Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

rf mems switch design (comsol 4.2a)

Amey Revandkar
HeLLO,
i'm working on designing a switch that should work on rf range. I'm able to draw 2d structure in work plane but still not able to fix one side of the cantilver beam. i hve following questions,

1) When i drawn a random structure i was able to get domains and boundaries, but when i restricted dimensions at micron scale, i didnt found domains, only i was able to get 11 boundaries, plz help me in knowing how to fix one side of the beam on anchor...

2) moreover when i chose xy plane in workplane how to add dimension of device in z plane? i had worked on quick plane section in workplane but i'm fiding extremely difficult to draw structure at micron scale. i'm working on coplanar waveguide .

kindly help


22 Replies Last Post Jan 17, 2017, 12:20 PM EST
Posted: 5 years ago Apr 14, 2013, 5:56 AM EDT
Hi

the CAD feature of COMSOL are not as advanced as those you can find in a full blow 3D CAD software (not the same purpouse). For elevation from a plane (extrusion) you can define only one height per node (note I havent really checked if I can add an array, as for copy ... but anyhow it extracts all objects to that/those heights

And when you have really very small scales, check the units and the relative resolution features (also of the graphics rendering as sometimes the geometry is OK, but the graphics rendering is poor, this happens seldom or me)

--
Good luck
Ivar
Hi the CAD feature of COMSOL are not as advanced as those you can find in a full blow 3D CAD software (not the same purpouse). For elevation from a plane (extrusion) you can define only one height per node (note I havent really checked if I can add an array, as for copy ... but anyhow it extracts all objects to that/those heights And when you have really very small scales, check the units and the relative resolution features (also of the graphics rendering as sometimes the geometry is OK, but the graphics rendering is poor, this happens seldom or me) -- Good luck Ivar

Amey Revandkar
Posted: 5 years ago Apr 15, 2013, 4:10 AM EDT
Thanks sir,
i'm finding it out extremely difficult to simulate the structure ..please can u suggest any other alternative to get the structure drawn at micron scale. .
Thanks sir, i'm finding it out extremely difficult to simulate the structure ..please can u suggest any other alternative to get the structure drawn at micron scale. .

Posted: 5 years ago Apr 15, 2013, 4:52 AM EDT
Hi
Use the geometry scale set in microns, and also make a new Definition View2, go to the sub node "Camera" and turn off "Preserve Scale ratio"

Then you can switch easily between view 1 and view 2

If you have a large bulk block you can hide it in view 2 by adding a sub-node Hide "Entities" and select the large "thick" structure(s)

This often helps for MEMS modelling ;)
--
Good luck
Ivar
Hi Use the geometry scale set in microns, and also make a new Definition View2, go to the sub node "Camera" and turn off "Preserve Scale ratio" Then you can switch easily between view 1 and view 2 If you have a large bulk block you can hide it in view 2 by adding a sub-node Hide "Entities" and select the large "thick" structure(s) This often helps for MEMS modelling ;) -- Good luck Ivar

Posted: 5 years ago Apr 15, 2013, 9:29 AM EDT
Hallo, i'm new using the comsol software.anyone can help me how to modeling the shunt capacitive mems switch?
Hallo, i'm new using the comsol software.anyone can help me how to modeling the shunt capacitive mems switch?

Amey Revandkar
Posted: 4 years ago Apr 18, 2013, 3:13 AM EDT
Thanks sir,
I did simulate my structure on cantilever and i obtained following results for study type eigen frequency in solid mechanics model:

in 3D plot group first 4 eigenfrequencies,

1. -93983.02i

2. -93983.0354i

3. -3.128e5i(1)

4. -3.12897e5i(2)

i dont know to interpret these reults.
morover in solver configuration ,in dependant variables subnode, in initial variables solved for under 3subnode Eigenfrequencies
i obtained 4 eigenfrequencies but only one eigenmode was shown..r these results authentic?
kindly help,
regards.
Thanks sir, I did simulate my structure on cantilever and i obtained following results for study type eigen frequency in solid mechanics model: in 3D plot group first 4 eigenfrequencies, 1. -93983.02i 2. -93983.0354i 3. -3.128e5i(1) 4. -3.12897e5i(2) i dont know to interpret these reults. morover in solver configuration ,in dependant variables subnode, in initial variables solved for under 3subnode Eigenfrequencies i obtained 4 eigenfrequencies but only one eigenmode was shown..r these results authentic? kindly help, regards.

