FastFieldSolvers Forum
FastFieldSolvers Forum
Home | Profile | Register | Active Topics | Members | Search | FAQ
 All Forums
 FastFieldSolvers
 FasterCap and FastCap2
 Can FasterCap 2D handle this configuration?

Note: You must be registered in order to post a reply.
To register, click here. Registration is FREE!

Screensize:
UserName:
Password:
Antispam question: What do MOONwalk and MOONdance have in common?
Answer:
Format Mode:
Format: BoldItalicizedUnderlineStrikethrough Align LeftCenteredAlign Right Horizontal Rule Insert HyperlinkInsert EmailInsert Image Insert CodeInsert QuoteInsert List
   
Message:

* HTML is OFF
* Forum Code is ON

 
   

T O P I C    R E V I E W
Michele Ancis Posted - Sep 30 2021 : 17:56:07
Hi,

I think it probably can, but just making sure :)

ht*ps://i.postimg.cc/bJvzv3wK/3.png

I am building such geometry by using vertical metal pieces,
in this kind of fashion:

C m_top.txt e1 0 0 +
C m_side_long.txt e2 0 0 +
C m_side_short.txt e3 0 0 +
C m_bottom.txt e4 0 0

<skipping the file/end formality here>

m_top.txt

S -a y_top a y_top

m_side_long.txt

S -a y_top -a (y_top-off1)
S a y_top a (y_top-off1)

...

I hope it is clear enough, a,y_top,off1 and so on are variables specific to the structure/stack.

I do get values from the simulator but they are "off" with respect to "my expectations" (30%)
Of course there can be 1001 reasons for that, but somewhere I've got to start :)

Thanks for helping!
Michele
4   L A T E S T    R E P L I E S    (Newest First)
Michele Ancis Posted - Oct 03 2021 : 21:06:48
Hi Enrico,

the actual file I'm using - btw I always keep everything in one source nowadays, I would find it cumbersome to jump from file to file, since I still make a lot of mistakes when defining geometries - is much more complex because it entails a full 1P8M stack.

I don't want to share it here in order not to risk to incur in any copyright infringement and stuff like that...
This being said, let me try a full description but with dummy variables for the dielectric constants, only the part of stack for Metal1, and also fake heights...


-------------

*** --- Metals --- ***
* Top to bottom

* M1 geometry comes out quite perverse - try to use #absolute# coordinates in the files
* and use here just (0,0) as reference.
C M1_top.txt eps1 0 0 +
C M1_side_long.txt eps2 0 0 +
C M1_side_short.txt eps3 0 0 +
C M1_bottom.txt eps4 0


*** --- Dielectrics --- ***
* Positions copied from above, but file is single for 'slabs'
* and specific for each interface that includes metal (of varying width)
* Bottom to top

D slab.txt eps7 eps6 0 0.410 0 -0.0
D slab.txt eps6 eps5 0 0.445 0 0.41
D M1_int.txt eps5 eps4 0 0.525 0 0.545
D M1_int.txt eps4 eps3 0 0.675 0 0.645
D M1_int.txt eps3 eps2 0 0.770 0 0.675
D slab.txt eps2 eps1 0 0.820 0 0.77

File M1_int.txt

0 dielectric interface M1
*
S diel -15.55 0 -5.04 0
S diel 5.04 0 15.55 0


File slab.txt

0 continuous dielectric interface
*
S diel -15.55 0 15.55 0

End


*** Metal geometries ***

File M1_top.txt

0 M1 conductor - top
* Absolute coordinates due to quirky dielectrics
S M1 -5.04 0.77 5.04 0.77

End

File M1_side_long.txt

0 M1 conductor - side in xxx
* This is a vertical side going from top across the xxx region
* It starts at yyy offset and ends at xxx offset values (from table above)
S M1 -5.04 0.77 -5.04 ddd
S M1 5.04 0.77 5.04 ddd

End

File M1_side_short.txt

0 M1 conductor - side in zzz
* This is a vertical side going from long side end across the zzz region
* It starts at xxx offset and ends at zzz offset values (from table above)
S M1 -5.04 ddd -5.04 sss
S M1 5.04 ddd 5.04 sss

End

File M1_bottom.txt

0 M1 conductor - bottom U-shape
* This U-shaped conductor sits in ILD1b dielectric and goes as deep as it's
* top - thickness y coordinate
S M1 -5.04 sss -5.04 lll
S M1 -5.04 lll 5.04 lll
S M1 5.04 lll 5.04 sss

End

I hope I didn't mess up too much, trying to cover my tracks! ;)

Regards,
Michele
Enrico Posted - Oct 03 2021 : 17:55:44
Yes I'm able to see the picture.

Thanks for the explanation; so the conductor should be specified correctly, however I'm not sure if you specified the dielectric-dielectric interfaces as well, as in your example reference file I only see 'C' statements and no 'D' statements.

You may want to share a full file maybe? You can also use a single file definition if you want (see the embedded online help on how to do that).

Best Regards,
Enrico

Michele Ancis Posted - Oct 03 2021 : 16:59:36
Hi Enrico,

first of all: can you open the image?
This should be the starting point, it is the structure I am trying to represent.
The file describing the geometry is just a framework, so maybe I was too imprecise with that.
Let me try again.

The structure is a metal rod embedded in four different dielectrics.
Therefore, I split the metal conductor (prefix C) into four "files".

1) Top file: just a segment, embedded in the top dielectric layer
2 Long/short side files: two couples of vertical segments, disconnected (they are opposite sides of a rectangle, if you want), each couple embedded in a different dielectric. The reason for "long" and "short" labels should be apparent from the pic.
3) Bottom file: this is a U shape conductor, embedded in the fourth dielectric.

Please refer to previous post for the way I define the side conductors, but it should be clear by now, they are just two segments (prefix S) going vertically for the required length.

So if my rod is 2a wide (x direction), they will look like:

file m_side.txt

S -a 0 a 0
S -a -h a -h

end

(imagining to offset them when used in the C declaration, h is the actual length in y dir of these shapes)

I left out the dielectric definitions in my previous description, as I didn't think it posed any issue. In general, the idea is what I found in your coupled microstrip examples.

Where you have an uninterrupted dielectric interface, you just declare a segment.
Where you have a dielectric interface that touches a metal, you declare a couple of disconnected segments to "work around the metal".

So, in my real file, I have disconnected segments all the way, just a single segment for the top where there's no metal crossing.

The segments you see in my previous post are all describing metal geometries.

Is it clearer now?

Onto your last comment: <<f so you need to model them, as well as segment the conductor in a way that each segment is in contact with a single material type.>>

I interpret that as if you were saying that a dielectric interface cannot encroach a conductor geometry, right? That is why when diel interfaces cross conductors (like it is in my case, but in your microstrip example as well), you declare TWO pieces, before the metal, and after the metal (going from -x to +x, so to speak).

If my understanding is correct, I think I'm doing correctly.

Regards,
Michele
Enrico Posted - Oct 02 2021 : 11:46:38
Hi Michele,

I'm not sure about your structure. What are the horizontal lines? Dielectric interfaces?

If so you need to model them, as well as segment the conductor in a way that each segment is in contact with a single material type.

Best Regards,
Enrico

FastFieldSolvers Forum © 2020 FastFieldSolvers S.R.L. Go To Top Of Page
Powered By: Snitz Forums 2000 Version 3.4.06