The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Modeling >> Transmission Lines and Other Distributed Devices >> mtline with spectre
https://designers-guide.org/forum/YaBB.pl?num=1188127992

Message started by setb2609 on Aug 26^th, 2007, 4:33am

Title: mtline with spectre
Post by setb2609 on Aug 26^th, 2007, 4:33am

Hi,

I have modeled with mtline and the LMG in Cadence 5.1 a differential (Zdiff = 100 ohm) microstrip tramsission line (for use in a LVDS-driver testbench). The mtline-model requests a parameter fmax which shold correspond to the max signal frequency. I started just by using 1 GHz.

For evaluation of the model I first have run a SP-analysis on a decent testbench on the mtline with baluns with termination on each side and a port on one side in order to calculate S11 for differential signals. The frequency range for the SP-analysis is 1 to 10 GHz ...

For fmax = 1GHz the result looks "reasonable" - but when I change fmax to 10 GHz then things get wired and I get already strong reflections at frequencies below 100 MHz.

I'm a little bit puzzled beacuse in my understanding a higher fmax shouldn't change the S-parameter-curve over frequency much - am I wrong ???

I used the lossless models and also the lossy, wide band model - besides "microstrip" I also tried "sublossline" - the only reasonable s-parameter simulation was with fmax = 1GHz - but I have to admit - I don't relay trust it ...

I hope someone can help me out to run transmission line simulations under spectre/cadence as good as with the w-line-model under hspice (where no "fmax" parameter to my nowledge is needed ...

If someone likes to see more of what I have "done" - let me know - and I'll post it ...

Title: Re: mtline with spectre
Post by setb2609 on Aug 26^th, 2007, 8:56am

furtheron it seems to that the value of zeven in the lmg is clearly wrong (1/2 the value) and I fugured out that impedance matching of the line that I simulate varies form the expected 50 ohm impedance by 40% - as impedance matchig takes place at 70 ohm - realy scary ...

Title: Re: mtline with spectre
Post by Andrew Beckett on Aug 27^th, 2007, 10:45pm

Are you providing RLGC matrices to mtline, or using a lumped model?

In recent versions of spectre (MMSIM61, MMSIM62), mtline supports the physical parameters directly on the component and the solver is run from spectre, rather than having to run lmg first.

I think you probably need to post your example, and I'll take a look.

Regards,

Andrew.

Title: Re: mtline with spectre
Post by setb2609 on Aug 28^th, 2007, 7:40am

OK - one point I have to point out is that I'm using 5.10.41 - this is the version the university has because of their agreement with europractice - can't realy say right now, when they are upgrading ...

Are there any known flaws which might be wiped out meanwhile ?

How is posting of examples handled here - what would you like to have ? Simfiles - just the spec ?

RIght now I'm using LMG to provide the subcircuit an d the RLGC in the .dat-file. In my version there is an option in the properties called "use lmg subckt" and a field "LMG subcircuit file" for the link to the subcircuit - in this way I assume spectre gets over the scs-file the RLGC from .dat-file . Was that "clear" enough ;)

Title: Re: mtline with spectre
Post by setb2609 on Aug 28^th, 2007, 7:48am

btw. are there any restriction towards the possible simulation modes with the mtline model - are there any restrictions towards transient simulation ?