AnsweredAssumed Answered

Re: Mesher input does not contain any valid shapes

Question asked by ALEPEREZ

on Sep 12, 2008
Latest reply on Nov 4, 2008 by ALEPEREZ

Like • Show 0 Likes0
Comment • 14

Send your DSN file to the Agilent EEsof support team in your region. Â That'll give you the quickest answer. Â It is most likely a mapping issue plus the geometry probably has too many vertex points in the polygons and can be cleaned up signficantly (especially with ADS2008 Update 2 release which contains a new Layout Preprocessor that does this automatically).

No one else has this question

Outcomes

14 Replies

tkhan Sep 18, 2008 8:24 AM
Mark Correct
Correct Answer
Here's my situation:

Our university is bound by NDAs from IBM and TSMC so there's no possibility of sending DSN files as they contain process parameters (specifically substrates). I have verified this by copying the DSN file to another computer and creating a new project logged on as a different user; the layers are not in the palette but the substrate information is still there. Is there any other file I can send which does not contain information releveant to the design kit?

On top of this, I have encountered several other problems migrating from ADS2005A to ADS2008U2. We have installations under Linux and Solaris 10. On Solaris 10, I simulated a design made in ADS2005A I am skeptical that MomEngine is doing anything at all. Although CPU usage is up to 100%, i ran 'pmap <pid of MomEngine>' (pmap display information about the address space of a process) several minutes apart and each time dumped the contents into a text file. Doing 'diff <txtfile1> <txtfile2>' showed the results to be excatly the same. Additionally, no progress is made in the simulation status window, it just says Layout Processing Started.

Now I made a new, very simple structure under both Solaris and Linux and I'm getting this error when trying to preview the mesh in RF mode:

ERROR :
Invalid or corrupted substrate description!
Please check the substrate definition and regenerate the substrate database.

You mentioned the layout preprocessor - I tried this on a complex design but I get an error (under Linux) saying:

Could not create the layer definition file - using system layer defaults.

Any ideas?
Like • Show 0 Likes0
Actions
ALEPEREZ Sep 18, 2008 9:35 PM
Mark Correct
Correct Answer
The 8U2 code is being changed, so don't use it anymore. It has a problem with the File Browser, but I just mention this as an aside. The 8U1 code does not have the layout preprocessor, yet. In 8U2, you can turn it off if it doesn't work with the settings that are defined.

Regarding NDA based DSN's, Agilent works with these all the time, although it does require getting the foundary's permission. Since we work with all RFIC/MMIC foundaries, I have never heard of them having a problem letting a customer get support from us.

To track down problems, the obvious thing to do is to simplify everything as much as possible - substrate definitions, models, etc. You can use Edit > Component > Flatten to convert component artwork into polygon level artwork and start reducing things from there. You can preprocess and simulate one layer at a time to see where things are falling apart. But without being able to see the details, it is impossible to give a precise reason for the problems you are seeing.
Like • Show 0 Likes0
Actions
tkhan Sep 19, 2008 4:18 AM
Mark Correct
Correct Answer
Thanks for the reply.

The invalid substrate definition is probably due to me choosing the wrong .slm and .lay files when I opened my designs for the first time in 8U2. (Our kit has 5 options for layer/substrate). I am resimulating now with the correct option selected, hopefully this resolves that problem.

"
The 8U2 code is being changed, so don't use it anymore. It has a problem with the File Browser, but I just mention this as an aside. The 8U1 code does not have the layout preprocessor, yet. In 8U2, you can turn it off if it doesn't work with the settings that are defined.
"

Which version is 'safe' to use? We have the base, U1 and U2 installed. When will a hotfix/update be available? We are downloading an update for our IBM kit right now, but it does not have any updates for ADS. We will contact our foundry before sending the DSN file.
Like • Show 0 Likes0
Actions
ALEPEREZ Sep 19, 2008 4:30 AM
Mark Correct
Correct Answer
U2 as released last week has one serious defect in the File > Browser that I haven't been able to reproduce, but the factory wants to be sure. A newer version will be released shortly. Â An email should be sent to notify you and others who downloaded previous versions.
Like • Show 0 Likes0
Actions
tkhan Sep 19, 2008 4:46 AM
Mark Correct
Correct Answer
Ok, but until the new version is released which version should I be using? So far I have not experienced the problem you are describing.
Like • Show 0 Likes0
Actions
ALEPEREZ Sep 19, 2008 4:48 AM
Mark Correct
Correct Answer
You can keep both U1 and U2 - if you want to try the Layout preprocessing on some designs that have lots of curved polygons in it, you need U2. U1 contains the NlogN solver which is very efficient for large problems.
Like • Show 0 Likes0
Actions
MF01 Sep 19, 2008 8:15 AM
Mark Correct
Correct Answer
I would be very careful using the initial U2 release. If you do, be safe and save often!
Like • Show 0 Likes0
Actions
tkhan Oct 25, 2008 6:58 AM
Mark Correct
Correct Answer
We have still not resolved this problem. As per our NDA we can not provide Agilent support with any design files created using our design kit. We are waiting to hear from the foundry as to whether or not we can seek support from Agilent directly or if there is someone in-house who can help us.

