Get email delivery of the Cadence blog featured here
The SPB16.2 PCB SI release now contains the Electromagnetic Solution 2D Full Wave field solver (EMS2D).
High density interconnect on PCB and packaging designs with signal switch rates over 5 Gpbs require model characterizations that can support frequency ranges from DC up to THz. Within this wide spectrum, electrical resonance, oscillation, signal dispersion and EM radiation are all likely and must be accounted for. Static or Quasi-static characterization such as BEM2D is not able to address these high frequency issues. Skin effect and dielectric loss are analyzed by simple formulation or empirical equations. Therefore, a full-wave solution is needed to handle these electromagnetic interaction effects.
The Electromagnetic Solution 2D Full Wave field solver (EMS2D) provides the full-frequency range analysis from DC, through the middle frequency range which covers the skin effect, to the THz range of the electromagnetic interactions which address resonances, radiations and EM signal integrity issues. EMS2D is implemented using the finite element method (FEM), which complements Allegro's moment-based BEM2D field solver. EMS2D combines multiple EM computation modules, static, quasi-TEM, and full-wave analysis. Additionally, EMS2D is able to analyze arbitrary transmission line-type and waveguide structures over PCB cross-sections and provide characterized models in table format.
The EMS2D full wave field solver supports the following capabilities:
Coplanar Waveguide Characterization
EMS2D supports analyses of coplaner waveguide (CPW) structures, including single and differential coupled CPWs in differential pair, microstrip, or stripline types. There's a good bit of technical details to this capability and I'd be happy to provide details to anyone who request it.
Interconnect libraries in products that support EMS2D contain a number of enhancements. They include:
Trapezoidal shapes All trace models support trapezoidal cross-section shapes. A trapezoidal keyword in the model syntax defines traces with a trapezoidal cross-section.
Frequency dependent materialsMaterial statements in interconnect models are enhanced to accommodate frequency dependent materials.
EMS2D extracts S-Parameters when the segment length of interconnect is specified. In such cases, the S-Parameter is output in Touchstone file format (.snp) that you can view in SigWave. The associated frequency points will be specified in the frequency point (.frequency) file. The command line option for this feature is
-sparam <filename.snp> –length <a_number_in_meters> –frequencypointfile <filename.frequency>
These parameters can also be set in the EMS2D Preferences form, accessed from the Analysis Preferences dialog boxes in PCB SI and SigXplorer.
Lossy Transmission Line Modeling in HSPICE
EMS2D analyzes lossy transmission line models with skin effects when provided with sufficient multiple frequency points to cover the skin effect range. The command line option for this feature is
Sample lossy transmission line model
