hi jian -help reg. field solvers

[mailgvenkat]

hi jian,

among the several field solver techniques(such as MOM,FDTD,FIT,FDFD) for EM problems, which method is advantageous?....and most widely used......

with regards,
venkat

 

[jian]

Hi, Venkat:

Again, this is a frequently asked question. Every algorithm has its own advantage in different things. For general 3D problems, FEM is good in flexibility and accuracy. However, it is slow and requires much resources for large structures. FDTD and FIT are also good for general 3D structures. It is also more capable for larger structures. However, its accuracy is normally lower. I have not studied FIT. I understand that it might fit arbitrary boundaries better than FDTD. However, my understanding is that finite integration for conformal structures may also involve solving a sparse matrix which may end up with the same kind of large computational resources. MOM is certainly the best choice for planar 3D structures and wire structures no matter whther speed, accuracy and easy of use are concerned. However, MOM normally will be less capable for full-3D structures or structures with complicated dielectrics. The bottom line is no single algorithm will be the best for every type of problems. Best regards.

Jian

 

[mailgvenkat]

hi jian,

thankyou for the reply.... do u have the algorithm for solving maxwells equation using MOM?....

Also what do u mean by guide wavelength?.....

Among various structures(microstrip,stripline,CPW,etc) in practice, which one is widely used for nanometer designs?....

please forward me the same.....expecting ur reply..

with regards,venkat.G.

 

[jian]

HI, Venkat:

I have been working on MOM for 23 years (since I was a senior in my colleage study). I have developed quite some algorithms for it. Most of my latest 16 year's works are proprietary. You can try to find my papers on AP-T and APS and MTT-T and MTT-S. in 1992-1994.

For nano applications, I am not sure which one of the planar circuits are more popular. Howver, I can see that you may need to make sure the strip is not too thin. If it is too thin, there will be much metallic loss.

Best regards.

Added after 4 minutes:

HI, Venkat:

I forgot to answer your question on guide wavelength. I mean "waveguide wavlength". However, I called it "guide wavelength" when I want to save some space. Regards.

 

[mailgvenkat]

hi jian ,

thankyou very much for the valuable suggestions......Can u give me the links of papers related to MOM done by you sir i.e., the links of your paper.......

do any alternate structures(other than those mentioned above in previous reply) exist in practice for MMIC design?.......If so please give me the alternate structures and the links related to it....

eagerly anticipating your reply....

with regards,
Venkat.G.

 

[mailgvenkat]

hi jian,
do any alternate structures exists in practice for MMIC design?.....If so please give me the details regarding it.....

Is there any MATLAB CODES for MOM based field analysi with u.....
i have given a detailed mail regarding this to ur personal mail ID....

Anticipating ur help...
With regards,
venkat.G

 

[jian]

Hi, Venkat:

I don't have any MOM code on MATHLAB. Sorry for it. Can you explain your first question? I am not sure what you mean in your sentence "alternative structures...".

Bset regards,

 

[mailgvenkat]

hi jian,
is there any structures other than microstrip,stripline,CPW,etc used in practice for MMIC design....

sorry tht i have mentioned as alternate structures?....

Sir u have given the geometry of CPW,sgCPW with micrometer dimensions......i have tried to change this to nm...but it indicated some errors regarding port definition......How can i rectify it....please help me sir....

with regards,
venkat.G.

 

[jian]

Hi, Venkat: If you use conductors as the main parts of the waveguides, these should be the most popular planar TLN structures.

For nm, the dimensions are very small. You need to increase the meshing density (meshing frequency and the cells per wavelength) so that the cell size will not be extremely bigger than the trace width etc. Otherwise, the simulator may detect some port dimensions are too small compared to the wavelength of the meshing frequency and it will indicate error to you. Regards.

 

[mailgvenkat]

hi jian,
thanks for the reply...

Is there any option in IE3D to calculate crosstalk,delay,ringing effects.etc.,between two planar tx.lines.....If so please give me the steps to follow...

With regards,
venkat.G.

 

[jian]

Hi, Venkat: Please create two coupled strip structure and define 4 ports on the 2 coupled strips. Then, after simulation with MODUA displaying the 4-port s-parameters, you can use PROCESS->LC-EQUIVALENT or PROCESS->PI-NETWORK EQUIVALENT commands on MODUA to find the C and L values involved.

 

[mailgvenkat]

hi jian,
how to create two coupled strips in IE3D....then by simulating we do have only s-parametrs,LC euivalent parmaters..

but how to do crosstalk analysis....how to find the coupling coefficients(near end crosstalk and far end crosstalk)..

anticipating ur reply..
with regards,
venkat.G.

 

[jian]

Hi, Ventkat: Here is a coupled TLN structure from your gcpw.

 

[mailgvenkat]

hi jian,
thanks for the reply...i will try to simulate the structure...

with regrds,
venkat.G.

 

[mailgvenkat]

hi jian,

Is there any option in IE3D to simulate for crosstalk between two lines i.e.,giving a signal input to the aggressor line(say port1&2)..and to see how much signal have been coupled to the victim line(say port 3&4)...

Please help me to simulate for nm subsrates also...

With regards,
]venkat.G

 

[jian]

Hi, Venkat: Just define 4 ports on the coupled lines and you can see it. In fact, you can use the coupled structure I uploaded some days ago. You can compare the freuency response of the cross-talk on MODUA while you can see the waveform on MDSPICE of hte Zeland products. Regards.

 

[mailgvenkat]

hi jian,
Thanks for the reply...I have simulated the coupled gcpw structure.....i wil try for the freqiency response now.....hope u have done for micrometer.....if i scale down by 10 times....it will result in a lossy structure.....

But i hve tried for scaling...it showed some errors on port definition......so i request u to change the dimensions further i.e, scaling down by 10 times.....
please give me the scaled goemetry file for the coupled_gcpw.......so that i can find tht the coupling is more between lines for nm substarte.....

awaiting for ur reply...
with regards,
venkat.G.

 

[jian]

Hi, Venkat: I tried to scaled it down by 10 and every thing is ok.

 

[mailgvenkat]

hi jian,

thankyou very much for the goemetry file.....i will try to simulate it.....

As far as interconnects is concerned...is there any structures other than CPW,microstrip,stripline exists in practice.......If so please give me the structures.....

also other than these parameters like s-parameters,tx.line paramters,RLC equivalents,nearfield distributions,etc.....is there any other parameters which characterizes the interconnect structure(means affects its performance)?......

do we simulate those in IE3D?.....

anticipating your reply....
with regards,
venkat.G.

 

[jian]

Hi, Venkat: Besides CPW, microstrip and stripline, I believe coupled strips are also used. Basically, for those TLN spanning from 0 to high frequency, you need at least 2 conductors in order to have at least one propagating mode without cutting off frequency. However, you also don't want to use more than 2 because more than 2 conductors will create more than 1 propagating mode. When more than 1 propagating mode happens, it will cause dispersion and it will affect transmission quality no matter time or frequency domain signals are concerned. CPW contains 3 conductors and it has more than 1 propagating mode and it is not good in this aspect. Designers have to connect the two grounds together here or there in order to suppress the 2nd propagating mode. Certainly, CPW has other advantage such as low dispersion on Ereff, easy manufacturing because of no need for vias and easy to connect lumped elements to them because we can connect them on the surface and we do not need to dig holes.

All those parameers (s-parameters, TLN parameters, RLC equivalent circuits) are related anyway. IE3D is excellent for planar cricutis and you can use it for all these kinds of interconnects. Regards.