'Virtuoso Meets Maxwell' is a blog series aimed at exploring the capabilities and potential of Virtuoso® RF Solution and Virtuoso MultiTech. So, how does Virtuoso meets Maxwell? Now, the Virtuoso platform supports RF designs, and the RF designers measure the physical and radiation effects by using the Maxwell's equations. In addition to providing insights into the useful software enhancements, this series broadcasts the voices of different bloggers and experts about their knowledge and experience of various tools in the Virtuoso IC-Packaging world along with the nuances of RF, microwave, and high frequency designs. Watch out for our posts on Mondays.

Hi Folks, I am back. Before I dive into today's topic, I wanted to let you know that you will find the link to all the previous parts of this series at the very bottom of this post.

Today, I had originally wanted to discuss port setup in the Virtuoso RF Solution, but instead I am going to focus on a slightly different topic which comes up every now and then. Recently, one of my readers sent me an email about this and I wanted to share part of my response with all of you. BTW, in case you were wondering, I love getting questions/responses from you, so please keep sending those.

The question was: how does Virtuoso RF electromagnetic (EM) analysis compare with other third-party EM tools in terms accuracy, performance, capacity, memory consumption, and usability. This is a very natural question to ask and the simple answer is, we compare very favorably. I know you're probably smiling when you read my response (after all, what else would you expect me to say). Hence, I would like to provide a more detailed response below.

But before that, here's one more clarification. As I have mentioned elsewhere, I have worked as a Product Engineer (PE) for nearly 16 years at Cadence. I have dealt with EM analysis/extraction tools for a significant chunk of my tenure at Cadence and also my prior life at Motorola and for my MS degree at University of Arizona. Obviously, that does not make me an expert by any stretch of imagination, it's just that I think it is helpful for you to know my background.

As a PE, I have worked on several tools in the Virtuoso platform, and I have always focused on usability of our tools and flows. One of my guiding principles is that even if we believe that we have a great solution, but a majority of our users (e.g., 7 out of 10 users) say that the feature or the flow is difficult to use, then we definitely need to focus on improving usability. In other words, IMO, while accuracy, performance, memory consumption, and capacity are key figures of merits for the actual EM engine, without a useful flow and tight integration to the layout environment, the tools become me-too and users will see no value-add from such a flow. So, with this introduction, let me now try to address this question about how the Virtuoso RF EM solvers compare with competition in terms of accuracy, performance, memory consumption, capacity, and usability.

Accuracy: Usability is definitely important, but let me tell you that even if we give you the most usable EM solver, but if the accuracy is off, then that serves you no purpose. Hence, first let's talk about accuracy. The EM engines that we have integrated in the Clarity and AXIEM solvers have been around for ~10 years and have undergone many benchmarks with measurement data and third-party tools. Based on all of that, we definitely have high confidence in the accuracy of our EM solvers. Please note that while comparing with other third-party tools is important, but ultimately, we need to fare against lab measurements.

The key in any successful accuracy benchmark (whether against measurement or third-party solver) is to ensure that the comparison is done as orange-to-orange. That's a heavily loaded terminology; so, what does that mean when you're comparing an EM engine to another one or to measurement? It means the following key items:

1. Analysis technique: There are a few mainstream EM analysis techniques that are well established in the industry and academia. These are Method of Moments (MOM), Finite Element Method (FEM), and Finite-Difference Time-Domain (FDTD) - there are some other techniques too but let's focus on these three. While doing a benchmark, we need to pay attention to the following:
   1. Each of these analysis techniques have their advantages and disadvantages. None of these techniques are well suited for all kinds of problems and even if several of these can solve the same problem, they may have additional advantages/disadvantages in terms of speed and capacity.
   2. Even though all three techniques solve Maxwell's equations, certain commercial solutions employ additional techniques like Partial Element Equivalent Circuit (PEEC) and Quasi-Static, which use a partial solution of Maxwell's equations. 
   3. Having said all of this, there are only a few commercial EM solvers in the industry that provide complete (aka full-wave) solution of Maxwell's equations using MOM or FEM or FDTD, i.e., treat the electrical and magnetic fields as coupled. This treatment of E and H (magnetic) fields as coupled is very important for the accuracy of the solution at high frequency (note that "high" is subjective because that's tied to electrical length and not absolute frequency or absolute dimension) as E and H fields cannot be solved separately (or as decoupled) if the solution needs to work for GHz frequencies.
   4. When comparing different EM engines, it is very important to compare a MOM engine to another MOM engine, a FEM to a FEM, etc. But, it's equally important to make sure that the two engines both provide the complete solution to Maxwell's equations and not use PEEC or Quasi-Static techniques, which will make the solution faster, but will have frequency limitations on the accuracy of the solution.
   5. On the other hand, if you're comparing an EM engine to measurement then you have to ensure that the measurement environment/setup is replicated in the EM solver, to the extent possible.
   6. If these above conditions are not met, then the benchmark is not fair (aka orange-to-orange).
   7. In case you're wondering, both AXIEM (MOM) and Clarity (FEM) provide complete solutions to Maxwell's equations and treat electrical and magnetic fields as coupled.
2. Physical stack-up: The physical stack-up needs to be the same. For example when doing measurement using the Vector Network Analyzer (VNA), if a 300um Silicon chuck is used, then the same Silicon thickness needs to be used for the EM engine. When comparing against another engine, if certain simplification/abstraction/modification has been done to the stackup for the other EM engine, then the same ones need to be done for the EM engines in the Virtuoso RF Solution as well.
3. Port setup: The port setups for the EM engines being compared need to be the same and when comparing against lab measurement, the EM engine port setup should mimic the VNA probe setup. For example,
   1. If certain part of the DUT is grounded, then the same structure in the test case needs to act as a reference port for the EM engine.
   2. If unreferenced ports are used in a third-party Method of Moments (MOM) solver, then the same needs to be used in AXIEM.
   3. If port calibration/deembedding is done for some/all ports in one solver, then the same needs to be done in the Virtuoso RF EM solvers.

