Upload
mohammad-hossein
View
305
Download
16
Embed Size (px)
Citation preview
8/18/2019 EMC / EMI in HFSS v8
1/42
1
EMC / EMI
in
HFSS v8
Jim ShermanAnsoft Applications Engineer
East Coast
8/18/2019 EMC / EMI in HFSS v8
2/42
2
Summaryw What is EMC / EMI ?
w Practical Examples:
w Trace over split in ground plane(HFSS EMI Wizard)
w Heatsink Emissions(HFSS Eigenmode Analysis)
w EMI from shielding enclosures(HFSS user Exercise)
• Will be available as download or disk:
• Listing of EMI Wizard macro• Complete EMI exercise
8/18/2019 EMC / EMI in HFSS v8
3/42
3
EMC/EMI ?
w EMC - ElectroMagnetic Compatibilityw Ability of equipment to function without error in it’s
intended EM environment
w EMI - ElectroMagnetic Interferencew
EM emissions from the equipment that interfere withnormal operation of other equipment
EMC Margin
Frequency E M
D i s t u r b a n c e
L e v e
l
Emission Limit
Immunity Limit
Equipment
8/18/2019 EMC / EMI in HFSS v8
4/42
4
Three Elements of the EMC Problem
EM Sourc e EM Recepto r Path
Conducted(Electric Current)
Inductively Coupled(Magnetic Field)
Radiated(Electromagnetic Field)
Capacitively Coupled(Electric Field)
Antenna
Connector
Apertures
Lightning
Power Line
Electronics
Cell Phone
Grounding
Electronics
Antenna
Apertures
Cell Phone
PeopleGrounding
Transistor
Diode
8/18/2019 EMC / EMI in HFSS v8
5/42
5
Solution
w HFSS – High Frequency Structure Simulator w Provide fast, accurate EMC/EMI predictionsw Use it early in product developmentw Understand the EM interaction
w EMI Lab bench measurementsw Uses specialized test equipmentw Test done late in product development
w EM interaction hard to understand
TestEquipment
Data
HFSSSoftware
EM Model
Data
8/18/2019 EMC / EMI in HFSS v8
6/42
6
Practical Examples **
w ACES Standard Problem 2000-2w Trace over split in ground plane
(HFSS EMI Wizard)
w ACES Standard Problem 2000-4w Heatsink Emissions
(HFSS Eigenmode Analysis)
w ACES Model Validation Paper w EMI from shielding enclosures
(HFSS user Exercise)
** Reference:(ACES) Applied Computation Electromagnetics Society:http://aces.ee.olemiss.edu/
8/18/2019 EMC / EMI in HFSS v8
7/42
7
Example 1:**ACES Standard Problem 2000-2
Trace Over Split in Ground Plane
** Linda Walling - Ansoft AE 1999
8/18/2019 EMC / EMI in HFSS v8
8/42
8
Problem Description
w EMI
Dimensions:
plane size: 10” x 12”
trace: 5 mil wide; 10” long
(trace is 5mil above plane) 80 Ohm
substrate: FR4 er = 4.5
slot: 8” long; 20 mil wide.
Dimensions:
plane size: 10” x 12”
trace: 5 mil wide; 10” long
(trace is 5mil above plane) 80 Ohm
substrate: FR4 er = 4.5
slot: 8” long; 20 mil wide.
Stitching capacitor
0.1uF/470pF
with 2 nH/4.88 nH indu ctance
with 0.5 Ohm seriesres i s tance
Stitching capacitor
0.1uF/470pF
with 2 nH/4.88 nH ind uctance
with 0.5 Ohm series
res i s tance Source/Load3.3mV
100 MHz to 2 GHz
50 Ohm load
Source/Load
3.3mV
100 MHz to 2 GHz
50 Ohm load
Find:
Maximu m EMI 3m f rom c i rcu i t
acros s 100 MHz to 2 GHz Band
Find:
Maximum EMI 3m f rom c i rcu i t
acros s 100 MHz to 2 GHz Ban d
ACES Problem 2000-2
8/18/2019 EMC / EMI in HFSS v8
9/42
9
Model Reduction Tricks
w Increase substrate thickness and strip width one order of magnitude.w This helps to relax aspect ratio for smaller mesh.w Line impedance remains the same.w Emissions are not affected.
w Use a virtual objectw Use mesh seeding to reduce number of adaptive passes.w Put virtual object in the air above the substrate.
w Simplify metal layersw Use Perfect Electrical Conductors.w Make all conductors 2D objects.
