Modelithics, Inc Logo
World’s Best RF & Microwave Simulation Models


  • Supported Simulators/Operating Systems
  • Ansys® HFSS™: Windows 10, 8, Linux
  • Cadence AWRDE®: Windows 10, 8
  • Cadence Spectre RF: Linux
  • Keysight ADS: Windows 10, 8, RHEL 7, RHEL 6
  • Keysight Genesys: Windows 10, 8
  • Sonnet Suites: Windows 10, 8, Linux
  • Licensing Servers: Windows, Linux
  • Supported Simulators/Versions
  • Ansys® HFSS™ : 2020 R1, R2, 2021 R1, R2, 2022 R1
  • Cadence AWRDE®: 16, 15, 14
  • Cadence Spectre RF: 21.1,20.1,19.1,18.1
  • Keysight ADS: 2022, 2021, 2020 and all updates
  • Keysight Genesys: 2022, 2020, 2018
  • Sonnet Suites: 18.52, 17.52
  • Current Release Versions
  • Simulator Modelithics Library
    Ansys® HFSS™ 22.0
    Cadence AWRDE 22.2
    Cadence Spectre RF 21.8
    Keysight ADS 22.5
    Keysight Genesys 22.3
    Keysight SystemVue 21.4
    Sonnet Suites 21.9

Quick Start & Licensing Videos

Modelithics QuickStart for Keysight ADS

Modelithics QuickStart for Cadence AWRDE®

Modelithics QuickStart for ANSYS® HFSS™

Modelithics QuickStart for Sonnet® Suites™

Modelithics Licensing Setup™

QuickStart Tutorial For Keysight Pathwave RF Synthesis (Genesys)™


FAQ - Filter


Can I perform statistic simulations using the Part Value input parameter in CLR library models?

For statistical simulations Monte Carlo, Yield, etc the Tolerance parameter should be used, and NOT the Part Value parameter. The nominal value for the Tolerance parameter should always be equal to 1. Statistics should be attached to this parameter as *percentage* variations around the nominal part value. This information can generally be obtained from the component vendor.

If the statistics are instead attached to the Part Value parameter itself then all parasitic elements of the model will vary along with the nominal element capacitance, inductance or resistance depending on the model and this may introduce excessive performance variation into the simulation.

In general the parasitic and higher order effects do not scale in direct proportion to the nominal element value and more realistic predictions are typically obtained by applying statistical distributions to only the nominal element. Note that the simulator will not issue warnings if statistics are attached to the nominal part value as the internal code has no way of detecting this.

If the Modelithics part models the component-to-substrate characteristics and EM based layout simulations model the pad and trace within the overall EM structure, does the effect of the pad-to-substrate get duplicated when using Microwave Office?

It is our position that the pad effects are best modeled using the Modelithics models, primarily because the end effects beneath the part will not be the same with and without a part soldered in place. Our models are based on measurements of this actual environment including the component/pad interaction, while the planar simulator will not take things into account in the same way.

The recommended approach is to terminate the input feedline with a stepinwidth that is appropriate to match the dimensions of the feedline to that of the pad stack, and place the separation between EM simulation and circuit simulation at that point. Having this occur in an automated fashion can be simulator specific, as is the translation from a schematic to a layout. In some cases, the simulator will connect the pad stack edges to the edge of a feedline as one might expect while layout translation may move the center of the pad stack to the edge of a feedline.

We are discussing similar issues with the EDA vendors, to try to determine the most user-friendly and transparent solution. Again, our suggested approach is to exclude the pads from the EM simulation and always to include steps-in-width as necessary.

What is meant by the term Modelithics Global Model™?

A Modelithics Global Model is a specific method of using a single model to represent an entire range of surface mount components on a wide variety of substrate materials. Since a single model represents every available value for a specific component, that object can be used in optimization routines that choose the best value to satisfy your design objectives.

The second unique feature of Global Models is that they can be applied to most substrate materials. Simply specify the substrate thickness and dielectric constant, and the model will accurately emulate the component on the specific material chosen for your design.

