CPW impedance do not match between Sonnet and ADS linecalc

Hi~

i have simulate an ungrounded coplanar waveguide with follloweing parameters

width 4mil
gap 2mil
substrate height 15mil
ε=9.8 (Al2O3)

it gave me the right answer (50Ω) when i used ADS linecalc
picture => http://www.badongo.com/pic/8993815

But when i used Sonnet 11.54, the result its about 55Ω

sonnet setting=> http://www.badongo.com/pic/8993956

it bother me a lot
i dont known if i can trust sonnet or not
Thanks~



PS : sry for my poor english

my main purpose is to find the characteristic impedance of finite ground cpw (FGCPW)

but linecalc and TXline only provide infinite ground CPW

i know that there are some analytical equations to use

but i'm not sure its right

is there any other method to solve this problem?

thank you~~

Hello,

what you get from Sonnet EM is exact values, when you impliment & measure you will get same...

Only you need to properly model in Sonnet exact practical structure,
say in this case (no ground) you can add differntial ports (-1, -2) for return current path & make sure you set the Box bottom Cover to Free space then you will get almost near to 50 ohm...

You can do CPW with ground analysis in Sonnet by making bottom cover to Metal/lossless metal

Modeling Co-Planar Waveguide (CPW) in Sonnet
Tips and App Notes
http://www.sonnetsoftware.com/suppor...r_waveguid.htm

http://www.sonnetsoftware.com/produc.../apps_cpw.html

See the post on CPW
https://www.edaboard.com/ftopic377496.html

Also using TXLINE you can do CPW with/without Ground cxalculations...
See the attached pictures....


---manju----

You can trust Sonnet if your model is correct :D

Your Sonnet model seems to use thin lossless metal, which is not realistic. When I simulate this way, with a reasonable fine cell size of 0.5μm, the Z0 is ~53Ohm.

You should change your model to use the actual metall losses. With the narrow gaps between conductors, you should also switch to "Thick Metal Model", so that the side wall capacitance between the lines is modelled more accurately.
When I simulate this way, with a conductor thickness of 0.5 mil, the Z0 is ~50Ohm.

I have attached both models.

Best regards
Volker

(PS: As a next step, we can check what error from mesh/discretization we have. This can be done by re-analyzing with a smaller cell size, and comparing results. The calculated Z0 will show monotonic convergence to the correct result when we reduce the cell size. This gives a very nice estimate of the error from discretization.)

Thanks guy~~~

i think there are three reason that i got the wrong value

1. memory/speed setup

2. thick metal mode

3. conductor LOSS (more loss will increase Z0 and thinner
conductor will iincrease Z0 too )





thanks again

:D

Be aware that when there is loss (dielectric and/or metal), the Z0 and Keff values are complex. Sonnet will give you both the real and imaginary values. For Z0, I think both Linecalc and TXline give an "impedance" and a "loss", but I don't know if the impedance is the magnitude or the real part of the impendance. Also, they give an effective dielectric constant, but don't say if this is the real part or the magnitude.