ACES Problem 2000-2
8/18/2019 EMC / EMI in HFSS v8
10/42
10
HFSS Model
Trace Slot
FR4 andground p lane
Virtual Object m aterial: air
surfaces us ed in EMI calculat ion
A i r Box material: air
wi th rad iat ion s ur faces
50 Ohmgap source
50 Ohmgap sourc e load
ACES Problem 2000-2
8/18/2019 EMC / EMI in HFSS v8
11/42
11
Solution Time Reduction Tricksw
Save time solving multiple geometriesw 1.Trace centeredw 2. 0.1 pF capacitor is placed across the slot close to tracew 3. 0.1 uF capacitor is placed across the slot close to tracew 4. Slot removedw 5. Trace moved to 2” from edge of board with slot and no
capacitors
w Reuse the meshw Capacitor 3D object
w First assigned to a vacuum dielectric.w Perform a solvew Solved project is then copiedw Mesh remains the samew Materials can be changedw Boundaries can be changedw Only the fast sweep is required again
w Use EMI Wizard
ACES Problem 2000-2
8/18/2019 EMC / EMI in HFSS v8
12/42
12
EMI Wizardw EMI Wizard:
w Field Post Processing Macrow Full complex vector field solution availablew Data spans the entire fast frequency sweep
w Allowed:w PML’sw Symmetry Walls
w Use inputs for EMI sweep customizationw Adjust frequency step to coarse or fine
w Provides results in dBuV/mw Finds maximum Etotal, Ephi, and Ethetaw Writes data to ASCII filew Requires a user supplies face list
ACES Problem 2000-2
8/18/2019 EMC / EMI in HFSS v8
13/42
13
EMI Wizardw The macro: emiwiz.mac **
** will be available as a download or disk
ACES Problem 2000-2
8/18/2019 EMC / EMI in HFSS v8
14/42
14
EMI Wizardw User Inputs
ACES Problem 2000-2
8/18/2019 EMC / EMI in HFSS v8
15/42
8/18/2019 EMC / EMI in HFSS v8
16/42
16
EMI Wizardw Specifiy sweep range (must be within fast sweep range)
ACES Problem 2000-2
8/18/2019 EMC / EMI in HFSS v8
17/42
17
EMI Wizardw Specifiy output file and path
ACES Problem 2000-2
8/18/2019 EMC / EMI in HFSS v8
18/42
18
HFSS Predicted Resultsw E total vs Frequency from Emi Wizard output
Best Case: No SlotBest Case: No Slot
3.3mV source3.3mV source
Capacitoris veryfrequencydependent
Capacitoris veryfrequencydependent
ACES Problem 2000-2
8/18/2019 EMC / EMI in HFSS v8
19/42
19
w Different cases produce modes that radiate in different directions.
Frequency (GHz)
Results forCases 1 and 2
Results for
Cases 1 and 2
Angle of
thetaor phi(deg.)
Maximum E Total
HFSS Predicted ResultsACES Problem 2000-2
8/18/2019 EMC / EMI in HFSS v8
20/42
20
Conclusions
w HFSS can be used to simulate EMI test structures.
w Only real problem is related to aspect ratio;w i. e. 5 mil substrates with 10 inch boards.
w The circuit model can be modified to improve aspectratio.
w Scale substrate and trace width.w Has very little effect on EMI results.w Don’t change the slot width. It will change EMI
results.
ACES Problem 2000-2
8/18/2019 EMC / EMI in HFSS v8
21/42
21
Example 2:**
ACES Standard Problem 2000-4Heatsink Emissions
** Richard Remski - Ansoft AE 2000
8/18/2019 EMC / EMI in HFSS v8
22/42
22
Problem Description
Heat Sink: 2.5 x 3.5 x 1.5 inch block,located 6 mm above a 6.29 x 4.74 inchground plane (2D conducting sheet)
Ground configurations: 6 mm squaregrounds, located in various combinationsof one, two, four, and eight locations at atime.
Model enclosed in volume of ‘vacuum’,with Perfect_H boundary walls.
First 4 eigensolutions obtained starting at0.1 GHZ.
ACES Problem 2000-4
Test cases
Grounds
1 Only
1, 2
3, 4
1, 2, 3, 4
5, 6, 7, 8
Test cases
Grounds
1 Only1, 2
3, 4
1, 2, 3, 4
5, 6, 7, 8
1
2
6
3
4
5
7
8
Ground plane edgeGround Pins
Air Circuit
8/18/2019 EMC / EMI in HFSS v8
23/42
23
HFSS Eigenmode Solution ?