How can the Global part-value scalable models in the CLR library be accurate if the parasitics vary from part to part, and even in a nonmonotonic fashion?
Modelithics Global part-value scalable models are extracted from measurement data and the variation of the equivalent circuit parameters—including all parasitics—is therefore characterized in a very accurate way. The resulting Global Models have been developed specifically to capture the variation in parasitics from part-to-part, even if that variation is nonmonotonic.
Receive ADS Error: Cannot load component symbol when I select a component symbol.
The Modelithics models are designed to run with ADS versions 2012 and later. This error usually results if you are using an earlier version of ADS.
What kinds of transistor models can Modelithics extract?
Generally speaking, Modelithics can extract the wide range of models available in Agilent`s ICCAP software for MOSFETS, MESFETs, HEMTs, BJTs and HBTs, plus other models not yet addressed by ICCAP. This includes models such as Curtice, EEHEMT, and Angelov for FETs/HEMTs, and GummelPoon and VBIC for BJTs/HBTs. Custom models are also developed depending on your project requirements
Does Modelithics only handle packaged devices?
Modelithics can handle on wafer, chip and wire, as well as packaged devices on a variety of mounting or circuit-board interfacing environments.
Some vendors suggest that their models are pad stack independent, and provide suggested ways to incorporate custom pad stack effects into the models they provide. Why are Modelithics models provided with fixed pad stack dimensions?
When an SMD is soldered to a pad stack on a PCB, a coupled system is formed comprised of the part, the pads and the substrate. It is our experience that pad stack effects cannot be removed from SMD models in a generic and accurate way. For example, if 20 different capacitors from the same family are mounted to identical pad stacks, there is not one representation of the pad stack effects that can be de-embedded from all the capacitor models. {Modelithics is currently investigating a modification of the model format in which the pad stack geometry will be variable. In the meantime, you can contact Modelithics with specific questions, and custom models with your pad dimensions can be provided.}
I am using a 59mil FR4 substrate in my design. Some of the models I am using from the CLR library report a warning during simulation that the valid substrate thickness has been exceeded. Is the model still valid, and how can I tell?
Each model in the Modelithics CLR library is developed using measurements from a set of substrates that span a range of thickness and/or dielectric constant. This information is provided with each model`s data sheets. The substrates used for characterization in turn define a range of substrate height-to-dielectric constant ratio over which the model is validated. For example, a typical H/Er range used to develop a substrate scalable model is 0.1mm/3.6=0.027mm to 1.524mm/3.6=0.427mm.

In case a model with the example H/Er range is simulated outside this range, the simulator will display warnings, however, our equivalent circuit scalable models can be used outside this range without losing accuracy.

The recommended model usage range, outside this validated substrate range, without sacrificing accuracy is 0.022 to 0.513 mm, or 20% below the model validated minimum H/Er and 20% above the model validated maximum H/Er.

If a particular design requires that the model is simulated the extended recommended range, the model’s accuracy will degrade and it is recommended to discuss additional validations to improve model’s accuracy – please contact our Support Team for options to address this.
For some models in the CLR library, why is the predicted Qfactor different from what is published in the vendor data sheets?
The Qfactor in vendor data sheets is calculated from measurements made with the part mounted in a specific test fixture or sometimes from expected trends in part performance. When a part is mounted on a board, the substrate related parasitics affect the overall impedance, which in turn change the calculated Qfactor.

Fairly significant Qfactor reduction can be observed just due to the loss of an FR4 PCB beneath the part, particularly for low-loss components. The effective series resistance included in the CLR library models is based on measurements made by Modelithics using either a Boonton resonant coaxial line or an RF impedance analyzer.

I observe a nonmonotonic variation in the predicted resonant frequency when I sweep the part value in some of the CLR models what is the explanation for this?
Surface mount components often demonstrate nonmonotonic variation of the resonant frequency as a function of the part value. Consider the example of a chip inductor: when increasing the part value a manufacturer may change the wire length, diameter, spacing, and winding pattern. It is thus possible for the interwinding capacitance to decrease while the inductance increases from one part to the next, and the resonant frequency may therefore increase with increasing part value. Although less common, similar effects can happen with chip capacitors as plate number, spacing, and overlap are modified to change the nominal capacitance value.

Being measurement-based, Modelithics CLR models will accurately predict resonant frequency variation versus part value. However, a user must be aware of the upper frequency limit for which a given model has been validated; beyond this upper frequency the prediction of resonant frequency cannot be guaranteed to be accurate. In some cases, the trend in resonant frequency versus part value may also vary non-monotonically at part values that do not correspond to actual vendor part numbers.

Data models based on S2P files seem a very cost-effective solution. What justifies the use of an equivalent circuit model?
In some cases, the use of S-parameter data is a good solution that provides results with sufficient accuracy. However, there are many good arguments for using equivalent circuit models, including: the ability to perform statistical analyses Monte Carlo simulations, the fact that equivalent circuit models will always be accurate at DC whereas extrapolation errors can occur when using S2P files below, and above, the frequency range of the data file, and the ability to accurately predict Q-factor.

