当前位置:文档之家› 2014-12_MWJ_EMC-Design-Flow-Modern-Electronics (1)

2014-12_MWJ_EMC-Design-Flow-Modern-Electronics (1)

EMC Simulation

in the Design Flow of Modern Electronics

Andreas Barchanski

CST AG, Germany

Jens Kr?mer, Pietro Luzzi

Festo AG1, Germany

E M C compliance is a necessary condi-

tion for releasing products to market.

National and international standards bodies such as the IEC2 and the FCC3 de?ne limits on the emissions a device is allowed to produce, and automotive and aerospace manu-facturers can set even stricter standards for their OEM suppliers. If these standards are not met, the product cannot be sold.

EMC engineering has traditionally been the domain of measurement alone, with the result that EMC compliance was only considered late in the design process, since it required a work-ing prototype. Correcting EM C problems at this late stage can be cost-intensive, especially if multiple prototype iterations are required.

In addition, although measurement can help identify the symptoms of EMC issues, it is only of limited use for identifying the root causes of the problems. This means EMC troubleshoot-ing using measurement is often limited to ap-plying countermeasures such as ?lters and ad-ditional shielding, which can increase the size and the cost of the product, instead of tackling the problem at its source.

Simulation allows EM C compliance to be analyzed earlier in the design process, and in greater depth than is possible with measure-ment alone. Because simulation doesn’t re-quire a physical prototype, it can be used to investigate many different con?gurations and

Fig. 1 Manufactured demonstrator board (top) and

the equivalent simulation model (bottom).

Reprinted with permission of MICROWAVE JOURNAL? from the December 2014 issue.

?2015 Horizon House Publications, Inc.

domain solver needs to simulate each

port separately, but it can simulate a

broad band of frequencies at once,

and is more ef?cient than the fre-

quency domain technique for electri-

cally large simulations.

Regardless of which simulation

method is used, the results are the

same: the S-parameters of the model

and the ?eld distribution inside it.

These S-parameters can be used to

produce a circuit-level model of the

component, which can be incorpo-

rated into the schematic of the device.

Circuit solvers are much faster than

3D solvers and, for a simple circuit, a

spectrum can be calculated within

seconds. This means that the circuit

elements can be adjusted interactive-

ly, and optimization of the circuit can

be carried out in a reasonable length

of time. This is especially useful for

extracting equivalent circuits for ele-

ments and for helping simulation and

measurement agree more closely.

In order to match the simulation

and measurement, the parasitic val-

ues of the mounted elements must be

known. To allow the input ?lter to be

simulated accurately, an equivalent cir-

cuit is added to the simulation model

representing the parasitics inherent in

the test ?xtures. By using the ‘Tune’

function, the values of the parasitics

are varied until the simulation’s behav-

ior matches the measurement. These

values are then applied across multiple

simulations, while maintaining good

agreement between measurement and

results (see Figure 2).

CONDUCTED EMISSIONS

Two parts of the system are in-

volved in conducted emissions: the

input ?lter and the voltage regulator.

For the analysis presented here, a

constant load of 2.5 ? is placed on the

board. This load leads to a high cur-

rent, so the emission is also quite high.

This is, therefore, a worst case set-up.

The conducted emission will be

measured using ports at the connector

that are connected on the other side

to an electric boundary, acting as a vir-

tual reference. This electric boundary

creates a low-impedance path for the

return current. It is not an ideal exter-

nal ground, as this cannot exist in a 3D

model.

As already discussed, the frequen-

cy domain solver is a good choice for

most conducted emissions simulations

C performance just

C

the interactions between signal lines

and components. Only a 3D simula-

tion can reveal these effects.

This means that circuit-level simula-

tion methods coupled to full-wave 3D

EM simulation can complement each

other. These can be combined in a

number of different ways: for example,

a ‘Combine Results’ task, based on the

superposition principle, can be used to

calculate the ?eld distribution when a

circuit is attached to a 3D model with-

out having to resimulate the entire 3D

model, while true transient/EM co-

simulation can be used to integrate cir-

cuits, including nonlinear components,

into a full-wave 3D simulation.

The ?rst step is to de?ne the mod-

el geometry and import CAD and

EDA data for PCBs and enclosures.

Lumped elements, ports, probes and

?eld monitors are also de?ned at this

stage. The ports are what will later

provide the link between the full-wave

simulation and the circuit simulation.