ACES Problem 2000-4
w Eigenmode looks for natural modes in structure.
w It can be used to simulate resonance effects in EMItest structures.
w Sources are not allowed
w HFSS output:w Resonant frequenciesw Full complex vector field solution
8/18/2019 EMC / EMI in HFSS v8
24/42
8/18/2019 EMC / EMI in HFSS v8
25/42
25
E-field magnitude on the ground plane for the single-ground configuration. Note that a probe too close tocenterline might not excite either Modes 3 or 4, but thisis not of too much concern since the fundamentalmode would specify the worst-case emissions threat.
The first three modes carry most of the E-field beneaththe heat sink, while Mode 4 carries most energy on theground plane edges around the sink. Judging from theMode 4 distribution, mid-side grounds would not likelyterminate this mode.
E-field magnitude on the ground plane for the single-ground configuration. Note that a probe too close tocenterline might not excite either Modes 3 or 4, but thisis not of too much concern since the fundamentalmode would specify the worst-case emissions threat.
The first three modes carry most of the E-field beneaththe heat sink, while Mode 4 carries most energy on theground plane edges around the sink. Judging from theMode 4 distribution, mid-side grounds would not likelyterminate this mode.
F1=603 MHz F2=1.34 GHz F3=1.41 GHz
F4=1.48 GHz
ACES Problem 2000-4
HFSS Predicted E field:Eigensolution with 1 ground only
8/18/2019 EMC / EMI in HFSS v8
26/42
26
E-field magnitude on the ground plane with twogrounds along the short sides. Peak forfundamental mode is beneath heat sink, likelypermitting good coupling in a probe-excitedanalysis. Modes 2 and 3 however have nulls alongcenterline of heat sink and could get missed (again,not a concern if the fundamental mode is found).
As expected, Mode 4 appears identical to that for thesingle ground case, and carries more energy on theground plane edges than beneath the heat sink.
E-field magnitude on the ground plane with twogrounds along the short sides. Peak forfundamental mode is beneath heat sink, likelypermitting good coupling in a probe-excitedanalysis. Modes 2 and 3 however have nulls alongcenterline of heat sink and could get missed (again,
not a concern if the fundamental mode is found).As expected, Mode 4 appears identical to that for thesingle ground case, and carries more energy on theground plane edges than beneath the heat sink.
F1=1.04 GHz F2=1.37GHz F3=1.40 GHz
F4=1.48 GHz
ACES Problem 2000-4
HFSS Predicted E field:Eigensolution with grounds on 1 and 2
8/18/2019 EMC / EMI in HFSS v8
27/42
27
E-field magnitude on the ground plane with two groundsalong the long side. Here the fundamental mode has arelative null at the heatsink center; therefore none of themodes might be excited by a probe feed too close to thecenter. The reference paper‘s feed location likely did notexcite Mode 1, but did excite Mode 2. This would explainwhy this configuration was reported to have an“advantage” better than that of the other 2-ground case,while HFSS shows the first resonance mode frequency isactually lower.
Again, Mode 4 is the same as for the single-ground andthe other two-ground case, as the ground locations donot prevent it from forming.
E-field magnitude on the ground plane with two groundsalong the long side. Here the fundamental mode has arelative null at the heatsink center; therefore none of themodes might be excited by a probe feed too close to thecenter. The reference paper‘s feed location likely did notexcite Mode 1, but did excite Mode 2. This would explainwhy this configuration was reported to have an“advantage” better than that of the other 2-ground case,while HFSS shows the first resonance mode frequency isactually lower.
Again, Mode 4 is the same as for the single-ground andthe other two-ground case, as the ground locations donot prevent it from forming.
F1=994 MHz F2=1.18 GHz F3=1.46 GHz
F4=1.48 GHz
ACES Problem 2000-4
HFSS Predicted E field:Eigensolution with grounds on 3 and 4
8/18/2019 EMC / EMI in HFSS v8
28/42
28
E-field magnitude on the ground for all fourside grounds. Fundamental mode hasrelative null beneath ground plane centeragain, and may have been missed in a probe-excited analysis. Modes 2 and 3 similarlyhave large nulls beneath one or the other
axis of the heat sink, making it easy to seehow a radiation model with a fixed probelocation as specified in the refererence papermight not couple well to them, either.
Mode 4 again appears the same as for theprior three cases illustrated, as anticipated.
E-field magnitude on the ground for all fourside grounds. Fundamental mode hasrelative null beneath ground plane centeragain, and may have been missed in a probe-excited analysis. Modes 2 and 3 similarlyhave large nulls beneath one or the otheraxis of the heat sink, making it easy to seehow a radiation model with a fixed probelocation as specified in the refererence papermight not couple well to them, either.