A more general question: has the format of the .slm and .lay files changed between 2005A and 2008? Under both Linux and Solaris, the 'invalid substrate description' error comes up with a new design in a new project.

The documentation from our kit says that it has been tested on ADS2006A, and we have asked the foundry if updated ADS files are available.

Thanks
Like • Show 0 Likes0
Actions
FCHASTEL Oct 29, 2008 2:45 AM
Mark Correct
Correct Answer
Hello,

I was trying to solve the next error in the RFDE-Momentum simulator, when I saw the open issue in the forum.

-- ERROR ---------------------------------------------------------------------

Mesher input does not contain any valid shapes. Your shapes might have been ignored because they were drawn as polylines on a strip or slot layer they were on an unmapped layer.

U2 is previously installed in our server and we have installed the â€œhotfix-20081008-rfde-momentumâ€
Like • Show 0 Likes0
Actions
tkhan Nov 1, 2008 3:23 AM
Mark Correct
Correct Answer
I have manually reentered all the process information for substrate and layer definitions and the mesher no longer gives errors related to invalid or corrupted substrate definitions. However I have asked MOSIS about this and they said this solution is also against our NDA, so I can not use the .lay and .slm that I wrote. The advice I received from the foundry is to revert back to ADS2006A because their testing was all done on that version.

I looked at the .slm that I wrote and the one provided by MOSIS and they are identical (except for some information related to GDSII; we do not plan to stream out from ADS so its not important). Also, another student has used the .lay and .slm from the foundry without and issues. So now I think the problem is not with the definitions themselves but something in ADS, particularly at the User Level. I have gone as far as removing my ADS designs and the hpeesof directory and starting over but I still encounter this problem. Is there any procedure for removing all files related to ADS at the user level? One thing different between myself and the student who got the kit to work is that they never used ADS2005A, only ADS2008A. I don't know how much significance that has.
Like • Show 0 Likes0
Actions
ALARCON Nov 4, 2008 1:49 AM
Mark Correct
Correct Answer
The U2 is given also different results for s-parameter calculations respect to 2006A_update2.I have submitted a ticket already on this issue.
=========================
With the new installed version of Ads2008_update2 I found differences in the s parameters of a transmission line. The phase and the amplitude, for instance of the s11 parameter, differ substantially. I cannot explain where these differences are coming from.
Like • Show 0 Likes0
Actions
ALEPEREZ Nov 4, 2008 8:14 AM
Mark Correct
Correct Answer
At the user level, the "hpeesof" subdirectory in $HOME (Unix) or where the projects are stored (Windows) is where all config type info is stored. ADS does permit customization at the SITE level which means that files can be loaded into the executable tree that would modify the behavior. It's good that you found a machine on which the problem does not show up, so I would think that by carefully comparing the install bits, HOME directory, hpeesof dir, ENV vars, you should be able to track this down.
Like • Show 0 Likes0
Actions
ALARCON Nov 4, 2008 6:18 PM
Mark Correct
Correct Answer
My variables at the user level were not changed when I started using 2008U2. The config files are unchanged locally and the results should be the same for the s-parameters calculations. All the other settings are the same and I expect to have the same output. My problem gets more complicated because when extracting the characteristic impedance fron the s-parameters, they are 8% off the values calculated with 2006_update2.
Like • Show 0 Likes0
Actions
ALEPEREZ Nov 4, 2008 6:38 PM
Mark Correct
Correct Answer
My last reply was in reference to "tkhan"'s post. For your ("alarcon") problem, I sent a detailed email reply last night. Hope that clarifies the issue.
Like • Show 0 Likes0
Actions

Re: Mesher input does not contain any valid shapes

Outcomes

Related Content

Recommended Content