Learning Forum

[schejarla]

I am trying to find an eigen modes of an air clad fiber. In this process, I am observing that there is limit of HE or EH mode number for particular core dimention. So if I wanted to go for HE/EH mode of order 100 what should be the best way to find in eigen mode solver ? I understand that the maximum mode number can be obtained as we increase the diameter of the core but there are lot of garbage modes also coming.

[Guilin Sun]

Depending on your purpose, you can use Metal BCs to quickly get some lower order of modes, such as the fundamental mode. you can also use PML to get the only propgation modes. However, you may need to signifiantly increase the number of trial mdoes in order to get ride of PML modes (not shown). If yoy believe the mode number can reach 100, and you want all of the modes, you may need to use 200 or more for the trial modes.

If you only need the modes with certain effective index, you can set the range of index, or begin from a certain index value (the result can be beyond this value).

Please try and see if you can get sufficient number of modes.

 

[schejarla]

Thanks for your reply. I have figured out the effective index of the higher order mode, say 100 then I have tried to keep that index range to get this mode but I dont see any of the modes nearby 100 rather it stops at mode number 38 at index value 1.27. It clearly says that for mode 100 we should go along the index value near 1 but still I could not find the desiredd order mode. Please help in this regard.

[Guilin Sun]

I guess you have used PML BCs? if so, since there are many PML modes to be discarded, the actual number of modes is not what expected. You can either try to use Metal BCs but to identify the true modes, or use more numbe of trial modes, say 999.

When metal BCs are used, you may also need to use more number of modes, since there are many non-propagation modes.

[schejarla]

 

Thank you for valauble reply. Yes, I am using PML bundaries as of now. I want only the propagating modes with the desired HE/EH  pattern. But if I use metal BCs I do see non-propagating modes as you mentioned then why to look at them, how to search for only propagating modes of higher order HE. ?

 

[Guilin Sun]

FDE uses a random seeds to solve the eigen values. So even if you input an exact value of the effective index for a certain "higher order" mode, the solve will search the solution around it, usually larger than it to begin with. So when you set PML BCs and give a search range, you will still need to set a large number of trial modes, say 999. The solver finds a lot of PML modes and then throw them away and keep the propagating modes only. Those PML modes are counted so you will need large number of trial modes.

[schejarla]

Thank you Guilin. You are right even if I knew the effective index of that desired higher order mode I need to search for the trial modes around 999.Even in this case also, I am not seeing that desired mode rather I see many garbage modes. Please kindly help.

[Guilin Sun]

Too many modes might be a waste of time. You may need to use smaller index there.

You can also increase the max number of trial modes from Advanced Options:

[schejarla]

Yes, I have chosen smaller index range and found one of the higher order HE mode (radial order one) but the intensity of that mode does not look like a continuous ring instead it looks like a circular chain ring. What was the reason for this ?

As you suggested, if I chose advanced options to search 10000 trail modes, does it give me the desired higher order mode of one radial order with perfect intensity ring ? and will it increase the computational time ?

[Guilin Sun]

Please post your result. I guess it is not the HE mode you expect. It is higher order. So please check the effective index and see if it is expected. You may need to use slightly larger reference index for the search.

As I said previously, there are a lot of PML modes that are not of our interest. The modal list only gives the propagation modes. Whether or not you can find the desired modes depends on the total number of PML modes and the max number of trial modes. If you use sufficiently large trial number you should be able to find that mode, if it exists.

The large number of trial modes definitely needs more computation time.

[schejarla]

You are right, I am using large number of trial modes around 999 within a index range of 1.2 to 1 but still i am not able to find the desired highest order mode. Please look into the below figure for your reference in which desired order mode is indicated right bottom corner.

[Guilin Sun]

Did you modify the max allowed number?

 

You can also reduce the index range, reduce the begining index for search;

and also, more number of modes will need more memory. So I would suggest you to use symmetry BCs. If you know the symmetry property, you can directly setup. If you do not know, you can use the combintaiton of Symmetrical and Anti-Symmetrical. in total 4 sets.

According to my knowlesge, such whispering gallery modes are many, and sometimes some may not be found since it is higherly degerated, and sensitive to mesh (2*pi*radius*neff is very large). You could try FEEM and see if it works better than FDE.

[schejarla]

 

Ok, Sure, Thank you very much for your suggestions and kind support.

If I use symmetry BCs, I will not see the even and odd modes in a sequential manner as I need two modes together with two different sources placed some distance apart, any comments on this ? How to use symmetrical and anti-symmmetrical in total 4 sets ? please suggest.

