HFSS lumped resistors

[tonino888]

resistor hfss

Hi all,
does anybody know how to design lumped resistors in HFSS (v 9.0)?
I'm trying to simulate a Wilkinson Power Divider in stripline.

Thanks in advance,

Antonio.

 

[madengr]

hfss lumped rlc

Place a rectangle where you want the resistor and assign it an RLC boundry condition. Better yet, model a thin film resistor with tantalum nitride. It will have some thinkess which may be very thin. Not sure how well it will model in HFSS. Do you have access to a 2.5D simulator? You will get better results for stripline.

 

[maverick09]

hfss resistor simulation

Hi,

I have copalanr strip lines on ground plane. I am defining a rectangle between the top edges of the two strips. IS that definition right?......I define the boundary of the rectangle as Lumped RLC.....
also how can I define the current line?....

 

[madengr]

lumped model for resistor

Yes, that sounds right. The current line should snap to the center of the square edges that you are defining as the RLC boundry. It works just line defining a lumped port. One thing I watch out for in HFSS is the meshing. Look at the current densities on your conductors. It should be high at the edges and uniform. If you have a blue cell with red cells on either side then your meshing is bad. This will happen even when HFSS converges to the default of delta_S=0.02 dB. You may want to limit the mesh size on teh conductors until the current density is correct. One needs to be able to specify a non-uniform mesh across the microstrip so the center is not over-meshed. Ansoft needs to implement this fearture.

I don't know if a lumped RLC boundry will give an accurate results. I think it will distribute the current in the boundy just as if it were a perfect conductor but I don't know how it will account for the resistance.

I recently did a 3-way, 2 stage Wilkinson on 0.010" alumina (thin film, sputtered metal and resistors). The simulation was done in Sonnet (via MWO) and the simulated vs. measured results were very good, within 0.2 dB for insertion loss. The L/W ratio of thin flim TaN resistors were tweaked with the EM simulator from their calculated DC values.

 

[maverick09]

hfss lumped

hi,

i am using rectangle of 0.001um wide and 0.32 um length with zero thickness as resistor between teh coupled lines......i defined teh current line halfway between teh rectangle perpendicular to the face of teh rectangle(as in lumped port excitation).....i assigned teh value of resistance as 25 Ohms to it.

I am simulating the structure with a waveport and differential port....the differential terminals were assigned 100 ohms for differential and 25 ohms for common mode. When I simulate it, I am actually expecting teh load resistance as 25 Ohm, but when I am calculating using teh standard transmission line equations for ZL, I am getting values not consistent with the load resistance (25 ohms)....

could anyone please point out what am I doing wrong with the simulation?....

 

[EdmundZheng]

high frequency lumped model of resistors

Yes, that sounds right. The current line should snap to the center of the square edges that you are defining as the RLC boundry. It works just line defining a lumped port. One thing I watch out for in HFSS is the meshing. Look at the current densities on your conductors. It should be high at the edges and uniform. If you have a blue cell with red cells on either side then your meshing is bad. This will happen even when HFSS converges to the default of delta_S=0.02 dB. You may want to limit the mesh size on teh conductors until the current density is correct. One needs to be able to specify a non-uniform mesh across the microstrip so the center is not over-meshed. Ansoft needs to implement this fearture.

Hi, friend! I dont know what you mean, about the hfss meshing. I dont know what you mean "a blue cell with red cells on either side then your meshing is bad."" One needs to be able to specify a non-uniform mesh across the microstrip so the center is not over-meshed." could you tell me in detail? thank you in advance!

 

[tonino888]

hfss normalize setting inertion loss

Thank you all for trying to help me;

unfortunately I do not have access to a 2.5 simulator that you suggested to be a better simulator for stripline design.

My questions are:

- how do I have to draw the rectangle to define the region to which assign the RLC boundary condition? I mean, do I have to draw it in the same plane of the striplines or in a perpendicular one starting from the striplines and reaching the ground plane?

- how do I have to define the current line if the rectangle is to be drawn in the same plane than the striplines, and how in the other case?


Thank you very much,

Antonio

 

[maverick09]

hfss thin film resistor

Hi Antonio,

A sfar drawing of rectangle is concerned.......chnage the plane from XY to YZ plane. Draw a rectangle between the strip lines and assign boundary to the rectangle by assigning values of R, L and C as required.

As far as current line is ocncerned .....I am also looking for some help....i am not quite sure but it should be defined perpendicular to the rectangle, i.e parallel to the microstrip lines.

Hope this helps........

 

[madengr]

resistor in hfss

Here is a 10 mil wide microstrip on 10 mil thick alumina driven at 20 GHz. The delta_S=0.02 dB for all cases. The mesh and current density magnitude are plotted.

This is the default mesh and convereged in 9 steps. The nulls and peaks indicate the line is about 1.25 lamda_e long. The current density is not well defined.





Now the mesh is limited to 10 mils on an edge. The edge currents start to appear.




5 mil mesh is better but theere are still yellow spots between the red.




2 mil mesh looks pretty good. The edge currents are well defined and taper off along the X axis. This convereged in 3 steps



Now, I didn't see much difference in S11 vs. frequency for all the cases and probably shouldn't since delta_S=0.02 dB however the 2 mil mesh "looks" right while the deafult is kind of lousy. This is open for debate.

About the non-uniform mesh. The line is about 5 cells wide and the currents are limited to edge cells so the middle of the line is over-meshed. It would be nice to specify the mesh edge length as a fuction of a vector that is snapped to the line. This may cut the mesh size down by 1/2, and allot more for large patches.