Two very important advantages of Modelithics` Global Models are the ability to perform continuous part-value optimization which generally converges more quickly than when using discrete part-value optimization and the ability to retain accuracy on a wide range of substrate types. It is also useful to consider that the time and expense needed to set up measurements of one surface mount part is about the same as the cost to purchase a fully-documented, part and substrate-scalable model from Modelithics!

In looking at Modelithics measured and modeled resonance behavior vs. frequency for series capacitors, the first self resonance tends to shift up in frequency, as the substrate thickness becomes thinner everything else held constant . What is the explanation for why the resonant frequency increases with decreasing H?
A simplified circuit model for a surface mount capacitor is a series RLC with two shunt C`s pi network. In a 2-port configuration the frequency at which ImZin = 0 increases as the shunt capacitance increases. The frequency also increases as the series L decreases. Both effects occur as the substrate height decreases.

In the RLC-shunt C pi-network the classical "resonance" where the impedance is real goes away as the shunt capacitance increases, so its more correct to refer to it as the S11 minimum. Also if the shunt capacitance starts to get relatively large larger than we typically see for surface mount RLCs the S11 minimum may decrease with decreasing substrate height.

How do changes in metallization thickness in the substrate definition affect Modelithics Global Model performance?
It affects the Zo and beta calculation for the solder pad models, which are treated like microstrip sections. This parameter, combined with the height from the ground affects the effective series inductance (ESL), for example, in ESL, in capacitor models. The effects are usually small except for very thin substrates (5 mil and below).
Is there a Modelithics installation guide for ADS?
Yes. Click here for the ADS installation guide for Windows, or here for the Linux DK installation guide.
I've run the Modelithics installer for ADS (or loaded the design kits) and can't find the models. Where are they?
In the ADS schematic window use the palette selection drop down in the toolbar to find the Modelithics library components. The Modelithics libraries all start with MDLX_ and (starting in version 7.0) should be located near the top of the list. In older versions the libraries will be toward the end of the list (mdlx_).

The Modelithics models can also be accessed using the ADS Component Library (bookshelf icon on the toolbar). The libraries will show in the Encoded Library block under Analog/RF Libraries.
Is there a Modelithics installation guide for AWR?
Yes. Click here for the AWR installation guide
After installation I can't find the models in AWR (Modelithics versions 6.0+)
Starting with version 6.0 of the Modelithics CLR library and the corresponding versions of the NLD and NLT the library were modified (the SLC was initially released in this format) to use the process library aspect of AWR. The Modelithics library will need to be loaded into the project to access the models. This is only required once per project. This procure will be required for every project. The library can be loaded using either of the following methods:

For new projects
From the AWR Design Environment main menu:
  1. Click on the menu bar File->New with Library->Browse...
  2. Double-click on the Modelithics folder
  3. Select the Modelithics_Library.ini file
  4. Press the Open button
  5. Open the Elements Browser
  6. The Modelithics Library models are in the Circuit Elements->X Libraries tree
    For existing projects or to load multiple process libraries:
    From the AWR Design Environment main menu:
  7. Click on the menu bar Project->Process Library->Add/Remove Library...
  8. Click the Add... button
  9. Double-click on the Modelithics folder
  10. Select the Modelithics_Library.ini file
  11. Press the Open button
  12. Click the OK Button
  13. Open the Elements Browser
  14. The Modelithics Library models are in the Circuit Elements->X Libraries tree
I'm using AWR and I have projects the use a previous version of the Modelithics library and they do not work with the current version. Is there a conversion tool or process?
Yes, the older versions of the Modelithcs library can be converted to use the latest installation. To convert an old project:

  1. Load the Modelithics Library in a new (empty) project
  2. Now open an existing project that uses an older version of the library
  3. Load the Modelithics Library again using the method for an existing project
  4. Save the project and reload.
When I run a simulation I get an error that the model is not licensed, what do I need to do?
If this is the first time you have tried to simulate a Modelithics model on this computer then you need to set the modelithic_LICENSE environment variable. The variable name is case-sensitive in Linux, but not in Windows

