Transcript

2-1

边界与端口设置

电子科技大学贾宝富博士

HFSS Boundary List

2-2

Perfect E and Perfect H/NaturalIdeal Electrically or Magnetically Conducting Boundaries‘Natural’ denotes Perfect E ‘cancellation’ behavior

Finite ConductivityLossy Electrically Conducting Boundary, with user-provided conductivity and permeability

ImpedanceUsed for simulating ‘thin film resistor’ materials, with user-provided resistance and reactance in Ω/Square

RadiationAn ‘absorbing boundary condition,’ used at the periphery of a project in which radiation is expected such as an antenna structure

SymmetryA boundary which enables modeling of only a sub-section of a structure in which field symmetry behavior is assured.“Perfect E” and “Perfect H” subcategories

Master and Slave‘Linked’ boundary conditions for unit-cell studies of infinitely replicating geometry (e.g. a slow wave circuit & an antenna array)

2-3

HFSS Boundary Descriptions: Perfect E and Perfect H/Natural

Parameters: NonePerfect E is a perfect electrical conductor*

Forces E-field perpendicular to the surfaceRepresent metal surfaces, ground planes, ideal cavity walls, etc.

Perfect H is a perfect magnetic conductorForces H-field perpendicular to surface, E-field tangentialDoes not exist in the real world, but represents useful boundary constraint for modeling

Natural denotes effect of Perfect H applied on top of some other (e.g. Perfect E)boundary

‘Deletes’ the Perfect E condition, permitting but not requiring tangential electrical fields.Opens a ‘hole’ in the Perfect E plane

Perfect E Boundary*

Perfect H Boundary

‘Natural’ Boundary

larperpendicuEr

continuousEr

parallelEr

*NOTE: When you define a solid object as a ‘perf_conductor’ in the Material Setup, a Perfect E boundary condition is applied to its exterior surfaces!!

Perfect H for 2D Aperture (I)

2-4

Monopole Over a Ground plane

Perfect H

Perfect H Surface Interior to the Problem Space Behaves Like an Infinitely Thin 2D Aperture

Perfect H for 2D Aperture (II)

2-5

Small Hole Can be “Cut” in infinitely Thin Septum Between the Upper and Lower Guide Using a Perfect H Surface at the Hole

Perfect H

HFSS Boundary Descriptions: Finite Conductivity

2-6

Parameters: Conductivity and Permeability

Finite Conductivity is a lossy electrical conductor

E-field forced perpendicular, as with Perfect EHowever, surface impedance takes into account resistive and reactive surface losses

User inputs conductivity (in siemens/meter) and relative permeability (unitless)Used for non-ideal conductor analysis*

Finite Conductivity Boundary

gattenuatinlarperpendicuE ,r

*NOTE: When you define a solid object as a non-ideal metal (e.g. copper, aluminum) in the Material Setup module, and it is set to ‘Solve Surface’, a Finite Conductivity boundary is automatically applied to its exterior faces!!

HFSS Boundary Descriptions: Impedance

2-7

Parameters: Resistance and Reactance, ohms/square ( Ω/χ)

Impedance boundary is a direct, user-defined surface impedance

Use to represent thin film resistorsUse to represent reactive loads

Reactance will NOT vary with frequency, so does not represent a lumped ‘capacitor’ or ‘inductor’over a frequency band.

Calculate required impedance from desired lumped value, width, and length

Length (in direction of current flow) ÷Width = number of ‘squares’Impedance per square = Desired Lumped Impedance ÷ number of squares

EXAMPLE: Resistor in Wilkenson Power Divider

Resistor is 3.5 mils long 4 m

(in direction of flow) andils wide. Desired lumped value is 35 ohms.

squareN

RR

N

lumpedsheet /40

875.35

875.045.3

Ω===

==

HFSS Boundary Descriptions: Radiation

2-8

Parameters: NoneA Radiation boundary is an absorbing boundary condition, used to mimic continued propagation beyond the boundary plane

Absorption is achieved via a second-order impedance calculation

Boundary should be constructed correctly for proper absorption

Distance: For strong radiators (e.g. antennas) no closer than λ/4 to any structure. For weak radiators (e.g. a bent circuit trace) no closer than λ/10to any structureOrientation: The radiation boundary absorbs best when incident energy flow is normal to its surfaceShape: The boundary must be concave to all incident fields from within the modeled space

Note boundary does not follow ‘break’ at tail end of horn. Doing so would result in a convex surface to interior radiation.