Posted: 4 years ago Apr 18, 2013, 5:06 AM EDT
Hi

are you sure those are the "eigenfrequencies, and not the eigenvalues ?

Check the solver name, and your doc, as there are both ;)

--
Good luck
Ivar
Hi are you sure those are the "eigenfrequencies, and not the eigenvalues ? Check the solver name, and your doc, as there are both ;) -- Good luck Ivar

Amey Revandkar
Posted: 4 years ago Apr 22, 2013, 1:59 PM EDT
hi,
This time i got 4 eigenfrequencies without imaginary part. i beleive it must be correct. i hve following querries:

1. i hve simulated the cantileverbeam on the substrate and found eigenfrequencies, length of the beam is 70 micron,
thickness is 3 micron,but the width is in z direction and i want to measurethat. plz let me know how it can be done?

2. what will be the unit for eigenfrequency?

i'm attaching the file hereby
Regards
hi, This time i got 4 eigenfrequencies without imaginary part. i beleive it must be correct. i hve following querries: 1. i hve simulated the cantileverbeam on the substrate and found eigenfrequencies, length of the beam is 70 micron, thickness is 3 micron,but the width is in z direction and i want to measurethat. plz let me know how it can be done? 2. what will be the unit for eigenfrequency? i'm attaching the file hereby Regards


Posted: 4 years ago Apr 23, 2013, 5:11 AM EDT
Hi

you solver says eigenvalue ? I suspect something has been messed up in the solver sequence, try making a new study with an Eigenfrequency solver, this could explain the complex results

--
Good luck
Ivar
Hi you solver says eigenvalue ? I suspect something has been messed up in the solver sequence, try making a new study with an Eigenfrequency solver, this could explain the complex results -- Good luck Ivar

Amey Revandkar
Posted: 4 years ago Apr 23, 2013, 3:33 PM EDT
hi,
i'll check the solver sequence, meanwhile i have calculated the spring constant of the cantilever beam which is0.0036.
the dimensions are,
length = 70 micron
width= 1 micron
thickness =0.3micron
is the value correct?

also can i know what could be the simplest formulae to find first eigenfrequency/free running frequency for above dimensions?
regards
hi, i'll check the solver sequence, meanwhile i have calculated the spring constant of the cantilever beam which is0.0036. the dimensions are, length = 70 micron width= 1 micron thickness =0.3micron is the value correct? also can i know what could be the simplest formulae to find first eigenfrequency/free running frequency for above dimensions? regards

Posted: 4 years ago Apr 24, 2013, 5:35 AM EDT
Hi

the physics formulas do not change with scale at most the cuplings due to the change in volume to area ratio. So take any valid canteliever beam formula and enter your dimensions and material properties E, nu, rho

--
Good luck
Ivar
Hi the physics formulas do not change with scale at most the cuplings due to the change in volume to area ratio. So take any valid canteliever beam formula and enter your dimensions and material properties E, nu, rho -- Good luck Ivar

Amey Revandkar
Posted: 4 years ago Apr 30, 2013, 2:27 PM EDT
Hi,
May i know if the results of my simulation for cantilever beam are correct or not for the given data:
L= 90 micron, W=40 micron, thickness t= 0.3 micron for aluminium cantilever with E= 70 GPA, poissons ratio=0.35,Density=2700 kg/m3,

results:

spring constant= 0.025

pull down voltage= 11.12 V

eigen frequency1= 31486.5373 ef2= 1.48e5, ef3= 1.9655e5, ef4= 4.7977e5

regards

Hi, May i know if the results of my simulation for cantilever beam are correct or not for the given data: L= 90 micron, W=40 micron, thickness t= 0.3 micron for aluminium cantilever with E= 70 GPA, poissons ratio=0.35,Density=2700 kg/m3, results: spring constant= 0.025 pull down voltage= 11.12 V eigen frequency1= 31486.5373 ef2= 1.48e5, ef3= 1.9655e5, ef4= 4.7977e5 regards