Once these are set up, the simulation

can begin.

Full wave solvers can be broadly di-

vided into time domain and frequen-

cy domain methods. For conducted

emissions, where the frequency range

is narrow and the ?elds are con?ned

to the electrically small board, the fre-

quency domain solver is usually the

best choice. A frequency domain di-

rect solver only requires one pass to

simulate all the ports, which is useful

for densely populated boards.

For radiated emissions, a time do-

main method is a better ?t. The time

components within the device – are

usually dif?cult to change later on and

can be crucial for the EM C compli-

ance of a device.

Simulation is also helpful at the

troubleshooting stage, where it is

a complementary tool to measure-

ment. Current and ?eld visualization

can serve to highlight the locations of

EMC problems. They offer more in-

sight into the behavior of the device

than a blip on a spectrum analyzer

graph can provide, although the mea-

surement is still critical for the ?nal

decision as to whether the device

passes or fails the test.

This article considers how to test

and develop the EM C work?ow for

modern electronic devices. The dem-

onstrator board (see Figure 1) is

based on modern electronic designs,

and the principles outlined here are

applicable to a wide range of devices.

On this board, the power is routed

through an input ?lter to a voltage

regulator module (VRM) that drives a

power delivery network (PDN) on the

PCB. A number of single ended and

differential drivers are placed on the

board. Some of them are routed over

a connector and all of them are ?nally

terminated on the board.

The demonstrator board contains

two versions of the same system on a

common substrate – one designed to

offer good EMC performance (the up-

per half of the board in Figure 1, called

the “good” layout), and one designed

to offer poorer performance (the lower

half, called the “bad” layout).

30

10.15

10100806040200–20–40Frequency (MHz)3010.15

10Frequency (MHz)

Simulated Measured

100806040200–20–40

PCB and outputs them as a near-?eld

diated emissions (in this case, 30 MHz to 1 GHz). In this example, the emis-

–40–60–80n i t u d e i n d B )

s Fig. 6 Electric ?eld around the PCB at

100 (top), 350 (center) and 700 MHz (bottom).

source ?le. This near-?eld source is then imported into a second simulation of the anechoic chamber, using a much coarser mesh. According to Huygens’ Principle, the second simulation will create the same radiated emissions as the full 3D model, and it only requires a few minutes. The same approach can be used for modeling other structures which can affect emissions, such as the housing of a PCB.

CONCLUSION

Simulation is a useful tool for the EM C analysis of complex electronics and allows engineers to identify possi-ble EMC problems early in the design process. Simulation can identify trends which are subsequently con?rmed by the measurements. Broadband simulations, in particular, are fast and straightforward and can provide useful data even in cases where the complex-ity of the measurements makes a direct comparison between simulation and measurement dif?cult. ■References

1. Festo, https://www.doczj.com/doc/1a5909225.html,

2. IEC/TR EN 61000, Electromag-

netic Compatibility (EMC)

3. Code of Federal Regulations, Ti-

tle 47, Part 15 (47 CFR 15), Fed-

eral Communications Commission

(FCC)

4. CST STUDIO SUITE, https://

https://www.doczj.com/doc/1a5909225.html,

5. LT8610: 42 V, 2.5 A Synchronous

Step-Down Regulator with 2.5 μA

Quiescent Current, http://www.

https://www.doczj.com/doc/1a5909225.html,/product/LT8610

6. CISPR 16, Speci?cation for radio

disturbance and immunity mea-

suring apparatus and methods

Andreas Barchanski is the EMC

market development manager at

CST. He holds an M.Sc. degree in

physics and a Ph.D. in numerical

electromagnetics from the Technical

University Darmstadt. He joined CST

as an application engineer in 2007.

Besides EMC, his main interest lies

in the simulation of various electronic

systems ranging from high speed

digital to power electronics.

Jens Kr?mer is the EMC simulation

and testing manager at Festo, a

leading world-wide supplier of

automation technology and a leader

in industrial training and education

programs. He holds a diploma in

electronics from the University

Stuttgart. His main interest lies in

energy propagation on PCBs and

through connectors – both intentional

and unintentional.

Pietro Luzzi is an EMC consultant

at Festo. He holds a bachelor’s degree

in electrical engineering. In 2012 he

joined the EMC Simulation & Testing

department, where he plans circuit

boards and simulates a variety of

interactions between EMC and the

PCB.

相关主题
文本预览
相关文档 最新文档