Boundary is λ/4 away from horn aperture in all directions.

HFSS Boundary Descriptions: Radiation, cont.

2-9

Radiation boundary absorption profile vs. incidence angle is shown at left

Note that absorption falls off significantly as incidence exceeds 40 degrees from normalAny incident energy not absorbed is reflected back into the model, altering the resulting field solution!

Implication: For steered-beam arrays, the standard radiation boundary may be insufficient for proper analysis.Solution: Use a Perfectly Matched Layer (PML) construction instead.

Incorporation of PMLs is covered in the Advanced HFSS training course. Details available upon request.

-100

-80

-60

-40

-20

0

20

Ref

lect

ion

Coe

ffici

ent (

dB)

0 10 20 30 40 50 60

theta (deg)

Reflection Coefficient (dB)

70 80 90

Reflection of Radiation Boundary in dB, vs. Angle of Incidence relative to boundary normal (i.e. for normal incidence, θ = 0)

ETM

θ

HFSS Boundary Descriptions: Symmetry

2-10

Parameters: Type (Perfect E or Perfect H)Symmetry boundaries permit modeling of only a fraction of the entire structure under analysisTwo Symmetry Options:

Perfect E : E-fields are perpendicular to the symmetry surfacePerfect H : E-fields are tangential to the symmetry surface

Symmetry boundaries also have further implications to the Boundary Manager and Fields Post Processing

Existence of a Symmetry Boundary will prompt ‘Port Impedance Multiplier’ verificationExistence of a symmetry boundary allows for near- and far-field calculation of the ‘entire’structure

Conductive edges, 4 sides

This rectangular waveguide contains a symmetric propagating mode, which could be modeled using half the volume vertically....

Perfect E Symmetry (top)

...or horizontally.

Perfect H Symmetry(left side)

HFSS Boundary Descriptions: Symmetry, cont.

2-11

Geometric symmetry does not necessarily imply field symmetry for higher-order modesSymmetry boundaries can act as mode filters

As shown at left, the next higher propagating waveguide mode is not symmetric about the vertical center plane of the waveguideTherefore one symmetry case is valid, while the other is not!

Implication: Use caution when using symmetry to assure that real behavior in the device is not filtered out by your boundary conditions!!

Perfect E Symmetry (top)

Perfect H Symmetry(right side)

TE20 Mode in WR90

Properly represented with Perfect E Symmetry

Mode can not occur properly with Perfect H Symmetry

HFSS Boundary Descriptions: Master/Slave Boundaries

2-12

Parameters: Coordinate system, master/slave pairing, and phasing

Master and Slave boundaries are used to model a unit cell of a repeating structure

Also referred to as linked boundariesMaster and Slave boundaries are always paired: one master to one slaveThe fields on the slave surface are constrained to be identical to those on the master surface, with a phase shift.

Constraints:The master and slave surfaces must be of identical shapes and sizesA coordinate system must be identified on the master and slave boundary to identify point-to-point correspondence

Unit Cell Model of End-Fire Waveguide Array

WG Port(bottom) Ground Plane

Perfectly Matched Layer(top)

Slave BoundaryMaster Boundary

Origin

V-axis

U-axis

HFSS Source List

2-13

PortMost Commonly Used Source. Its use results in S-parameter output from HFSS.Two Subcategories: ‘Standard’ Ports and ‘Gap Source’ PortsApply to Surface(s) of solids or to sheet objects

Incident WaveUsed for RCS or Propagation Studies (e.g. Frequency-Selective Surfaces)Results must be post-processed in Fields Module; no S-parameters can be providedApplies to entire volume of modeled space

Voltage Drop or Current Source‘Ideal’ voltage or current excitationsApply to Surface(s) of solids or to sheet objects

