Continue to Site

Welcome to

Welcome to our site! is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

The difference between CST and Sonnet

Not open for further replies.


Member level 3
Dec 13, 2001
Reaction score
Trophy points
City of Angels
Activity points

Dear Friends,

Anyone can explain me about C S T Vs. S0nnet...?

I have learn in S/o/n/n/e/t web site that they offered C S T to the north americ@ customers.....Confuse! :?

Thxs in advance..... :lol:

Best regards,EDIN

Hi, Edin!
CST is good for large and complex 3D structures with freedom in use of dielectric, loosy materials, metals, gyrotropic... It works in time domain (but it has frequency and eigen solver too).
Sonnet is 2.5D simulator (method of moments) appropriate for layered structures as planar circuits. The aim of those simulators is thus different.
I've found Sonnet much more user friendly compared to IE3D which is relatively easy to use and gives good results.
This is very short introduction, but I expect that there will be posted more opinions from our experienced EMans.

Sonnet simulates planarly layered microstrip-type circuit enclosed in a conducting rectangular box. The result would be definitely better than IE3D if the circuit fits into a uniform grid.

Hi, loucy!

You think that uniform mesh is better that non-uniform? Why?
I'm still waiting for somebody who compares my IE3D results of patch with Sonnet (I'm not so experienced in this simulator).

Cst vs sonnet

Both are Em simulation softwares. But methods are different. You can think this just as collaboration of two companies. Sonnet was selling KCC Microstripes one year ago. KCC is also another 3D simulation program. They are not selling it now.

4 months ago, AWR MW office and sonnet had an aggrement. AWR will implement sonnet software inside their softwares.


Re: Cst vs sonnet

lkuzu said:
Both are Em simulation softwares. But methods are different. You can think this just as collaboration of two companies. Sonnet was selling KCC Microstripes one year ago. KCC is also another 3D simulation program. They are not selling it now.

4 months ago, AWR MW office and sonnet had an aggrement. AWR will implement sonnet software inside their softwares.


it is a nice news.

Does this mean M/W/O will replace their existing solver with the S/o/n/n/e/t version? The geometry editor in M/W/O is sure easy to use.

i'd like to add that sonnet results when achieved are really accurate however you can't trust ohmic loss - better to neglect this effect and use loseless metallization.
also i agree that sonnet analysis is not the fastest of know simulators and i agree that it consumes lots of memory, but it's worth to check your design if you use ie3d...
no CST soft characterization coz i'm 3 days exerienced user :>

With Regards,

Hi, Eirp,

I didn't mean that uniform mesh is better than non-uniform mesh. I think IF the circuit layout fits into a retangular grid used by sonnet, then the its result would be more accurate than IE3D's. My reason is that, in this case, the formulation of sonnet is more rigorous. I believe IE3D calculates the moment matrix by using some accerleration techniques which involves subtle assumptions, and the result MIGHT be less accurate.


Concerning the conductor loss, I think IE3D and Sonnet use the same idea in their model. I suspect the lack of accuracy in Sonnet is due to:
1. the geometry of the circuit is not modelled exactly. Because the current along the edge is more significant and contribute more to the loss, if a curved line is approximated by staircasing, the edge current would not be accurate.
2. some loss mechanism such as the surface wave and radiation is not modelled well in Sonnet. One can blame the enclosure box for that.

Can anyone upload some example projects along with measurements concerning the model of loss?

Can anyone confirm the old news that "AWR MW office and sonnet had an aggrement... AWR will implement sonnet software inside their softwares"?
This looks similar to Ansoft's buying out the HFSS from Agilent.

$ONNET just uses a an approximate formula for the complete region from DC loss to Skin-Effect losses. This causes me many head-aches.
What I did, is compare a measured MMIC inductor with the loss of the simulation and then just fit the Rdc and Rrf values.

But what else do you want to do with a 2.5D simulator? If your gap between the lines is the same magnitude as the thickness of the metal (Gold) than how would you estimate the skin-effect on top, bottom and edges?
You could add arbritrary amount of metal layers in $ONNET and set the skin-loss to zero but the computation time....
The maximum I can afford is two metal layers.
I also used C/S/T to show me the skin-loss but so-far it gave me false results, even after making a ridiculously small grid.

Does anybody have any good suggestion on how to simulate a MMIC inductor (thick metal compared to skin-depth) or a coupling structure from DC to 30 GHz?
Is IE-3D better? Please let me know.

I think doing it in a 3D simulator is impossible? Think about the fine grid you would need to do a lossy inductor.


Hi, wave-maniac

I agree with most of your comments. But I think the Green's functions implemented in IE3D is not completely 3D. It can't handle 3D dielectric objects. It can be used for metallic 3D object such as sphere, but not dielectric sphere. For sonnet, the requirement of uniform grid comes from the fact that FFT (hence, uniform samples) are used to calculate the impedance (moment) matrix. The difference between 2.5D and 3D lies mainly in the way z-directed currents are modelled. Thus if you don't have via etc., 2.5D and 3D green's function should give you the same answer if there were no numerical error.


We are using the MWO to analyze the microstrip filters up to 6GHz and obtains very good comparision between calculated and measured data, and I can note that the metal box (around filter) accounting very good, because it is strongly affects on filters characterics.

With responce,

I think Kit-the-Great pointed out the most significant advantage of Sonnet and MWOffice over IE3D: accurate modeling of the enclosure box.
I noticed IE3D would be very slow if you include 10 and or images to simulate the box.

Sonnet vs. MWS

Here is my 20 cents:)

I know for sure that MWO implemented absolutely same formulation as Sonnet and Sonnet had no choice but to provide Sonnet Lite - reasoning was that Sonnet rejected an initial offer by MWO to cooperate. After all, Sonnet code is published alreadyand AWR folks did it with no problem.
Next, if one should design MMIC I'd never recommend IE3D - their formulation and losses thereof are horrible - and they just did nothing to improve on that, however they claim - Try to make a simple 50 ohm line on Rogers and then deembed both sides to the mid-point and simulate 1-20GHz and you'll be astonished by what your eyes will see. Numerical errors are huge with IE3D.
Third, IE3D uses open Green Functions formulation and naturally calculates reponses (currents) on antenna surface and thus calculates far-field patterns with ease - as opposed to Sonnet, which uses boxed Green Functions. In SOnnet simulating a patchy antenna is a bit loousy and has strict limitation. IE3D, on contrary, uses Free Space Green Functions and can easily simulate radiation problems. I am not very happy with the weighting and expansion functions IE3D uses, but for many problems results are OK.

For thick metal and real 3D dielectric bricks both SW are second to EM3DS. EM3DS is superior to them in many respect.



Not open for further replies.

Part and Inventory Search

Welcome to