Posted: 4 years ago Apr 30, 2013, 2:44 PM EDT
Hi

if you check on wikipedia the single sided fixed-free beam deflection tells you F/disp = 3E*I/L^3 where F is the force applied, disp the tip displacement, E the young modulus, L the beam length and I the Area moment of inertia typically some thing like "tickness^3*width/12" for a rectangular section beam deflecting along an axis paralle to the width

Another rule of thumb I often use is the maximum deflection "disp" of the first fundamental mode frequenacy "f" often follows

2*pi*f = sqrt(k/m) = sqrt(g/disp)

where g=9.81[m/s^2], m the mass bt often one take 2/3 of the total beam mas in a fixed-free case

But you fins also more precise formulas on Wiki i.e Euler - Bernoulli beam or in the COMSOL doc theory section

--
Good luck
Ivar
Hi if you check on wikipedia the single sided fixed-free beam deflection tells you F/disp = 3E*I/L^3 where F is the force applied, disp the tip displacement, E the young modulus, L the beam length and I the Area moment of inertia typically some thing like "tickness^3*width/12" for a rectangular section beam deflecting along an axis paralle to the width Another rule of thumb I often use is the maximum deflection "disp" of the first fundamental mode frequenacy "f" often follows 2*pi*f = sqrt(k/m) = sqrt(g/disp) where g=9.81[m/s^2], m the mass bt often one take 2/3 of the total beam mas in a fixed-free case But you fins also more precise formulas on Wiki i.e Euler - Bernoulli beam or in the COMSOL doc theory section -- Good luck Ivar

Amey Revandkar
Posted: 4 years ago May 1, 2013, 8:24 AM EDT
Thanks sir,
I have used the same formulas u stated above and i got the the mentioned results. Now i'm performing Electromechanics modelling to check displacement for the actuation voltage i obtained.

As i'm using comsol 4.2a i'm following the instructions given under resonator study model pdf. with necessary instructions relevant to my model. i request you to kindly check the mph file attached hereby and help me in modelling as i have just 7 domains in my structure.

regards
Thanks sir, I have used the same formulas u stated above and i got the the mentioned results. Now i'm performing Electromechanics modelling to check displacement for the actuation voltage i obtained. As i'm using comsol 4.2a i'm following the instructions given under resonator study model pdf. with necessary instructions relevant to my model. i request you to kindly check the mph file attached hereby and help me in modelling as i have just 7 domains in my structure. regards


Posted: 4 years ago May 1, 2013, 9:21 AM EDT
Hi

I see you have a challenging model there.
For me my main comments are that to get a electrostatic field to develop, you must include some "air" or vacuum between the electrodes, and also above to allow the field to loop around

As you model is basically an extruded 2D model it would be far simpler and faster to work in 2D and define the appropriate depth d (along Z) for the absolute values calculations. I can only recommend to start "simple" in 2D then go to 3D.

Even have you started in 2D structural alone,m checked your eigenfrequencies, the material selections etc, often useful. But if you are well used to COMSOL then jumping straight into EMI is possible.

Finally you could reduce the solid AlO3 block to have less elements to mesh, and to allow for at least 3-5 mesh elements in the thickness of your thin beams to correctly resolve what is happening in the beam/plates you have

--
Good luck
Ivar
Hi I see you have a challenging model there. For me my main comments are that to get a electrostatic field to develop, you must include some "air" or vacuum between the electrodes, and also above to allow the field to loop around As you model is basically an extruded 2D model it would be far simpler and faster to work in 2D and define the appropriate depth d (along Z) for the absolute values calculations. I can only recommend to start "simple" in 2D then go to 3D. Even have you started in 2D structural alone,m checked your eigenfrequencies, the material selections etc, often useful. But if you are well used to COMSOL then jumping straight into EMI is possible. Finally you could reduce the solid AlO3 block to have less elements to mesh, and to allow for at least 3-5 mesh elements in the thickness of your thin beams to correctly resolve what is happening in the beam/plates you have -- Good luck Ivar

Amey Revandkar
Posted: 4 years ago May 2, 2013, 6:45 AM EDT
Hii,
Thanks indeed for the previous reply. I'm still trying to get the EMI perfect. meanwhile i'm trying to find the biased modes .i.e. i'm trying to apply 11 V dc bias voltage and trying to get the biased mode shapes.
Will u please go through the file i'm attaching hereby that has the unbiased modes and help me in getting biased shapes for 11 V.