 

[madengr]

do lambda refinment

Whoops, that previous post's line length is about 3/4 lambda_e.

Here is an example of a Lumped_RLC boundry. It is a capacitor (actually where a SMT one will go) with vias on either side. I defined the current where it would flow if it were DC.

 

[maverick09]

lumped rlc hfss

Hi,

I have 2 coplanar strips on dielctric and ground plane. I am using a waveport at one end, with 2 modes to simulate, also a differential terminal. I have a load of say 25 ohms attached at one end, which I define by a rectangle with length equal to the gap between the two strips,and width of small thickness. I define a current line going from left to right ?....is this setup right?....also I am trying to calculate the Z0 and Z. I am not getting the same value of resistance (say 25 Ohms) applied as load (load resistance). I am using the reference impedance of 100 ohm for differential mode and 25 ohms for common mode, also the waveport is normalized to 50 Ohms......can anyone explain what am i doing wrong ?

 

[madengr]

ansoft hfss resistor

I have not used diff ports in HFSS so can't help you there. I believe you mentioned your resistor (RLC bondry) length was 32 um with a 1 um width. That ratio seems too much. I would use L/W=2 (which is what you will get anyway for a 100 ohm deposited reistor).

 

[maverick09]

hfss voltage across edge

Hi,

I have 2 coplanar strips on dielctric and ground plane. I am using a waveport at one end, with 2 modes to simulate, also a differential terminal. I have a load of say 25 ohms attached at one end, which I define by a rectangle with length equal to the gap between the two strips,and width of small thickness. I define a current line going from left to right ?....is this setup right?....also I am trying to calculate the Z0 and Z. I am not getting the same value of resistance (say 25 Ohms) applied as load (load resistance). I am using the reference impedance of 100 ohm for differential mode and 25 ohms for common mode, also the waveport is normalized to 50 Ohms......can anyone explain what am i doing wrong ?

i am using submicron lengths.....

 

[tonino888]

hfss model resistor

Hi madengr,

I've been trying to set the mesh dimension of my project to a lower value (as you suggested me showing the four pictures) to improve the accuracy of my results.

I'm not sure about where to set it; I set the new value into the field:

"Mesh Options -- > Maximum Length of Elements"

Is that correct?

Thanks a lot,

Antonio.

 

[madengr]

50 terminal hfss

Yes, thats correct. Just be sure you have the metal selected when you set it. You should then plot the mesh after simulation to be sure it is correct. I had some problems when going from 10,5, to 2 mil mesh. It would sometimes not update unless I deleted all the solutions. I'm not sure if this is a bug with 9.2 or not. I did this before last year and don't remember any problems setting the mesh size.

 

[SAJ25]

lumped resistor model

hi madengr
please upload the hfss files of these simulation:
resistor, capacitor,...

 

[lkminz]

resistor s parameters

hi
i need to simulate a structure with lumped inductor between two microstrip line of certain thickness so what i should do ........i just draw a rectangle sheet and assign RLC boundary over it or there is something else that i can do to take care of thickness of microstrip lines too ???:?:

 

[manikandan_5025]

current densities in hfss

Modeling a Powerbus with HFSS

Geometry and setup

Double-sided PCB:
Size: 125 mm × 100 mm × 1 mm
Top and bottom metal: PEC
Dielectric: FR4, εr = 4.5, dielectric loss tangent = 0.015
Solution type: Driven terminal
Excitation: Voltage source (1 V, 50 ohms)
Boundary: Radiation
Analysis Setup:

Solution Frequency: 1.15 GHz
Maximum Number of Passes: 50
Maximum Delta S: 0.001
Do Lambda Refinement: 0.2
Maximum Refinement Passes: 40%
Sweep:

Sweep type: Discrete
Frequency Setup: 5 MHz - 2 GHz, Step Size = 5 MHz
hfss_powerbus.zip

Simulation result

Simulation Time: 12 hrs
Number of passes completed: 7
Number of tetrahedra: 39325

Decisions the user must make that affect the accuracy of the result

Solution type: driven terminal
Location of absorbing boundary: cylinder, radius=150 mm, height=300 mm
Maximum delta S: default = 0.2, this model = 0.001
Do lambda refinement: default = 0.333, this model =0.1
Maximum refinement passes: default=20%, this model=30%

Comments

How do we select the solution type?

Two kinds of solution types were used in this model, driven modal and driven terminal. Choose the Driven Modal solution type when you want HFSS to calculate the modal-based S-parameters. The S-matrix solutions will be expressed in terms of the incident and reflected powers of waveguide modes. Choose the Driven Terminal solution type when you want HFSS to calculate the far-zone radiation.

How do we select the excitation type?

In HFSS, impedance matrix parameters are computed from the S-parameters and port impedances. Since Lumped ports compute S-parameters directly at the port, it is more efficient to use a lumped port when you want HFSS to calculate the input impedance. Assign a voltage source when you want to specify the voltage and direction of the electric field on a surface. In this model, the power bus is driven by a voltage source with 50 ohm in series resistor. To simulate the voltage source, an idea voltage source was assigned to a rectangle from the edge of the upper layer to the RLC boundary. The RCL boundary is modeled as a 50 ohm resistor which is in series with the idea voltage source.

Screen shots


Fig. 1. Simulation model


Fig. 2. Simulation meshes


Fig. 3. Input impedance


Fig. 4. Electric field at 3 m, θ=0°, φ=0°


Fig. 5. Electric field at 3 m, θ=90°, φ=0°


Fig. 6. Electric field at 3 m, θ=90°, φ=90°