I agree with you that (2*pi*radius*neff) actually decides the order of the mode number but if i wanted to go for a higher order mode, i may have to increase the radius but in that case I am really worried about the mesh resolution and computational time. Since I am dealing with circular waveguides, mesh resolution is playing a crucial role, otherwise I am not seeing any of the desired modes.

 

[Guilin Sun]

"If I use symmetry BCs, I will not see the even and odd modes in a sequential manner as I need two modes together with two different sources placed some distance apart, any comments on this ? "

I do not see the importance of the sequential manner. The important thing is to find them. For example, you can first find the even mode, and then odd mode, and then no-symmetry mode. Not sure what "two different sources placed some distance apart," means. Do you mean the dipole source? You are using FDE for mode analysis. There is no source for FDE.

"How to use symmetrical and anti-symmmetrical in total 4 sets ? please suggest."

You are simulating 2D cross section, let's say in xy plane. both xmin and ymin can be set "symmetrical" and "anti-Symmetrical". Therefore there are 4 combinations. Hopefully the desired mode has such symmetry property. If not, you will need to use full simulation region, which is slow. Symmetry BCs can significantly save simulation time and memory. Please try.

Please refer this article for more information:https://optics.ansys.com/hc/en-us/articles/360034382694-Symmetric-and-anti-symmetric-BCs-in-FDTD-and-MODE

 

For larger bending redius, please try "FEEM", as it is a finite element method which can use coarser mesh in the central region where there is almost no fields. 

Discrete algorithm needs time and memory to conclude the result. If it is a simple disk, you may find some analytical solutions from published papers.

Please try.

 

 

 

 

How to use symmetrical and anti-symmmetrical in total 4 sets ? please suggest.

[schejarla]

Thank you for your immediate response.

I am not using the dipole sources, I am using mode sources, and I wanted to excite two mode sources at a time with two different frequencies placed a diferent distances inside the waveguide. For this, I am really looking for a higher order mode of HE\EH of index100.

But in this case, I am using 36 um diameter of a cylindrical waveguide and I am searching for round 2000 modes (have used your suggestion here using advanced options) between the index 1.1 to 1. I would rather see only HE-69 mode at 1.002139 index (Please find the attached picture), do I really get HE-100 in this case? Please suggest.

Symmetry BC's are working fine for me, thanks for this. But as I am looking for two consecutive sources, even and odd, if I use symmetry BC's the FDE solver is showing only one mode (either even or odd), so it is critical for me. Do you get this point?

 

[Guilin Sun]

My suggesion is: using symmetry BCs to find the odd/even modes, so you know what they look like, and their characteristics.

then not using symmetry BCs to identify them. This will be faster, and ensure the modes are what you expect.

 

I am not sure if it is the HE-100. You should have some criteria for this mode, right?  I strongly suggest to check some papers regarding to HE modes, in particular the higher order modes, the whispering gallery modes.

 

[schejarla]

OK, Thank you. I will try it out and get back to you.

[schejarla]

Hi Guillin,

As I am searching for many number of modes, it is occupying lot of memory for the entire simulation. Is there any way to save that particular mode and import it as a source in my case ? Please kindly help.

[Guilin Sun]

If you find the desired mode, you can either export it:

or using script:

E1=getresult("mode1","E");

H1=getresult("mode1","H");

EH=E1;

EH.addattribute("H",H1.Hx,H1.Hy,H1.Hz);

matlabsave("EG_mode1.mat",EH);

 

You may need to narrow down the index range to possibly reduce the number of calculated modes

 

[schejarla]

Could you please tell me which expression is used for the "magnitude" as shown in the below picture?

[Guilin Sun]

It is sqrt(abs(Ex)^2+abs(Ey)^2+abs(Ez)^2).

[schejarla]

Thank you very much for your help.

I wanted to know whether there is a way to turn ON and OFF the mode sources in my simulation model ?

[Guilin Sun]

You can disable any object in GUI or by script:

setnamed("obj_name","enabled",0);

 

Please write a new post later for anything not directly related to the topic. Thank you!

[schejarla]

Thank you for your immediate reply, but how can we operate the source, like switching mode ON and OFF ? If we use disable option, it will be completely OFF.

[Guilin Sun]

You want to swtich on and off an object or source DURING the simulation? unfortuately it is not doable, as all the solvers are deterministic.

In some cases you may use material plugin to have a refracgtive index change with time, but it is out of support scope. Please refer this "step index" example:

https://optics.ansys.com/hc/en-us/articles/360034394734

also refer how does it work:

https://optics.ansys.com/hc/en-us/articles/360034915213-Flexible-Material-Plugin-Framework