Mode 4 again appears the same as for theprior three cases illustrated, as anticipated.
F1=1.35 GHz F2=1.38 GHz F3=1.46 GHz
F4=1.48 GHz
ACES Problem 2000-4
HFSS Predicted E field:Eigensolution with 4 side grounds
8/18/2019 EMC / EMI in HFSS v8
29/42
29
Emissions Testing: Eigensolution Example:Results (Corner Grounds)
E-field magnitude on the groundplane for the cornergrounded case. Here, only the first mode carriessignificant energy beneath the heat sink, while the othersstrongly excite the ground plane edges, but may notcouple to a probe beneath the sink.
The first mode should likely be excited by a probe feedlocated beneath the heat sink, but may not radiate as itcarries very little energy to the ground plane edgesunlike the prior fundamental modes. Therefore this is agood example of a potential on-board (component tocomponent beneath the sink) EMI issue which mightnever show up in radiation measurements or analysis.
E-field magnitude on the groundplane for the cornergrounded case. Here, only the first mode carriessignificant energy beneath the heat sink, while the othersstrongly excite the ground plane edges, but may notcouple to a probe beneath the sink.
The first mode should likely be excited by a probe feed
located beneath the heat sink, but may not radiate as itcarries very little energy to the ground plane edgesunlike the prior fundamental modes. Therefore this is agood example of a potential on-board (component tocomponent beneath the sink) EMI issue which mightnever show up in radiation measurements or analysis.
F1=1.36 GHz F2=1.43 GHz F3=1.48 GHz
F4=1.49 GHz
ACES Problem 2000-4
8/18/2019 EMC / EMI in HFSS v8
30/42
30
Conclusions
w HFSS Eigensolution results appear to fit fairly well with reported Emission Effectivenessw In first 3 cases, fundamental resonance is above frequency where emissions were not
improved by the groundsw However, HFSS appears to imply that the (3, 4) configuration isn’t quite as good as the (1, 2)
w Field plots show how probe may have missed case (3,4)’s fundamental modew HFSS also indicates that there is a lower frequency resonance for the 4-side and corner
grounded cases than shown in the referencew One should have been excited, but does not couple much to the ground edges to
radiate. The other has a relative null at the reported probe location and may not havebeen excited by that technique.
w Solutions took very little time (approx 1 hr total on PI450 for 5 combinations)
Grounds F1 F2 F3 F4 Reported ‘Effectiveness’1 Only 603 MHz 1.34 GHz 1.41 GHz 1.48 GHz to 450 MHz
1, 2 1.04 GHz 1.37 GHz 1.40 GHz 1.48 GHz to 750 MHz
3, 4 994 MHz 1.18 GHz 1.46 GHz 1.48 GHZ to 850 MHz
1, 2, 3, 4 1.35 GHZ 1.38 GHz 1.46 GHz 1.48 GHz to 1.5 GHz
5, 6, 7, 8 1.36 GHz 1.43 GHz 1.48 GHz 1.49 GHz to 2.5 GHz
Grounds F1 F2 F3 F4 Reported ‘Effectiveness’
1 Only 603 MHz 1.34 GHz 1.41 GHz 1.48 GHz to 450 MHz
1, 2 1.04 GHz 1.37 GHz 1.40 GHz 1.48 GHz to 750 MHz
3, 4 994 MHz 1.18 GHz 1.46 GHz 1.48 GHZ to 850 MHz
1, 2, 3, 4 1.35 GHZ 1.38 GHz 1.46 GHz 1.48 GHz to 1.5 GHz
5, 6, 7, 8 1.36 GHz 1.43 GHz 1.48 GHz 1.49 GHz to 2.5 GHz
ACES Problem 2000-4
8/18/2019 EMC / EMI in HFSS v8
31/42
31
0.085 semi- rigid coaxial feed
50 ohm source
3 cm x 4 cmAperature
SMT Termination47 ohm
Metal Enclosure
Example 3:**
ACES Validation Paper (Min Li)EMI from shielding enclosures
** Jim Sherman - Ansoft AE 2001
ACES Validation Min Li 1
8/18/2019 EMC / EMI in HFSS v8
32/42
32
Problem Description**
w Shielding enclosures require apertures (holes).w Heat dissipationw Unused or open I/O connector portsw Weight reductionw Non-metal shielding
w Compute EMI at distance from aperture.w FCC Class B radiation limits
w Write Post Processing Macros:w
Compute magnitude of E Field 3m from aperture vs frequencyw Compute power dissipated in load resistor in cavity
ACES Validation Min Li 1
** Complete user exercise will beavailable as download or disk.
8/18/2019 EMC / EMI in HFSS v8
33/42
33
Model and Solution Reduction Tricks
w Use 2D conductors for all metal.
w Make all metal Perfect Electrical Conductor (PEC).
w Use symmetrical H wall to reduce model in half.
w Replace complex coax feed with lumped gap port.
w Model the resistor load as a 2D surface impedance.
w Model the aperture as a simple 2D H boundary.
w Model the input probe as a narrow 2D rectangular strip.
w Use Fast Frequency Sweep with post processor macro.