Magnetic BiasInternal H Field Bias for nonreciprocal (ferrite) material problemsApplies to entire solid object representing ferrite material

HFSS Source Descriptions: Port (I)

2-14

HFSS Source Descriptions: Port (II)

2-15

Parameters: Mode Count, Calibration, Impedance, Polarization, Imp. Multiplier

A port is an aperture through which guided electromagnetic field energy is injected into a 3D HFSS model. There are two types:

Standard Ports: The aperture is solved using a 2D eigensolution which locates all requested propagating modes

Characteristic impedance is calculated from the 2D solutionImpedance and Calibration Lines provide further control

Gap Source Ports: Approximated field excitation is placed on the gap source port surface

Characteristic impedance is provided by the user during setup

EXAMPLE STANDARD PORTS

EXAMPLE GAP-SOURCE PORTS

2-16

Impedance, Calibration and Polarization Lines

Impedance line, calibration line and polarization line are optional in port setup.They are located in the port and have a starting point and an end point.

I,C, and/or P Line

Port = cross section of waveguide

2-17

Impedance Line

Without impedance line, HFSS computes port impedance from power and current: Zpi

With impedance line, a voltage can be defined: ∫ E•dl .Two more port impedances result: Zpv and Zvi . These are not the same for non-TEM transmission lines.

2-18

Calibration Line

Calibration lines remove 180o phase ambiguity.This helps to obtain correct phase in S21 and S12 .

waveguide, side view

2-19

Polarization Line

Imposes polarization in case of ambiguity,e.g. in square or circular guides with degenerate modes.

Port = cross section of square waveguide

HFSS Source Descriptions: Incident Wave (I)

2-20

HFSS Source Descriptions: Incident Wave (II)

2-21

Parameters: Poynting Vector, E-field Magnitude and Vector

Used for radar cross section (RCS) scattering problems.Defined by Poynting Vector (direction of propagation) and E-field magnitude and orientation

Poynting and E-field vectors must be orthogonal.Multiple plane waves can be created for the same project.

If no ‘ports’ are present in the model, S-parameter output is not provided

Analysis data obtained by post-processing on the Fields using the Field Calculator, or by generating RCS Patterns

In the above example, a plane incident wave is directed at a solid made from dielectrics, to view the resultant scattering fields.

HFSS Source Descriptions: Voltage Drop and Current Source (I)

2-22

Voltage Drop

Current Drop

2-23

HFSS Source Descriptions: Voltage Drop and Current Source (II)

Example Voltage Drop (between

trace and ground)

Example Current Source (along trace

or across gap)Parameters: Direction and Magnitude

A voltage drop would be used to excite a voltage between two metal structures (e.g. a trace and a ground)A current source would be used to excite a current along a trace, or across a gap (e.g. across a slot antenna)Both are ‘ideal’ source excitations, without impedance definitions

No S-Parameter OutputUser applies condition to a 2D or 3D object created in the geometry

Vector identifying the direction of the voltage drop or the direction of the current flow is also required

Sources/Boundaries and Eigenmode Solutions

2-24

An Eigenmode solution is a direct solution of the resonant modes of a closed structureAs a result, some of the sources and boundaries discussed so far are not available for an Eigenmode project. These are:

All Excitation Sources:PortsVoltage Drop and Current SourcesMagnetic BiasIncident Waves

The only unavailable boundary type is:Radiation Boundary

A Perfectly Matched Layer construction is possible as a replacement

HFSS Source Descriptions: Magnetic Bias

2-25

Parameters: Magnitude and Direction or Externally Provided

The magnetic bias source is used only to provide internal biasing H-field values for models containing nonreciprocal (ferrite) materials.

Bias may be uniform field (enter parameters directly in HFSS)...

Parameters are direction and magnitude of the field

...or bias may be non-uniform (imported from external Magnetostatic solution package)

Ansoft’s 3D EM Field Simulator provides this analysis and output

Apply source to selected 3D solid object (e.g. ferrite puck)

HFSS Ports: A Detailed Look

2-26

The Port Solution provides the excitation for the 3D FEM Analysis. Therefore, knowing how to properly define and create a port is paramount to obtaining an accurate analysis.Incorrect Port Assignments can cause errors due to...

