Development of EMTools

1

• Development of EM-Tools

Stefan Vossen, Peter Zwamborn and Rene van
Wijk TNO Defence, Security and Safety Tel. 070
3740349 Email stefan.vossen_at_tno.nl
2
Contents

• Available tools.
• What is the use of these tools?
• New approaches.
• Example of EM topology and statistical EM.
• Conclusions.

3
Numerical approaches

• Solving an electromagnetic interaction problem
  numerically we use
• a local technique (FDTD, TLM, FE),
• a global technique (Integral Equation (MoM)),
• (semi) hybrids.
• Solve in Time or Frequency domain.
• Methods
• Matrix inversion,
• Iterative,
• Direct space-time discretization,
• Etc

4
What did we (try to) solve?

• Enclosures.
• Printed circuit boards.
• Lumped elements.
• Cross talk
• Cabling,
• Printed circuit paths,
• Effects from cable/antenna entries.
• (Very) high frequency problems.
• Digital signals still vary with voltages and not
  just 0 and 1.
• Etc
• Did we reach a solution?

5
Not exactly Common problems

• Problems
• Speed,
• Memory,
• Most of all Obtaining EMC is not an exact
  science.
• Think of first electronics lesson versus later
  courses in school
• From network theory to a practical approach.
• Modern day EMC needs to much input to find a
  direct network theory solution for a complete
  system.

6
Where does this leave computer techniques?

• Numerical tools should be used WHERE and WHEN
  appropriate.
• The computer can never provide a complete answer,
  it is an aid to reach the answer.
• (Experimental) verification stays important.
• Try taking another approach to a problem.

7
Designer
Design (change)
Ready
Specs
Input
Evaluation
Knowledge Based System
Design rules
Database
EM-Simulation tools
EMC/EMI
measurements
FDTD
Transmission lines
Cable structures
Coupling to cables
MoM
equations
Rules of thumb
Antenna structures
Integral
Material properties
PCBs
Iterative / MOiA
Antennas
FEM
(Semi) Hybrid tools
Construction
EM Topology
Statistical EM
8
EM Topology

• Decompose the geometry into volumes.
• Make an inventory of penetration paths in each
  volume.
• Each volume must be separated by geometrical or
  physical barriers.
• Example Network switch
• Employ appropriate tools to subproblems.
• Combine results through a network formulation.

Box

• Inside box
• Printed circuit boards
• Cabling
• (Swithed) Power supply
• Chips
• Etc

Cables, cable entries and ventilation holes
9
StatEM Introduction

• Statistical electromagnetics enables to treat the
  problem of interior responses of complex systems.
• The principle quantities of interest are cable
  and pin-currents not field coupling or
  scattering.
• Given an electromagnetic environment and a
  system, the probability of the systems
  performance can be examined.
• Necessary requirement An analytical technique to
  predict the statistical distribution of
  electromagnetic fields.

10
Why statistical electromagnetics?

• Simulation wire altered around the centre-line.
• A sample of transfer functions for different wire
  positions inside the structure.
• The variation between different wire positions
  reaches a magnitude range over 10dB (bold line
  corresponds to the centre position.)
• Transfer functions are extremely complex and are
  very sensitive to variations.

11
The statistical nature of a measurement I

• Imagine a cavity in which a single mode is
  excited and a sensor is located at a fixed but
  arbitrary location.
• Since the sensor is at an arbitrary location
  relative to the peaks and nodes of the mode, it
  may measure a field strength with any value
  between the true mode amplitude and zero.
• If one changes the frequency slightly and excites
  a different mode, the spatial pattern will jump
  to a new configuration, and the sensor will be at
  a different position relative to the peaks and
  nodes.

12
The statistical nature of a measurement II

• Thus, far from measuring the mode amplitude, in
  any few measurements the sensor only measures
  samples greater then zero and less than the true
  mode amplitude.
• The true mode amplitude remains undetermined, as
  does the energy density in the cavity and the
  maximum field that a sensor or component might
  experience at some other location in the cavity.
• In reality this picture is further complicated by
  the fact that in most real cavities many modes
  are excited simultaneously. The field at a
  particular location may be the sum of
  contributions of hundreds of modes.

