FastFieldSolvers Forum
FastFieldSolvers Forum
Home | Profile | Register | Active Topics | Members | Search | FAQ
Username:
Password:
Save Password
 All Forums
 FastFieldSolvers
 FasterCap and FastCap2
 FasterCap Warnings/Errors
 New Topic  Reply to Topic
 Printer Friendly
Author Previous Topic Topic Next Topic  

ahsantipu1

Pakistan
2 Posts

Posted - May 29 2013 :  06:30:09  Show Profile  Reply with Quote
I am getting the following warning when I input a .lst file in faster cap and I need some help with it urgently :-

"degenerate quadrilateral panel 'Q' found (co-linear vertexes) corner coordinates are :-
(-3.5e-009,5e-008,-3.5e-009)(-3.5e-009,5e-008,3.5e-009)(2.35e-008,5e-008,3.5e-009)(2.35e-008,5e-008,-3.5e-009)
panel defination found at line 10 of input file "bottom.qui"
skipping the panel"


Somebody help please!!!

ahsantipu

Enrico

412 Posts

Posted - May 29 2013 :  15:40:15  Show Profile  Reply with Quote
Looks like you defined a very small panel, below the geometric tolerance threshold.

You have two ways out:

1. Since pure capacitance scales linearly with the dimensions, you can rescale your structure multiplying all coordinates by 1e9, and then dividing the result by 1e9

2. You can consider the unit of measurement in the permittivity value, so again rescale your structure multiplying all coordinates by 1e9, and divide the relative permittivity constants by 1e9 (remark: including the air permittivity which usually is 1!). So the permittivity has the dimension of F/nm instead of F/m, and you directly get the result without the need of further scaling.

Best Regards,
Enrico

Go to Top of Page

ahsantipu1

Pakistan
2 Posts

Posted - May 31 2013 :  09:02:02  Show Profile  Reply with Quote
Dear Enrico,
Thank you very much for your help so far.
I am running model1.lst (emailed to you already) file in fastercap but i'm getting errors like "non-negative off diagonals" and "matrix not diagonally dominant".
The model is basically a cube with spheres inside. The dielectric constant inside the cube is 3.9. You will get a better idea if you run the file in fastmodel.
Kindly help me out with this.

Regards,
Ahsan Tipu.

ahsantipu
Go to Top of Page

Enrico

412 Posts

Posted - May 31 2013 :  16:01:36  Show Profile  Reply with Quote
You are simulating a conductive box containing conductive two spheres. The box is completely enclosing the spheres. To understand why you are getting this warning, let's connect the sides of the box all together, to form a single conductor (in your original file there are no '+'s):

* this is the first sphere
C final3.qui 3.9 0000000005 0 0

* this is the second sphere
C final3.qui 3.9 0000000015 0 0

* this is the box, composed by six plates, that I connected toghether with the '+' instruction
C plateairinterfacefront.qui 3.9  -0000000003.5 -0000000010 0000000003.5  +
C plateairinterfacefront.qui 3.9  -0000000003.5 -0000000010 -0000000003.5 +
C plateairinterfaceside.qui  3.9 -0000000003.5 -0000000010 -0000000003.5 +
C plateairinterfaceside.qui  3.9 0000000023.5 -0000000010 -0000000003.5 +
C bottom.qui  11.7 -0000000003.5 -0000000010 -0000000003.5 +
C bottom.qui  11.7 -0000000003.5 0000000050 -0000000003.5 

Now if you run the simulation you get:

g1_sphere  9.28601e-010 -6.87077e-012 -9.5492e-010 
g2_sphere  -7.12464e-012 9.27649e-010 -9.51469e-010 
g3_1  -9.23075e-010 -9.24135e-010 1.23681e-008 
Warning: capacitance matrix is not diagonally dominant due to row 1
Warning: capacitance matrix is not diagonally dominant due to row 2

If you check the 'misbehaving' values, they are related to the total capacitance of each sphere with respect to the capacitance of the sphere to the box. This can be expected since, being the sphere(s) fully encapsulated in the box, the two values should be exactly balanced (no field lines to infinity: they are all closed on the box internal sides), but this is very difficult to match numerically.
So while in principle you should have C11 = -(C12 + C13) for row 1 in this situation, you get a slightly unbalanced value.

Now we can argue that the value has a non-negligible unbalancing, even for high discretization and low compression of the matrix (that should lead to higher accuracy).
This is due to a problem in the input file: you can notice that the two bottom panels define a relative permittivity of 11.7 instead of 3.9 like the other part of the structure, but there are no dielectric interfaces.
So this is artificially unbalancing the result.
Correcting to

C bottom.qui  3.9 -0000000003.5 -0000000010 -0000000003.5 +
C bottom.qui  3.9 -0000000003.5 0000000050 -0000000003.5 

you get

g1_sphere  9.28601e-010 -6.87077e-012 -9.22867e-010 
g2_sphere  -7.12464e-012 9.27649e-010 -9.23392e-010 
g3_1  -9.23075e-010 -9.24135e-010 9.87059e-009

that still has a diagonal dominance warning but as you can see the balance of C11 with C12+C13 is very close.