...Excitation of the wrong mode structure

...Bisection by conductive boundary

...Unconsidered additional propagating modes

...Improper Port Impedance

...Improper Propagation Constants

...Differing phase references at multiple ports

...Insufficient spacing for attenuation of modes in cutoff

...Inability to converge scattering behavior because too many modes are requested

Since Port Assignment is so important, the following slides willgo into further detail regarding their creation.

HFSS Port Selection: Standard or Gap Source?

2-27

When would you choose to use a Gap Source Port over a Standard Port?

When the model has tightly-spaced lines

When a port reference location is difficult to determine using a Standard port

When you’d like to use a voltage gap, but want S-parameter output

Gap Source Ports (blue)

2-28

HFSS Ports: Sizing

A port is an aperture through which a guided-wave mode of some kind propagates

For transmission line structures entirely enclosed in metal, port size is merely the waveguide interior carrying the guided fields

Rectangular, Circular, Elliptical, Ridged, Double-Ridged WaveguideCoaxial cable, coaxial waveguide, square-ax, Enclosed microstrip or suspended stripline

For unbalanced or non-enclosed lines, however, field propagation in the air around the structure must also be included

Parallel Wires or StripsStripline, Microstrip, Suspended StriplineSlotline, Coplanar Waveguide, etc.

A Coaxial Port Assignment

A Microstrip Port Assignment (includes air above substrate)

HFSS Ports: Sizing, cont.

2-29

The port solver only understands conductive boundaries on its borders

Electric conductors may be finite or perfect(including Perfect E symmetry)Perfect H symmetry also understoodRadiation boundaries around the periphery of the port do not alter the port edge termination!!

Result: Moving the port edges too close to the circuitry for open waveguide structures (microstrip, stripline, CPW, etc.) will allow coupling from the trace circuitry to the port walls!

This causes an incorrect modal solution, which will suffer an immediate discontinuity as the energy is injected past the port into the model volume

Port too narrow (fields couple to side walls)

Port too Short(fields couple to top wall)

HFSS Ports: Sizing Handbook I

2-30

Microstrip Port Sizing GuidelinesAssume width of microstrip trace is wAssume height of substrate dielectric is h

Port Height GuidelinesBetween 6h and 10h

Tend towards upper limit as dielectric constant drops and more fields exist in air rather than substrateBottom edge of port coplanar with the upper face of ground plane(If real structure is enclosed lower than this guideline, model the real structure!)

Port Width Guidelines10w, for microstrip profiles with w ≥ h5w, or on the order of 3h to 4h, for microstrip profiles with w < h

w

h

6h to 10h

10w, w ≥ hor

5w (3h to 4h), w < h

Note: Port sizing guidelines are notinviolable rules true in all cases. For example, if meeting the height and width requirements outlined result in a rectangular aperture bigger than λ/2 on one dimension, the substrate and trace may be ignored in favor of a waveguide mode. When in doubt, build a simple ports-only model and test.

HFSS Ports: Sizing Handbook II

2-31

Stripline Port Sizing GuidelinesAssume width of stripline trace is wAssume height of substrate dielectric is h

Port Height GuidelinesExtend from upper to lower groundplane, h

Port Width Guidelines8w, for microstrip profiles with w ≥ h5w, or on the order of 3h to 4h, for microstrip profiles with w < h

Boundary Note: Can also make side walls of port Perfect H boundaries

wh

8w, w ≥ hor

5w (3h to 4h), w < h

HFSS Ports: Sizing Handbook III

2-32

Slotline Port GuidelinesAssume slot width is gAssume dielectric height is h

Port Height:Should be at least 4h, or 4g (larger)Remember to include air below the substrate as well as above!

If ground plane is present, port should terminate at ground plane

Port Width:Should contain at least 3g to either side of slot, or 7g total minimumPort boundary must intersect both side ground planes, or they will ‘float’ and become signal conductors relative to outline ‘ground’

g

Approx 7g minimum

h

Larger of 4h or 4g

HFSS Ports: Sizing Handbook IV

2-33

CPW Port GuidelinesAssume slot width is gAssume dielectric height is hAssume center strip width is s

Port Height:Should be at least 4h, or 4g (larger)Remember to include air below the substrate as well as above!

