ADS schematic vs momentum

P

Philip Newman

Guest
Hi, my colleagues and I are trying to simulate interdigital capacitors using
ADS. We have found our value of capacitance using the-3dB cut-off point of
the S11 characteristics at the frequency of interest. We have also used the
equation given in GD Alley's paper, from which we can calculate the length
of the fingers. The width of the capacitor is limited, as is the length of
the fingers to less that lambda/10 for operations at 10GHz.

The problem we have found is that simulations done with the schematic view
give very different answers to those given by momentum. The design is
exactly the same, but the answers are different. We tested the capacitor
using two dog bone structures coupled together with the capacitor to give a
bandpass filter. however, the momentum sim shows that the passband is
shifted by 2GHz to the left!

We have updated the substrate so that everything is the same. the dog-bones
are the same size, as is the dimensions of the capacitor.

this problem is very annoying and is holding back our work!

If anyone has had similar problems, or know how to fix them, then please let
me know

Cheers

Phil
 
On Thu, 4 Dec 2003 19:01:38 -0000, "Philip Newman"
<nojunkmail@ntlworld.com> wrote:

Hi, my colleagues and I are trying to simulate interdigital capacitors using
ADS. We have found our value of capacitance using the-3dB cut-off point of
the S11 characteristics at the frequency of interest. We have also used the
equation given in GD Alley's paper, from which we can calculate the length
of the fingers. The width of the capacitor is limited, as is the length of
the fingers to less that lambda/10 for operations at 10GHz.

The problem we have found is that simulations done with the schematic view
give very different answers to those given by momentum. The design is
exactly the same, but the answers are different. We tested the capacitor
using two dog bone structures coupled together with the capacitor to give a
bandpass filter. however, the momentum sim shows that the passband is
shifted by 2GHz to the left!

We have updated the substrate so that everything is the same. the dog-bones
are the same size, as is the dimensions of the capacitor.

this problem is very annoying and is holding back our work!

If anyone has had similar problems, or know how to fix them, then please let
me know

Cheers

Phil


I've used Ensemble instead of Momentum, but the Method of Moments
Click to expand...
calculations are similar.

Have you tried increasing the number of triangles in the Momentum
simulations? In Ensemble, I noticed frequency changes in patch antenna
designs when the metal layers were "meshed" tighter. Ansoft's
suggestion was to vary the mesh size until the results began to
stabilize, suggesting that the simulation was nearing an optimum
point. Another thing to check (especially for operation out to 10 GHz)
is the specification of the metal layers (thickness and material).
Conductivity will begin to add to the simulation errors.

Mark
 
thanks, we'll try looking at the difference changing the mesh today. we
have update the substrate from the schematic, so the same one is used, with
the correct conductivity for copper and permittivity of substrate and that,
so no change there, though momentum could make a difference here as it
simulates reality more than the schematic.

any other suggestions would be greatly appreciated

cheers

Phil

"Mark" <spam@spam.com> wrote in message
news:3fd012a8.14583375@news.east.cox.net...
On Thu, 4 Dec 2003 19:01:38 -0000, "Philip Newman"
nojunkmail@ntlworld.com> wrote:

Hi, my colleagues and I are trying to simulate interdigital capacitors
using
ADS. We have found our value of capacitance using the-3dB cut-off point
of
the S11 characteristics at the frequency of interest. We have also used
the
equation given in GD Alley's paper, from which we can calculate the
length
of the fingers. The width of the capacitor is limited, as is the length
of
the fingers to less that lambda/10 for operations at 10GHz.

The problem we have found is that simulations done with the schematic
view
give very different answers to those given by momentum. The design is
exactly the same, but the answers are different. We tested the capacitor
using two dog bone structures coupled together with the capacitor to give
a
bandpass filter. however, the momentum sim shows that the passband is
shifted by 2GHz to the left!

We have updated the substrate so that everything is the same. the
dog-bones
are the same size, as is the dimensions of the capacitor.

this problem is very annoying and is holding back our work!

If anyone has had similar problems, or know how to fix them, then please
let
me know

Cheers

Phil


I've used Ensemble instead of Momentum, but the Method of Moments
calculations are similar.