  1. Shut down the simulator
  2. Open the Windows Control Panel
  3. Open the System component
  4. Select Advanced
  5. Click the Environment Variable button
  6. Click the New... button
    For just the current user use the one under the User panel
    For all users (and if you have the proper security) for all users
  7. In the name field type MODELITHIC_LICENSE
  8. In the value filed type the license file location
    For node- and dongle-locked licenses this is a path (i.e. c:\Modelithics\license)
    For floating licenses this is a server name or IP address (i.e. or @
  9. Click OK
  10. Click OK
  11. Click OK
  12. Restart the simulator
What type of license do I have?
Modelithics has 4 types of licenses: node/dongle-locked, floating and cloud. Node and dongle locked will have a HOSTID parameter in each of the LICENSE lines. Floating licenses have a SERVER and VENDOR line at the top of the file. Cloud licenses begin with the keyword CUSTOMER.
When does my license expire?
Each line in the license file contains the date when that feature will expire.
Is there a user's manual?
Yes. The user's manual and release notes are both included in the installation. During the installation process a shortcut is placed in the start menu for both documents. Go to Start->Programs->Modelithics for the shortcuts.
Where are the release notes?
The user's manual and release notes are both included in the installation. During the installation process a shortcut is placed in the start menu for both documents. Go to Start->Programs->Modelithics for the shortcuts. You can also see the release notes for each version here.
What are the limits/range/properties for <model>?
Each model has a descriptive datasheet which will answer most of your questions regarding the range of the model and its parameters as well as the model performance. The datasheets are available from the Windows Start menu in the Programs->Modelithics sub-menu. You can also access the datasheet in each of our supported simulators.

In ADS the datasheets can be viewed by opening the model properties window and clicking the Help button.

In AWR the datasheets can be found by either right-clicking on the model (in either the schematic or the element browser) and selecting Element Help or by opening the model properties window and click the Vendor Help button.

To view the datasheets from Genesys select the model in the part selector and press the Reference Info button (white letter i on a blue back ground with a yellow circle on the part selector toolbar). Prior to Genesys 2009.04 the Reference Info button can also be accessed in the model properties window.
If a model or S-parameter data set is available from the component or device manufacturer or "vendor", why might a designer want to use a Modelithics model for that part?
For many component types, Modelithics offers unique proprietary model formats in most cases are the most accurate and flexibly scalable models available world-wide. For all models we offer best-in-class models that also often offer scalability features that go beyond the capabilities of typically available vendor models in terms of temperature, quiescent bias, substrate (board properties) and geometry.

We are the only company we know of providing substrate-scalable models for surface mount components. Many surface mount component models also have built-in solder pad scalability and pad de-embedding features. As an independent provider – we can offer uniformly high quality and consistent and thorough documentation of every model . Our models are extracted from our own measurements, so they also represent a 3rd party representation of the manufacturers devices.

That said, some passive vendors do a good job in providing S-parameter libraries for their components, but these are generally only valid on the type of substrate used in their test fixture. Some foundries and transistor manufactures do a good job providing non-linear models, however, if you pick 10 suppliers or foundries you will find 10 different formats, levels of quality and usually little or no documentation of measured to modeled performance, reference planes, validation space and so forth. In summary, Modelithics provides consistently accurate, highly scalable and very well documented models for all components included in The Modelithics Library.

Is there a Modelithics Library installation guide for Genesys?
Yes. Click here for the Genesys installation guide
How does the Modelithics’ Maintenance Program work?
We have two levels of maintenance: Platinum and Silver.

Platinum (15%): This level of maintenance is for library and expandable sub-library (e.g. single vendor sub-libraries) purchases. Customers will receive upgrades that include new component models and updated models and Modelithics will ensure compatibility with simulation software upgrades along with technical support.

Silver (10%): This level of maintenance is for fixed size custom sub-libraries. Customers will receive upgrades that include applicable model updates and Modelithics will ensure compatibility with simulation software upgrades along with technical support.
ADS Simulation error on a Windows based computer. The application failed to start because its side by side configuration is incorrect. This can be caused by missing dependent libraries or trying to load libraries of the wrong architecture. For example you cannot use 32 bit libraries with 64 bit applications (vice versa)
One of 2 issues can cause the error:

First if using ADS2009U1, verify the hotfix installed is 351.508 or greater. If applicable, after installing the hotfix simulate the model again.

If you still have the problem after the hot fix has been verified or your using a different version of ADS then a problem exists with Windows installed components.