If ground plane is present, port should terminate at ground plane

Port Width:Should contain 3-5g or 3-5s of the side grounds, whichever is larger

Total about 10g or 10sPort outline must intersect side grounds, or they will ‘float’ and become additional signal conductors along with the center strip.

Larger of approx. 10g or 10s

s

h

Larger of 4h or 4g

g

CPW Wave Ports: Starting Recommendations

2-34

Wave Port SizeThe standard recommendation for most CPW wave ports is a rectangular aperture

Port width should be no less than 3 x the overall CPW width, or 3 x (2g + w)Port height should be no less than 4 x the dielectric height, or 4h

Wave Port LocationThe wave port should be centered horizontally on the CPW trace

If the port is on GCPW, the port bottom edge should lie on the substrate bottom ground plane

If the port is on ungrounded CPW, the port height should be roughly centered on the CPW metal layer

Wave Port RestrictionsAs with all wave ports, there must be only one surface normal exposed to the field volume

Port should be on exterior model face, or capped by a perfect conductor block if internal

The wave port outline must contact the side grounds (all CPWs) and bottom ground (GCPW)

The wave port size should not exceed lambda/2 in any dimension, to avoid permitting a rectangular waveguide modal excitation

3 (2g + w)

w

h

4h minimum

g

Grounded CPW

(Port height begins at lower ground)

3 (2g + w)

w

h

4h minimum

g

Ungrounded CPW

(Port height centered on trace)

2-35

HFSS Ports: Sizing Handbook V; Gap Source Ports

Gap Source ports behave differently from Standard Ports

Any port edge not in contact with metal structure or another port assumed to be a Perfect H conductor

Gap Source Port Sizing (microstrip example):“Strip-like”: [RECOMMENDED] No larger than necessary to connect the trace width to the ground“Wave-like”: No larger than 4 times the strip width and 3 times the substrate height

The Perfect H walls allow size to be smaller than a standard port would beHowever, in most cases the strip-like application should be as or more accurate

Further details regarding Gap Source Port sizing available as a separate presentation

Perfect H

Perfect H

Perfect E

Perfect E

Perfect H

Perfect H

Perfect E

Perfect H

2-36

HFSS Port Selection Example: Parallel Traces

Spaced by 8 or more times Trace WidthInputs sufficiently isolated that no coupling behavior should occur

Sufficient room for Wave port apertures around each trace

Use Wave Ports as shown

Spaced by 4 – 8 times Trace WidthInputs still fairly isolated, little to no coupling behavior should occur

Insufficient room for Wave port apertures around each trace without clipping fringing fields

Use Lumped Ports as shown

Spaced by less than 4 times Trace WidthTraces close enough to exhibit coupling

Even and Odd modes possible; N modes total for N conductors and one ground reference [odd mode shown at right]

Lumped Ports from trace to ground neglect coupling behavior and are no longer appropriate

Use multi-mode Wave Port

Terminal line assignments can permit extraction of S-parameters referenced to each ‘trace’

HFSS Ports: Spacing from Discontinuities

2-37

Structure interior to the modeled volume may create and reflect non-propagating modes

These modes attenuate rapidly as they travel along the transmission line

If the port is spaced too close to a discontinuity causing this effect, the improper solution will be obtained

A port is a ‘matched load’ as seen from the model, but only for the modes it has been designed to handleTherefore, unsolved modes incident upon it are reflected back into the model, altering the field solution

Remedy: Space your port far enough from discontinuities to prevent non-propagating mode incidence

Spacing should be on order of port size, notwavelength dependent

PortExtension

HFSS Ports: Single-Direction Propagation

2-38

Standard ports must be defined so that only one face can radiate energy into the model

Gap Source Ports have no such restriction

Position Standard Ports on the exterior of the geometry (one face on background) or provide a port cap.