ACES Validation Min Li 1
8/18/2019 EMC / EMI in HFSS v8
34/42
34
HFSS Model
Half modelHalf model
airbox
cavity
probe
port1
res_47
hole
50 ohmGap source
ACES Validation Min Li 1
8/18/2019 EMC / EMI in HFSS v8
35/42
35
HFSS vs Measured ResultsACES Validation Min Li 1
HFSS predicted
after 6 adaptivepasses
Measured Data
TMy101
TMy111
TMy201
8/18/2019 EMC / EMI in HFSS v8
36/42
36
sV
Source voltage 1mV
0 Z Coax Input
Input Probe
= 50 Ohm
11S
Inside the enclosure
w Use the Maxwell Plot Utility to generate plots of delivered power
( )2
110
2
18
S Z
V P s −=
Power delivered to the enclosure:
HFSS vs Measured ResultsACES Validation Min Li 1
Measured Power Delivered vs Frequency
8/18/2019 EMC / EMI in HFSS v8
37/42
37
w Create and save S11 magnitude plot
w From Maxwell Executive Commandsw Select Post Process > Matrix Plot
w Plot > New Plotw Data Type S Matrixw Quantity
w Port1,Mode1;Port1,Mode1 (S11)w Cartesian vs frequencyw Plot scaling: Unscaled
w Plot > Save: s11mag.dat
Using The Plot Utility Calculator ACES Validation Min Li 1
8/18/2019 EMC / EMI in HFSS v8
38/42
38
w Modify S11 plot to show Delivered Power
w From Maxwell Control Panelw Select Utilitiesw
Select PlotData
w Plot > Open : s11mag.dat(located in project directory)
The plot data program will open
ASCII text files that containcolumns of data that are spacedelimited. You can import yourmeasured data simply byopening the ASCII text file.
The plot data program will openASCII text files that containcolumns of data that are spacedelimited. You can import yourmeasured data simply byopening the ASCII text file.
Using The Plot Utility Calculator ACES Validation Min Li 1
8/18/2019 EMC / EMI in HFSS v8
39/42
39
w Use plot calculator to modify plot
w Select Tools > Calculator w Perform steps:
w copy s11 to stackw 2 Enter, Y x
w 1 CHS Enter, *
w 1 Enter, +w 1e-3 Enter w 1e-3 Enter, *, *w 8 Enter, 50, Enter,w *w /w
1e-9 Enter, /w Load , Donew Plot > New
( )2110
2
18
S Z
V P s −=
Calculate Del ivered Pow er
Using The Plot Utility Calculator ACES Validation Min Li 1
8/18/2019 EMC / EMI in HFSS v8
40/42
40
ACES Validation Min Li 1
HFSS vs Measured Results
HFSS predictedafter 6 adaptivepasses
Measured Data
8/18/2019 EMC / EMI in HFSS v8
41/42
41
Conclusions
w Simplify the model for faster solution and reduced model space.
w Repetitive steps can be performed automatically with macros.
w Macro’s can be used to produce additional post processing
results.
w The Maxwell Plot Utility Calculator includespowerful math functions.
w The HFSS predicted results are very close after 6 passes.w However,
additional passes are required for pinpoint results.
ACES Validation Min Li 1
EMC
8/18/2019 EMC / EMI in HFSS v8
42/42
42
EMC / EMI References
[1] (ACES) Applied Computation Electromagnetics Society:http://aces.ee.olemiss.edu
[2] EMC at Univ. of Hamburg:http://www.tu-harburg.de/et1/Emc/index.html
[3] Henry W Ott., Noise Reduction Techniques in Electronic Systems, WileyInterscience, 2nd edition, 1988
[4] Tim Williams, EMC for Product Designers , Butterworth-Heinemann, 1992.
[5] C. R. Paul.(Introduct ion to) Electromagnetic Compatibi l i ty, Wiley Interscience, 1992.
[6] Tsaliovich, A., Cable Shielding for Electromagnetic Com patibi l ity , Van NostrandReinhold, 1995.
[7] Perez, Handbook of Elec t romagnet ic Compat ib il i ty , Academic Press, 1995