13
Exact treatment of the problem?

• Coupling to the sensor depends on polarization
  which is undetermined for many modes.
• In irregular cavities with nonorthogonal walls
  the field polarization for a given mode can vary
  with spatial location.
• To calculate the fields in a cavity of a
  particular shape, at a particular location and
  with the many other required parameters
  specified, is both complex and difficult to
  interpret.
• Any change in any of the parameters requires a
  new complex solution and leaves no general
  understanding of the behavior of the system.

14
So we do statistics!

• A useful solution must be statistical in nature
  and depend only on general properties of the
  system.
• The solution cannot depend in detail on such
  things as whether a small metallic can has been
  set down somewhere inside the test article, or
  the position of the pilot's arms, or whether some
  mechanical widget has moved from position A to
  position B, changing the mode structure.
• If the answer did depend on those things, all of
  the measurements would be useless, defeated by
  details present in all systems.

15
What do we want?

• For a given disturbed electromagnetic
  environment, what is the probability that a
  particular interior wire, pin or integrated
  circuit will not carry a current greater than
  some acceptable value?
• Prior to developing the protection requirements,
  there is a minimum set of information that must
  be developed first
• detailed descriptions of environments the system
  must survive or operate through,
• detailed descriptions of the immunity levels for
  the electronics comprising the system,
• the margin selected to compensate for risks and
  uncertainties,
• the general system layout from an electromagnetic
  protection point of view,
• detailed information on system capabilities, on
  performance requirements, and on allowable
  impacts on capabilities e.g., upsets, down time,
  recovery time.

16
Example computer in a building

• Tx represents an electromagnetic time-harmonic
  source which causes interference on a device Rx.

17
Example first step EM topology

• Identify the various electromagnetic zones in
  order to develop a proper coupling model.
• Each zone must be separated by geometrical or
  physical barriers.
• For simplicity, we assume only physical barriers
  which are perfectly conducting and have plane
  surfaces (i.e. no scattering only reflections
  occur at the barriers.)
• The solid lines represent the barriers.

18
Example second step calculation input power

• Determine the amount of electromagnetic power
  which is coupled into Rx.
• As Tx is a time-harmonic source, the steady-state
  time-average rate of doing work on the sources in
  Rx by the fields is equal to the average flow of
  power into the volume Rx through the boundary
  surface of Rx.
• Thus, the input power can be calculated using
  Poyntings theorem for time-harmonic variation of
  the fields.

19
Comparison practice and theory

• Objective
• To describe an analytic technique to predict the
  statistical distribution of field amplitudes in
  complex cavities, which result from the
  simultaneous excitation of hundreds of modes.
• To test the validity of the theory, comparison
  with experimental data is mandatory.
• To compare we use the Kolmogorov-Smirnov (K-S)
  test and probability plot.

20
Comparison experiment

• Measure field inside Rx and determine the
  transfer function.

21
Comparison theory

• Determine the statistical properties of the
  measured power which include the form of the
  distribution, and the parameters which determine
  its shape.
• Electromagnetic power is positive definite and of
  quadratic form.
• The asymptotic distribution of the quadratic form
  can be approximated by a gamma distribution.

input power Rx
number of excited modes
cavity Q
measured power inside Rx
distribution parameters
22
Comparison concluded
predicted
Measured
23
Example step three interior response

• How do the predicted fields translate into
  induced currents on printed circuit boards?
• Assume the printed circuit board to be a
  two-terminal, passive electromagnetic system.
• The complex Poynting theorem can be used to
  define the input impedance and corresponding
  input current and voltage.
• Calculate current and voltage levels at
  integrated circuitry and establish functional
  distortion and destruction thresholds.

24
Remarks non-linearities

• Due to semiconductor non-linearities, high
  frequency components of the irradiated field can
  also be converted to a low frequency component.
• This low frequency component then travels through
  the system like an in-band signal, which in turn
  can disrupt further signal processing features.

25
Conclusions

• Conventional numerical tools/methods are usefull
  for modelling simplified problems.
• The entire EMC design problem is very difficult
  or even impossible to model with conventional
  tools/methods.
• New approaches use i.e. EM topology and
  statistical EM.
• These methods take hybrid tooling one step
  further.
• Human interfacing remains an important factor in
  finding EMC solutions
• Question???

26
Remarks chaos I

• When the induced currents and voltages have a
  certain frequency and amplitude the systems
  behavior even can become chaotic.

27
Remarks chaos II
Period doubling
Period one
More period doubling
Chaos