* RLGC parameter for a 10-conductor lossy* frequency-dependent line10* L03.30259e-07 2.31592e-08 3.31989e-07 5.2425e-09 2.31151e-08 3.31968e-07 1.68377e-09 5.21857e-09 2.31153e-08 3.31983e-07 6.07076e-10 1.68377e-09 5.24252e-09 2.31592e-08 3.30253e-07 1.14758e-07 4.3743e-08 1.33025e-08 4.57592e-09 1.68688e-09 5.27234e-07 4.37807e-08 1.13685e-07 4.35694e-08 1.32506e-08 4.58298e-09 1.11103e-07 5.27028e-07 1.33205e-08 4.35704e-08 1.13627e-07 4.35706e-08 1.33205e-08 3.55001e-08 1.10656e-07 5.26878e-07 4.58298e-09 1.32506e-08 4.35696e-08 1.13684e-07 4.37805e-08 1.26163e-08 3.53548e-08 1.10657e-07 5.27021e-07 1.68688e-09 4.57592e-09 1.33025e-08 4.37428e-08 1.14758e-07 4.70793e-09 1.26163e-08 3.55002e-08 1.11103e-07 5.27227e-07 * C01.60927e-10 -5.001e-12 1.60325e-10 -6.73228e-14 -4.97538e-12 1.60325e-10 -6.6084e-15 -6.37752e-14 -4.97553e-12 1.60328e-10 -1.89605e-15 -6.6083e-15 -6.73204e-14 -5.00116e-12 1.6093e-10 -3.35033e-11 -5.08109e-12 -2.67057e-13 -4.97631e-14 -1.53303e-14 1.04442e-10 -5.15666e-12 -3.16128e-11 -5.01411e-12 -2.5613e-13 -5.51085e-14 -1.76851e-11 1.08041e-10 -2.86835e-13 -5.01436e-12 -3.16076e-11 -5.01442e-12 -2.86827e-13 -1.89344e-12 -1.72439e-11 1.08077e-10 -5.51105e-14 -2.56134e-13 -5.01418e-12 -3.16137e-11 -5.15674e-12 -4.36895e-13 -1.79772e-12 -1.72442e-11 1.08042e-10 -1.53306e-14 -4.97637e-14 -2.67051e-13 -5.08117e-12 -3.35043e-11 -1.24375e-13 -4.36891e-13 -1.89343e-12 -1.76855e-11 1.04444e-10 * R06.59632 4.98763e-05 6.59632 4.98763e-05 4.98763e-05 6.59632 4.98763e-05 4.98763e-05 4.98763e-05 6.59606 4.98763e-05 4.98763e-05 4.98763e-05 4.98763e-05 6.59606 4.98762e-05 4.98762e-05 4.98763e-05 4.98763e-05 4.98763e-05 6.59632 4.98762e-05 4.98762e-05 4.98762e-05 4.98763e-05 4.98763e-05 4.98762e-05 6.59632 4.98763e-05 4.98762e-05 4.98762e-05 4.98762e-05 4.98763e-05 4.98763e-05 4.98762e-05 6.59632 4.98763e-05 4.98763e-05 4.98762e-05 4.98762e-05 4.98762e-05 4.98763e-05 4.98763e-05 4.98762e-05 6.59606 4.98763e-05 4.98763e-05 4.98763e-05 4.98762e-05 4.98762e-05 4.98763e-05 4.98763e-05 4.98763e-05 4.98762e-05 6.59606 * G00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 * Rs4.09704e-08 4.17396e-10 3.85053e-08 3.07264e-11 3.38072e-10 3.85034e-08 4.08935e-12 2.44874e-11 3.37475e-10 3.85039e-08 1.27611e-12 4.29392e-12 3.08721e-11 4.17211e-10 4.09693e-08 3.46744e-09 1.03583e-09 1.38149e-10 3.44633e-11 1.22189e-11 4.84376e-08 1.03871e-09 3.61622e-09 9.80175e-10 1.25833e-10 3.28181e-11 2.34857e-09 4.66542e-08 1.35478e-10 9.80998e-10 3.61822e-09 9.80111e-10 1.35842e-10 4.80654e-10 2.22529e-09 4.66662e-08 3.26349e-11 1.26568e-10 9.79745e-10 3.61629e-09 1.0385e-09 1.5017e-10 4.57766e-10 2.22207e-09 4.66517e-08 1.20266e-11 3.43279e-11 1.3778e-10 1.03609e-09 3.46801e-09 5.83918e-11 1.5234e-10 4.84458e-10 2.34905e-09 4.84387e-08 * Gd2.22449e-20 -6.91289e-22 2.21617e-20 -9.30604e-24 -6.87747e-22 2.21618e-20 -9.1348e-25 -8.81565e-24 -6.87768e-22 2.21621e-20 -2.62092e-25 -9.13466e-25 -9.30571e-24 -6.9131e-22 2.22453e-20 -4.63117e-21 -7.02359e-22 -3.69153e-23 -6.87876e-24 -2.1191e-24 1.44371e-20 -7.12806e-22 -4.36984e-21 -6.931e-22 -3.54048e-23 -7.61765e-24 -2.44461e-21 1.49345e-20 -3.96492e-23 -6.93135e-22 -4.36912e-21 -6.93144e-22 -3.96481e-23 -2.6173e-22 -2.38362e-21 1.49395e-20 -7.61792e-24 -3.54054e-23 -6.93111e-22 -4.36996e-21 -7.12817e-22 -6.03921e-23 -2.48499e-22 -2.38366e-21 1.49347e-20 -2.11915e-24 -6.87884e-24 -3.69145e-23 -7.0237e-22 -4.6313e-21 -1.71923e-23 -6.03915e-23 -2.61729e-22 -2.44467e-21 1.44373e-20
Command Line Options
This option enables EMS to produce S-parameters to touchstone file filename.snp.