Regards
Hii, Thanks indeed for the previous reply. I'm still trying to get the EMI perfect. meanwhile i'm trying to find the biased modes .i.e. i'm trying to apply 11 V dc bias voltage and trying to get the biased mode shapes. Will u please go through the file i'm attaching hereby that has the unbiased modes and help me in getting biased shapes for 11 V. Regards

Amey Revandkar
Posted: 4 years ago May 2, 2013, 6:47 AM EDT
here is the file attached
here is the file attached


Posted: 4 years ago May 3, 2013, 1:31 AM EDT
Hi

when you look at your results you see that basically what is of interest for the eigenfrequency is only domain 5 or perhaps 3+5

if you domain 6 is a counter electrode you could add some boundary load on the beam domain 5 by estimating the force you would get from your voltage, the distance the related areas, and then do a pre-stresssed eigenfrequency analysis (stationary + eigenfrequency linearised from the stationary result, compare to a linear buckling analysis to see how the solvers are combined)

If you want to add ES and ES forces you need to add the surrounding air/vacuum above and below your electrode. This can also be done in EMI

I have only latest version so I cannot make any "older version models" but I'm sure you will get there easily

--
Good luck
Ivar
Hi when you look at your results you see that basically what is of interest for the eigenfrequency is only domain 5 or perhaps 3+5 if you domain 6 is a counter electrode you could add some boundary load on the beam domain 5 by estimating the force you would get from your voltage, the distance the related areas, and then do a pre-stresssed eigenfrequency analysis (stationary + eigenfrequency linearised from the stationary result, compare to a linear buckling analysis to see how the solvers are combined) If you want to add ES and ES forces you need to add the surrounding air/vacuum above and below your electrode. This can also be done in EMI I have only latest version so I cannot make any "older version models" but I'm sure you will get there easily -- Good luck Ivar

Sarath Gopalakrishnan
Posted: 4 years ago Jul 4, 2013, 2:26 AM EDT

Hi,
May i know if the results of my simulation for cantilever beam are correct or not for the given data:
L= 90 micron, W=40 micron, thickness t= 0.3 micron for aluminium cantilever with E= 70 GPA, poissons ratio=0.35,Density=2700 kg/m3,

results:

spring constant= 0.025

pull down voltage= 11.12 V

eigen frequency1= 31486.5373 ef2= 1.48e5, ef3= 1.9655e5, ef4= 4.7977e5

regards


Hi, I have been trying to find out the pull in voltage of a very basic cantilever based MEMS switch. I referred to the tutorial given in MEMS module for resonator pull in voltage calculation and tried to simulate a switch. However, I keep getting errors indication lack of convergence and NaN/Inf not found and so on. Were you successful in finding out the pull in voltage? I would appreciate it if you help me by giving some tips on this issue.
[QUOTE] Hi, May i know if the results of my simulation for cantilever beam are correct or not for the given data: L= 90 micron, W=40 micron, thickness t= 0.3 micron for aluminium cantilever with E= 70 GPA, poissons ratio=0.35,Density=2700 kg/m3, results: spring constant= 0.025 pull down voltage= 11.12 V eigen frequency1= 31486.5373 ef2= 1.48e5, ef3= 1.9655e5, ef4= 4.7977e5 regards [/QUOTE] Hi, I have been trying to find out the pull in voltage of a very basic cantilever based MEMS switch. I referred to the tutorial given in MEMS module for resonator pull in voltage calculation and tried to simulate a switch. However, I keep getting errors indication lack of convergence and NaN/Inf not found and so on. Were you successful in finding out the pull in voltage? I would appreciate it if you help me by giving some tips on this issue.