Please try this:

1) Close all applications.
2) Uninstall all Microsoft Visual C++ 2005 Redistributable Packages or Microsoft Visual C++ 2005 Redistributable Packages (x64) from the computer.
3) Reboot the computer - (This ensures no files are still locked by any application)
4) Install the following programs in the EXACT order shown:

For 32 bit Windows operating systems:

c) vcredist_x86.EXE

For 64 bit Windows operating systems:

c) vcredist_x64.EXE

5) Launch ADS and simulate the desired model
When simulating many of our high-power designs, it's critical that we monitor dissipated power in each capacitor across frequency. Since we don't know the exact ESR of the capacitor (particularly near resonance), I can't see how to do this from outside the model. I'm wondering what you can suggest?
A good estimate for the effective series resistance is obtained by assuming a pi-network representation for the capacitor, in which case ESR is approximately equal to re(-1/Y21). This value, in combination with the current passing through the capacitor - available through a current probe - can be used to compute the dissipated RF power.
Is it possible to compute the classical capacitor parameters (ESR, Q, SRF) from the model?
A good way to calculate the ESR is to simulate the model in a 2-port configuration using Sim_Mode = 2 (no pads) and calculate ESR = Re(-1/Y21); this result will be relatively insensitive to the substrate which is selected. The best way to calculate the Q-factor is to simulate the capacitor in a 1-port and calculate Q = -Im(Zin)/Re(Zin); this result will be sensitive to Q and it may be desirable to use a relatively thick (e.g. 60 mil) substrate to minimize substrate effects. The SRF is best determined by observing S11 for a series 2-port simulation; this will be sensitive to the substrate that is used.
Do you have examples of (ESR, Q, SRF) computed on various substrates?
ESR is rather insensitive to the substrate. Q and SRF are quite dependent upon substrate thickness, and the dependence varies on the component size, construction value and the substrate. We advise running trial simulations with models of interest as discussed in the first question.
What is the advantage of staying on Silver Maintenance Support?

The advantage to keeping maintenance for our models really comes into play as the simulators release new versions of their software. If for some reason their upgrades change the way our models work then we would be available to work with their technicians and resolve the issues. Additionally, if there are upgrades to your models you would receive them.

How do Modelithics transistor models differ from what I can download from the vendor website?

Vendor models vary in format and quality from no model at all or a poor model, to very good models.

How to release a Modelithics license in ADS ?
To release the license when using ADS, one of the three below options must be performed:

  1. From the ADS schematic windows click Simulator->Stop and Release Simulator
  2. Close the workspace or project
  3. Close ADS

An update in the version 21.0 of the Modelithics Library for ADS, a new feature automatically releases the Modelithics License after a specified time. The default time is 30 minutes and can be configured using the MDLX_LICENSE_TIMEOUT environment variable to specify the number of minutes before checking the license back in. The minimum value is 15 minutes.
I noticed board vendor information on Er values are in many cases different than those values used in Modelithics Substrate Library, why is this?
The provided parameters for Er values are based on Modelithics experience with these substrates as good values to use for microstrip simulations. Modelithics has collected significant quantities of repeated measurements of microstrip formatted dielectric constants as part of its characterization work on a wide range of boards.

The default values are those used by Modelithics in its microstrip simulations. Dielectric loss tangent information, on the other hand, typically comes from manufacturer’s data sheets. The substrate library is provided only for the convenience and potential utility of Modelithics customers.Modelithics makes no specific claim as to the absolute accuracy of this data other than we have found these values to be practically valid substrate definitions for the substrates specified.
Why doesn't the Modelithics license release automatically in ADS ?
The Modelithics library and its licenses are tied to the simulator. When the simulator loads the Modelithics library a license is checked out. The library cannot check a license back in until the simulator releases the library.

At this time the simulators do not indicate to our models when a simulation is complete, so we have no way to know when to release a license. Currently the only methods to release a Modelithics library license is to release/unload the simulator.

Below are 3 different methods to unload the simulator:

  1. From the ADS schematic windows click Simulator->Stop and Release Simulator
  2. Close the project or workspace
  3. Exit ADS
What are the library licensing options available ?
Licensing Options
How to release a Modelithics license in ADS ?
To release the license when using ADS, one of the three below options must be performed:

  1. From the ADS schematic windows click Simulator->Stop and Release Simulator
  2. Close the workspace or project
  3. Close ADS

An update in the version 21.0 of the Modelithics Library for ADS, a new feature automatically releases the Modelithics License after a specified time. The default time is 30 minutes and can be configured using the MDLX_LICENSE_TIMEOUT environment variable to specify the number of minutes before checking the license back in. The minimum value is 15 minutes.