Use -zc 50 to specify the reference impedance needed in S-parameter computation. 50 ohms if the default.
Use -length 0.0508 to specify the length of the model for S-parameter computation. The length is specified in meters. This line length must be provided when -Sparam is set.
0: Include self-inductance in RLGC and S-parameter list1: Exclude self-inductance in RLGC and S-parameter list
Fixed uses the same frequency points as in bem2d. The -fs and -np options are ignored.
Exact uses the frequency points calculated by EMS based on fs, fe, and np.
Fast uses default frequency points.
This option is for advanced use only. It is used to scale the additional layers added by the solver right above the microstrip structures to increase computational space to model open space.
All of the EMS2D field solver features are only available in products that support the new EMS field solver (Allegro PCB SI GXL and XL).
As always, I'm interested in those using this feature in their design analysis and any suggestions you may have.
The “2.5D” has different meanings in different situations. When we say bem2d is 2.5D, we mean we solve the cross-section of a transmission line and do consider longitude component of the field and assume this component is very small (quasi-tem). Is this what you thought?
EMS2D does not consider a via as a volume. EMS2D is solving a cross-section problem too, but in a full-wave sense. Our EMS3D solves via structure as a volume problem. EMS3D is currently scheduled to be in 16.6.
i have a small query.
BEM2D is a 2.5D solver?
Does the EMS2D consider via as volume (volume meshing)?
please share your knowledge on this.
Thanks for your questions. I've updated the RLC matrix example with a different set of data - this should make more sense. The first example was only used to demonstrate the file syntax. I have included a "good" RLC matrix now in the blog post.
The off-diagonal resistance at DC can be non-zero. One example is when a ground plane is an imperfect conductor. Meanwhile it is also possible that those numbers are nothing else but numerical round-offs due to equation and matrix computation. Actually we may truncate those tiny numbers to be zero in a future release.
For the Gd, the value is around ~10^(-16). I am not sure why you think it is a very large number. It just stands for the rate over frequency. Please Email me privately if you're interested in a reference paper in which you can see that the rates about the G matrix are in the same order. But, you are correct about the sign of the Gd. It should be negative as I've not updated in the data file in the original blog post.
Thanks for the update! I am highly curious about the formatting of the RLC matrix data you present here. The off-diagonals of the Ro matrix are non-zero, which implies either that the matrix is not at DC (as Ro should be) or that the Ro matrix is being used to adjust for other resistive effects -- the values are very small. Second, the off-diagonal terms in the Gd matrix are positive and very large, which I would not expect at all. What am I missing?
I have answers to some of your questions.
First though, I'm not a Signal Integrity expert, I'm more of a novice (if even that - ha-ha). So, I asked our Customer Support PCB SI expert - Sue Frederick - to chime in with answers. If you do need any follow-up, I'd encourage you to contact our Customer Support team and ask to speak directly with Sue.
Here are the answers to your questions:
1. Any transmission line structure with rectangle or trapezoidal cross-sections (that can be described in IML format) can be solved by EMS2D. In addition, the cross-section can contain any number of dielectric layers which can be dispersive (frequency dependent Er and loss tangent).
2. In addition to the S parameter output, EMS2D also provides RLGC data and a data file for HSpice to use.
3. The current version does not have a passivity/causality enforcement function. However, you can use our material file capability to specify causal material and then the output model from EMS2D will be causal. So far with all of our testing, the model extracted from EMS2D is always passive since EMS2D is a full-wave based engine. The enforcement will be added the in next version.
4. EMS2D does the EXACT computation for the input data provided by the user through a material file, as mentioned in 3.
5. The Surface roughness modeling is under implementation (see CCR# 622936). This feature should be available in a month or two.
I need urgently basic data for a purchase decision. I'm the user.
1. The allowed input types for describing the structure
2. The available output model types (obviously sparam is one)
3. Passivity / causality enforcement
4. How exactly is loss tangent vs. freq. accounted for by user in input (table, etc.). I know the phenomenon; I want the tool's requirements.
5. Adjustments done for surface loss (ie: roughness)? If yes, exactly how?