraged any time the results exceed the understanding of cause-effect relationships. This broach avoids acceptance of affected results bom the wrong specifications in complex simulation program settings. The adliteram acceptance of some design prescriptions , ,  must be encouraged only from the perspective of a situation in which any intervention tool is understood through the causeeffect consequence. In the same context, a list with design prescriptions for studied situations can be useful. Probably, a list with such rules cannot cover all the real situations. but we must encouraee the idea that by using the right tool the list can be completed.
Fig. 12 Trace current analysis
Anbambeault 6.: EMyEMC Computational Handbook, Kluwe Academic Puhlishm. 1998. Hall S. H.,Hall G. W., McCall J. A.: High-Sped Digital System Design - A Handbook of Intmonneet Theory and Design Practices, J o b Wiley & Sons, 2004.
Johoson H : High-Speed Digital h i p - A Handbook of . Black Magic, Prentics Hall, 1993.
Fig. 13 Magnetic field distribution analysis.
ar [ 1 Khan M.S.Z., Smith C. E., K f D.: NASWAVE - A 4 Rograms for Display of Standing Waves on Network Analysers, IEEE TraasaCtioos on Education, Vo1.434 No. 2 , May 2001, pp. 151-157 PI M d i g n i a n M.: Conmlling Radiated Emmisioos by Design, Kluwer Academic Publisben, 2001.
[GI Monuoss M.I.: Ronted Circuit B o d design Techniques for EMC Compliance, IEEE press, 1996.
The present broach of this paper demonstrates that in the analysis of complex phenomena it is important to have different investigation tools. For those who get involved in the investigation of Signal integrity and EMC on PCB it is important to be aware that the best method is the one which is most relevant for the studied phenomenon.
[ 1 Mont?? M. 1.: EMC and the Rinted Circuit Board - Design, 7 Theory and Layout Made Simple, IEEE Press, 1999. [ I Trueman C. W.: Intmctive Transmission Line Computer S Program for Undergraduate Teaching, IEEE Tranractions on Education, Vo1.43,No. I , February 2000, pp. 1-14.
WaUrer C.S.:Capacitance, Inductance and Crosstalk Analysis, Ate& House, 1990.
26th ISSE 2003