Have you tried increasing the number of triangles in the Momentum
simulations? In Ensemble, I noticed frequency changes in patch antenna
designs when the metal layers were "meshed" tighter. Ansoft's
suggestion was to vary the mesh size until the results began to
stabilize, suggesting that the simulation was nearing an optimum
point. Another thing to check (especially for operation out to 10 GHz)
is the specification of the metal layers (thickness and material).
Conductivity will begin to add to the simulation errors.

Mark
Click to expand...
 
spam@spam.com (Mark) writes:

On Thu, 4 Dec 2003 19:01:38 -0000, "Philip Newman"
nojunkmail@ntlworld.com> wrote:

Hi, my colleagues and I are trying to simulate interdigital capacitors using
ADS. We have found our value of capacitance using the-3dB cut-off point of
snip
this problem is very annoying and is holding back our work!

If anyone has had similar problems, or know how to fix them, then please let
me know

Cheers

Phil


I've used Ensemble instead of Momentum, but the Method of Moments
calculations are similar.

Have you tried increasing the number of triangles in the Momentum
simulations? In Ensemble, I noticed frequency changes in patch antenna
designs when the metal layers were "meshed" tighter. Ansoft's
suggestion was to vary the mesh size until the results began to
stabilize, suggesting that the simulation was nearing an optimum
point. Another thing to check (especially for operation out to 10 GHz)
is the specification of the metal layers (thickness and material).
Conductivity will begin to add to the simulation errors.

Mark
Click to expand...
Good points.
You could also look for articles by James C. Rautio (the Sonnet guy).
He has written about the accuracy of method of moment programs. He has
also covered topics like what to do when metal thicknes can not be
neglected.

What else? Well, it is always a good idea to double-check the accuracy
of the models used in the schematic-driven simulation. For what values
of the paramters is the model valid etc.

Remember that Momentum (and its ilk) can analyze 'any' geometry but
to get trustworthy results you need more insight than when running an
ordinary circuit simulator. There is quite a number of parameters
to tweek and eventhough the default is OK in most cases it might
not be in your case.

Then of course you have the posibility of building a test structure and
take it to the lab.

Cheers,

--
========================================================================
Martin Schöön <Martin.Schoon@ericsson.com>

"Problems worthy of attack
prove their worth by hitting back"
Piet Hein
========================================================================
 
thanks for the advice, I think we are slowing getting there!!

I hope this works because we are getting quite annoyed with it not working!

cheers

Phil

"Schöön Martin" <Martin.Schoon@dont.spam.ericsson.com> wrote in message
news:s5zekvjzm4j.fsf@yellow320.era-a.ericsson.se...
spam@spam.com (Mark) writes:

On Thu, 4 Dec 2003 19:01:38 -0000, "Philip Newman"
nojunkmail@ntlworld.com> wrote:

Hi, my colleagues and I are trying to simulate interdigital capacitors
using
ADS. We have found our value of capacitance using the-3dB cut-off
point of
snip
this problem is very annoying and is holding back our work!

If anyone has had similar problems, or know how to fix them, then
please let
me know

Cheers

Phil


I've used Ensemble instead of Momentum, but the Method of Moments
calculations are similar.

Have you tried increasing the number of triangles in the Momentum
simulations? In Ensemble, I noticed frequency changes in patch antenna
designs when the metal layers were "meshed" tighter. Ansoft's
suggestion was to vary the mesh size until the results began to
stabilize, suggesting that the simulation was nearing an optimum
point. Another thing to check (especially for operation out to 10 GHz)
is the specification of the metal layers (thickness and material).
Conductivity will begin to add to the simulation errors.

Mark

Good points.
You could also look for articles by James C. Rautio (the Sonnet guy).
He has written about the accuracy of method of moment programs. He has
also covered topics like what to do when metal thicknes can not be
neglected.

What else? Well, it is always a good idea to double-check the accuracy
of the models used in the schematic-driven simulation. For what values
of the paramters is the model valid etc.

Remember that Momentum (and its ilk) can analyze 'any' geometry but
to get trustworthy results you need more insight than when running an
ordinary circuit simulator. There is quite a number of parameters
to tweek and eventhough the default is OK in most cases it might
not be in your case.

Then of course you have the posibility of building a test structure and
take it to the lab.

Cheers,

--
========================================================================
Martin Schöön <Martin.Schoon@ericsson.com

"Problems worthy of attack
prove their worth by hitting back"
Piet Hein
========================================================================
Click to expand...
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top