Best Regards,
Enrico

Go to Top of Page

phillipus

Germany
2 Posts

Posted - Jun 25 2018 :  09:05:13  Show Profile  Reply with Quote
Hello everyone,

i got the same warning, but in my case your two solutions donīt work.

I am not sure whether i did everything right. So i got a question to the second way.

Can you please help me to understand the marked lines?
What is meant by <xtran>... <xref>?

PS: If you can't see my image, how do i insert it right?
Go to Top of Page

Enrico

412 Posts

Posted - Jun 25 2018 :  10:26:29  Show Profile  Reply with Quote
Hi Phillipuus,

inserting images in the Forum is not supported due to spam / virus prevention. You can send the image to me (see the 'contacts' page) and I will insert it for you, but if you want to show the log, you can simply copy & paste it into the text box.

Best Regards,
Enrico

Go to Top of Page

phillipus

Germany
2 Posts

Posted - Jun 25 2018 :  11:20:07  Show Profile  Reply with Quote
Thank you Enrico!

* <comment>
G <group name>
C <file> <outperm> <xtran> <ytran> <ztran> [+]
D <file> <outperm> <inperm> <xtran> <ytran> <ztran> <xref> <yref> <zref> [-]
B <file> <outperm> <inperm> <xtran> <ytran> <ztran> <xref> <yref> <zref> [-] [+]

The underlinded lines are the ones i don't understand. That's what i meant before.

I E-Mailed you my Code. So maybe you can find my fault.
Go to Top of Page

Enrico

412 Posts

Posted - Jun 25 2018 :  11:49:20  Show Profile  Reply with Quote
Please look at the FasterCap embedded online help, it describes the G, C, D statements (they are compatible with FastCap2); otherwise you can refer to the User's guide for FastCap, again accessible from FastCap2's embedded online help.
Here below you are an extract from FasterCap help page.

I will check your email, but be sure to read the documentation first, otherwise I can hardly help you.

Best Regards,
Enrico


Syntax: D <file> <outperm> <inperm> <xoffset> <yoffset> <zoffset> <xref> <yref> <zref> [-]


The 'D' element at the beginning of a line defines a dielectric. The geometry of the dielectric surface is further specified in the file <file>. The dielectric surface is intended as the interface between two regions with relative permettivities <outperm> and <inperm>. The relative permittivities <outperm> and <inperm> can be complex valued, in the format eRe-jeIm, where eRe is the real part of the complex permittivity value, and eIm is the imaginary part.
The dielectric can be translated with respect to the coordinates defined in <file> by an offset (xoffset, yoffset, zoffset), thus allowing to reuse the same geometric definitions multiple times.
The reference point (xref, yref, zref) and the optional - argument are used to specify which side of the dielectric interface has which permittivity. More specifically, the reference point is assumed to lie on the <outperm> side of all the panels in <file>. The optional - argument indicates that (xref, yref, zref) instead lies on the <inperm> side. The reference point is not translated, i.e. the offset only applies to the elements specified in <file>.

Remark: it is the user's responsibility to make sure that the reference point is on the same side of the dielectric interface for all panels. That is, each panel is evaluated as stand-alone with respect to the reference point, to define its <outperm> and <inperm> sides. There is no concept of a external and internal side of a surface specified by a group of panels, even if specified in the same file with the same conductor names, since FasterCap maintains no topological information.

For complex shaped surfaces, FasterCap supports also the option to specify the reference point panel by panel, see 3D Triangular panel definitions ('T' statement) and 3D Quadrilateral panel definitions ('Q' statement) for details about the syntax. You can also mix the two type of panel definitions, i.e. panels without and panels with a per-panel reference point specification. In this case, panels without a reference point specified per panel will use the reference point specified for the whole dielectric interface in the 'D' statement.

Example 1:

D sphere.txt  1.0 2.0  0.0 0.0 0.0  0.0 0.0 0.0  -


This input file fragments specifies a dielectric interface whose geometry is defined in the file sphere.txt. The interior of the sphere is filled with a material with relative permittivity equal 2.0, since the reference point is centered on the origin and the optional - argument is specified; while the dielectric medium ouside the sphere has a relative permittivity equal to 1.0.

Example 2:

D sphere.txt  1.0 3.0-j0.02  0.0 0.0 0.0  0.0 0.0 0.0  -


This input file fragments specifies a dielectric interface whose geometry is defined in the file sphere.txt. The interface is between air (relative permittivity equal to 1.0) and a lossy dielectric with complex relative permettivity equal to 3.0-j0.02 (e.g. with real part equal to 3.0 and imaginary part equal to 0.02).


Go to Top of Page
  Previous Topic Topic Next Topic  
 New Topic  Reply to Topic
 Printer Friendly
Jump To:
FastFieldSolvers Forum © 2018 FastFieldSolvers S.R.L. Go To Top Of Page
Powered By: Snitz Forums 2000 Version 3.4.06