Typically, I recommend that the comparison for accuracy be done with simpler structures, for example, transmission lines and inductors. This activity allows the user to iron out any differences in terms of stackup, port setup, or any other nuance.

Performance (speed), memory consumption, and capacity: Once the accuracy benchmark is done, then typically the next questions are: how fast does the solver run, how much memory does it consume, and how large the structures can be? At this stage of the benchmark, the user should proceed to more complicated structures like multiple inductors and nets or even to full blocks like all passive structures from an LNA or VCO.

We are very sensitive to the fact that even if our solvers have very good accuracy and very good use model, but takes much longer time or consumes higher memory, or cannot solve the structures that you typically solve, then the Virtuoso RF solvers are not useful for you. Please remember that EM modeling of an inductor has been solved many times over! That is not the real value add of the Virtuoso RF Solution. The EM engines in Virtuoso RF are really geared towards providing high frequency EM analysis of much more complex and real-life structures/layouts. 

Again, based on fair (orange-to-orange) benchmarks with other commercial full-wave EM solvers, we have not seen any limitations with the Virtuoso RF EM solvers for speed, memory consumption, and capacity. Actually, when using the same number of multi-threads on same machines, AXIEM is faster than competition and Clarity is many times faster than competition. In another future post, I will share some comparative data on results from Clarity and AXIEM. Please do keep in mind that, ultimately, the best way to demonstrate accuracy, performance, memory consumption, and capacity to you is to run a benchmark with your toughest structure on your machines/network! Please let us know and we will be happy to assist you.

Usability: Let me reiterate what I said above. Even if we gave you an EM solver, which is accurate i.e., it offers very good correlation with measurement (after all, measurement, not results from a third-party solver, is the golden data) and runs as fast or faster than your favorite EM tool and consumes the same or less memory when compared to your favorite solver and runs the same or larger test cases than (you guessed it) your favorite solver, but the usability is poor compared to another tool, then all we have done is given you another EM engine. And, trust me, the world does not need another me-too EM engine!

From the inception of the Virtuoso RF Solution, we have been focusing on tighter integration and better usability. Users have been running third-party EM analysis engines outside of and from Virtuoso for many years. However, for the EM engines inside the Virtuoso RF Solution, we have taken full advantage of our deep integrations within the Virtuoso platform, which is only possible for our R&D teams. In my personal discussions with customers, we are sometimes reminded that the customer does not need another EM engine. However, once we start demonstrating the state-of-the-art Virtuoso RF Solution with all its capabilities, then the true appreciation comes from the customer. And, this is one of the biggest satisfaction that I get, because I can sympathize with you since I have suffered some of the same challenges when I was using EM tools.

Here are some of the key usability benefits that you will enjoy when you use the Virtuoso RF EM analysis flow:

1. Select parts of the layout without having to delete other parts or create an EM only layout. This is important in a real life, because after several iterations with EM analysis, users end up with multiple variations of the same layout and it is virtually impossible to keep the manufacturable layout (the one that will be sent to the fab as GDS / Gerber) with the various EM-layouts.
2. Automated port setups with manual overrides.
3. Nearly identical integration and flow when using a MOM engine and a FEM engine.
4. Automated stitching of s-parameters for circuit simulation without breaking the schematic used for generating the layout (i.e., doing the Layout Versus Schematic check).
5. Ability to model parts of the layout with an RC extraction engine like Quantus(TM) and parts with a full-wave EM engine and combine these results without worrying about double counting of parasitics.
6. Ability to solve three-dimensional structures like wire bonded and bumped ICs from Virtuoso.
7. Ability to combine structures from IC, package, module, PCB in one EM analysis without manually merging stackups, metal layers, and dielectric layers.

As I mentioned in the beginning of the blog, my intention was to discuss port setup. However, I took a detour and discussed benchmarking a little bit. I hope this is useful to you and you're able to appreciate the objective approach that I have laid out. Until next time, TTFN.

Virtuoso Meets Maxwell: Help with Electromagnetic Analysis Part I, Part II, and Part III

Related Resources

• Virtuoso RF Solution
• What’s New in Virtuoso (ICADVM18.1 Only)
• Video - Running an Electromagnetic Analysis Using AXIEM
• Virtuoso Electromagnetic Solver User Guide

For more information on Cadence circuit design products and services, visit www.cadence.com.

Contact Us

For any questions, general feedback, or even if you want to suggest a future blog topic, write to custom_ic_blogs@cadence.com.

About Virtuoso Meets Maxwell

Virtuoso Meets Maxwell series includes posts about the next-generation die, package, and board design flow with a focus on reinventing and optimizing the design process to ensure that the designer remains a designer! Keep watching!

Click Subscribe to submit your email address for receiving notifications about our latest Virtuoso Meets Maxwell posts. 


(Sutirtha) Kabir