Amey Revandkar
Posted: 4 years ago Jul 18, 2013, 6:24 AM EDT
Sir,
I'm getting spring constant for my cantilever beam as low as 0.025N/m . so just want to know wheather the spring constant value can be as low as this. The dimensions for my beam are l=90micron, W=40micron, t=0.3micron,E=70Gpa for aluminium.
Sir, I'm getting spring constant for my cantilever beam as low as 0.025N/m . so just want to know wheather the spring constant value can be as low as this. The dimensions for my beam are l=90micron, W=40micron, t=0.3micron,E=70Gpa for aluminium.

Amey Revandkar
Posted: 4 years ago Jul 24, 2013, 7:47 AM EDT
Dear sir,
I'm doing Stationary Analysis for checking the displace ment of cantilever beam(RF Switch) . Ihave almost reachead the end of the modelling , but getting error like,'' Failed to evaluate the expression''. I request you to therefore help me out in completing the analysis. I'm attaching the file hereby,
Regards
Dear sir, I'm doing Stationary Analysis for checking the displace ment of cantilever beam(RF Switch) . Ihave almost reachead the end of the modelling , but getting error like,'' Failed to evaluate the expression''. I request you to therefore help me out in completing the analysis. I'm attaching the file hereby, Regards


Posted: 4 years ago Sep 18, 2013, 7:18 AM EDT
Hi,Ivar
Now I am trying to use emi physics to make a MEMS capacitor with vacuum sealed gap. The top plate of capacitor is thin as compared with the diameter. So the top plate(membrane) will be deflected to the bottom plate due to the surrounding air pressure. I have tried a way by adding material with relative permitivity of 1 in the gap. But the membrane did not deflected more as compared with the air inside the gap. Is there any wrong with this method?
Thank you very much!



Hi

when you look at your results you see that basically what is of interest for the eigenfrequency is only domain 5 or perhaps 3+5

if you domain 6 is a counter electrode you could add some boundary load on the beam domain 5 by estimating the force you would get from your voltage, the distance the related areas, and then do a pre-stresssed eigenfrequency analysis (stationary + eigenfrequency linearised from the stationary result, compare to a linear buckling analysis to see how the solvers are combined)

If you want to add ES and ES forces you need to add the surrounding air/vacuum above and below your electrode. This can also be done in EMI

I have only latest version so I cannot make any "older version models" but I'm sure you will get there easily

--
Good luck
Ivar

Hi,Ivar Now I am trying to use emi physics to make a MEMS capacitor with vacuum sealed gap. The top plate of capacitor is thin as compared with the diameter. So the top plate(membrane) will be deflected to the bottom plate due to the surrounding air pressure. I have tried a way by adding material with relative permitivity of 1 in the gap. But the membrane did not deflected more as compared with the air inside the gap. Is there any wrong with this method? Thank you very much! [QUOTE] Hi when you look at your results you see that basically what is of interest for the eigenfrequency is only domain 5 or perhaps 3+5 if you domain 6 is a counter electrode you could add some boundary load on the beam domain 5 by estimating the force you would get from your voltage, the distance the related areas, and then do a pre-stresssed eigenfrequency analysis (stationary + eigenfrequency linearised from the stationary result, compare to a linear buckling analysis to see how the solvers are combined) If you want to add ES and ES forces you need to add the surrounding air/vacuum above and below your electrode. This can also be done in EMI I have only latest version so I cannot make any "older version models" but I'm sure you will get there easily -- Good luck Ivar [/QUOTE]

Sai Pravallika Sakhamuri
Posted: 9 months ago Jan 17, 2017, 12:20 PM EST
sir this is the error iam suffering with
Undefined variable.
- Variable: comp1.emi.disp
- Geometry: geom1
- Boundary: 126
Failed to evaluate expression.
- Expression: 1000000*comp1.emi.disp
- Plot: surf1 (Surface

can u please solve it if possible
sir this is the error iam suffering with Undefined variable. - Variable: comp1.emi.disp - Geometry: geom1 - Boundary: 126 Failed to evaluate expression. - Expression: 1000000*comp1.emi.disp - Plot: surf1 (Surface can u please solve it if possible

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.