Cap should be the same dimensions as the port aperture, be a 3D solid object, and be defined as a perfect conductor in the Material Setup module

Port on Exterior Face of Model

Port Inside Modeled Air Volume; Back side covered with Solid Cap

HFSS Ports: Mode Count

2-39

Ports should solve for all propagating modesIgnoring a mode which does propagate will result in incorrect S-parameters, by neglecting mode-to-mode conversion which could occur at discontinuities

However, requesting too many modes in the full solution also negatively impacts analysis

Modes in cutoff are more difficult to calculate; S-parameters for interactions between propagating and non-propagating modes may not converge well

What if I don’t know how many modes exist?Build a simple model of a transmission line only, or run your model in “Ports Only” mode, and check!You can alter the mode count before running the full solution.

Degenerate mode ordering is controlled with calibration lines (see next slide)

Circular waveguide, showing two orthogonal TE11 modes and TM01 mode (radial with Z-component). Neglecting the TM01 mode from your solution would cause incorrect results.

HFSS Ports: Degenerate Modes

2-40

Degenerate modes have identical impedance, propagation constantsPort solver will arbitrarily pick one of them to be ‘mode(n)’ and the other to be ‘mode(n+1)’

Thus, mode-to-mode S-parameters may be referenced incorrectly

To enforce numbering, use a calibration lineand polarize the first mode to the lineOR, introduce a dielectric change to slightly perturb the mode solution and separate the degenerate modes

Example: A dielectric bar only slightly higher in permittivity than the surrounding medium will concentrate the E-fields between parallel wires, forcing the differential mode to be dominantIf dielectric change is very small (approx. 0.001 or less), impedance impact of perturbation is negligible

For parallel lines, a virtual objectbetween them aids mode ordering. Note virtual object need not extend entire length of line to help at port.

In circular or square waveguide, use the calibration line to force (polarize) the mode numbering of the two degenerate TE11 modes. This is also useful because without a polarization orientation, the two modes may be rotated to an arbitrary angle inside circular WG.

HFSS Ports: Impedance Definitions

2-41

HFSS provides port characteristic impedances calculated using the power-current definition (Zpi)

Incident power is known excitation quantityPort solver integrates H-field around port boundary to calculate current flow

For many transmission line types, the power-voltage or voltage-current definition is preferred

Slot line, CPW: Zpv preferredTEM lines: Zvi preferred

HFSS can provide these characteristic impedance values, as long as an impedance line is identified

The impedance line defines the line along which the E-field is integrated to obtain a voltageOften it can be identical to the calibration line

For a Coax, the impedance line extends radially from the center to outer conductor (or vice versa). Integrating the E-field along the radius of the coaxial dielectric provides the voltage difference.

In many instances, the impedance and calibration lines are the same!

HFSS Ports: Phase Calibration

2-42

A second purpose of the calibration line is to control the port phase references

The 2D port eigensolver finds propagating modes on each port independentlyThe zero degree phase reference is chosen at a point of maximum E-field intensity on the port face.

This occurs twice, with 180 degrees separation, for each 360 degree cycle

Therefore the possibility exists for the software to select inconsistent phase references from port to port, resulting in S-parameter errors

All port-to-port S-parameter phases, e.g. S21, will be off by 180 degrees

Solution: The calibration line defines the preferred direction for the zero degree reference on each port.

Which of the above field orientations is the zero

degree phase reference? Calibration Line defines...

HFSS Ports: Impedance Multiplier

2-43

When symmetry is used in a model, the automatic Zpi and impedance line-dependant Zpv and Zvi calculations will be incorrect, since the entire port aperture is not represented.

Split the model with a Perfect Esymmetry case, and the impedance is halved.Split the model with a Perfect Hsymmetry case, and the impedance is doubled.

The port impedance multiplier is just a renormalizing factor, used to obtain the correct impedance results regardless of the symmetry case used.The impedance multiplier is applied to all ports, and is set during the assignment of any port in the model.

Whole Rectangular WG(No Symmetry)

Impedance Mult = 1.0

Half Rectangular WG(Perfect E Symmetry)Impedance Mult = 2.0

Half Rectangular WG(Perfect H Symmetry)Impedance Mult = 0.5

...and for Quarter Rectangular WG?(Both Perfect E and H Symmetry)

Impedance Mult. = 1.0

Top Related