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.
a Voltage Divider simulation
Posted Feb 15, 2011, 11:13 p.m. EST 3 Replies
Please login with a confirmed email address before reporting spam
Hiya, My name is Leonardo Camargo and I’m a Electrical Engineering student. Currently I´m using de 4.1 version of Comsol to simulate a band-width Voltage Divider and I must to calculate their STRAY CAPACITANCES. The Divider must to reduce the Amplitude of a Lightning Impulse to measure it on a oscilloscope.
I already made the geometry completely and added the rest of items to make the wave shape, Even I prove the circuit in other file that works so well but I cannot connect it to the voltage Divider Model.
I have attached both Files:
-The first one is the simulation of the lightning Impulse
-The second one is the file where I´ve been making the geometry for the voltage divider
PD:I forgot to write down that there is other problem because the program cannot mesh the resistor, Even if I choose a finer mesh
Please Help me out!!!!!!!!!!!!!!!!!!!!!!!!!!!
I already made the geometry completely and added the rest of items to make the wave shape, Even I prove the circuit in other file that works so well but I cannot connect it to the voltage Divider Model.
I have attached both Files:
-The first one is the simulation of the lightning Impulse
-The second one is the file where I´ve been making the geometry for the voltage divider
PD:I forgot to write down that there is other problem because the program cannot mesh the resistor, Even if I choose a finer mesh
Please Help me out!!!!!!!!!!!!!!!!!!!!!!!!!!!
3 Replies Last Post Feb 16, 2011, 10:06 a.m. EST
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
the lazy reply I would say, if you look at your pulse in a lin-log scale, you could reproduce it rather simply by some analytical function to feed into the 2nd model
Next approach, you fill a "table" from the first simulation and use that as the V function over time
3rd you couple both by using "mod1.cir.C2_v" as input, and check that you separate the combine solving into two segregated steps
Concerning your model a general comment is that sharp edges/corners are singularities in ACDC and induce breakdowns in reality, it would be better to use "tubes" than square section wires
For the meshing, COMSOL has often some issues with looped geometries, you have typically a topology that contains multiple loops. I would say
1) check carefully that your boundaries are really matched, I believe your resistor wires are not really touching the torus, they do not share a common boundary (look at boundary 140 & 319 for example. then you should work in Geometry- Finish - Union mode as you need the continuity BC for your model (this will flag the error for you, telling you that the geometry is not "clean"). To improve it you need to have the volumes overlap and then perform boolean operations to cut the extra material and to avoid extra internal domains
2) if you still have meshing issues, cut your rings with internal boundaries such that you can mesh one quarter at the time
Finally, your system is so symmetric that you could consider one one quadrant.
By the way, for the COMSOL DEVELOPERS:
this is again a typical example where applying some affine symmetry or mirror operations on the geometry and linked them to the mesh, we users would gain a lot of time and have less errors.
Final comment: you should imperatively update your COMSOL version, there are a few errors in the formulas for the 4.1.0.88, there are two new patches on the main COMSOL site, see with your sysadmin or your local COMSOL rep
You have also something wrong in the materials data (empty entry) for your Voltage divider Model file, not sure what ? And you need to "box" the model with an air surrounding I believe"
Nice project by the way
--
Good luck
Ivar
the lazy reply I would say, if you look at your pulse in a lin-log scale, you could reproduce it rather simply by some analytical function to feed into the 2nd model
Next approach, you fill a "table" from the first simulation and use that as the V function over time
3rd you couple both by using "mod1.cir.C2_v" as input, and check that you separate the combine solving into two segregated steps
Concerning your model a general comment is that sharp edges/corners are singularities in ACDC and induce breakdowns in reality, it would be better to use "tubes" than square section wires
For the meshing, COMSOL has often some issues with looped geometries, you have typically a topology that contains multiple loops. I would say
1) check carefully that your boundaries are really matched, I believe your resistor wires are not really touching the torus, they do not share a common boundary (look at boundary 140 & 319 for example. then you should work in Geometry- Finish - Union mode as you need the continuity BC for your model (this will flag the error for you, telling you that the geometry is not "clean"). To improve it you need to have the volumes overlap and then perform boolean operations to cut the extra material and to avoid extra internal domains
2) if you still have meshing issues, cut your rings with internal boundaries such that you can mesh one quarter at the time
Finally, your system is so symmetric that you could consider one one quadrant.
By the way, for the COMSOL DEVELOPERS:
this is again a typical example where applying some affine symmetry or mirror operations on the geometry and linked them to the mesh, we users would gain a lot of time and have less errors.
Final comment: you should imperatively update your COMSOL version, there are a few errors in the formulas for the 4.1.0.88, there are two new patches on the main COMSOL site, see with your sysadmin or your local COMSOL rep
You have also something wrong in the materials data (empty entry) for your Voltage divider Model file, not sure what ? And you need to "box" the model with an air surrounding I believe"
Nice project by the way
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
about the cuadrant , You mean that I can simulate just the 25% of the model and I'll get the same results like I made all the geometry or I missunderstood
Due some problems with the boolean operations , I couldn´t revolve the wire resistors beacuse they were overlapping and Comsol didn`t let to split or to cut the model, so i must continute working with that but If You say that simulating just the 25% of the model is possible, everything would be easier
thaxs for your quick answers
Due some problems with the boolean operations , I couldn´t revolve the wire resistors beacuse they were overlapping and Comsol didn`t let to split or to cut the model, so i must continute working with that but If You say that simulating just the 25% of the model is possible, everything would be easier
thaxs for your quick answers
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
Yes in your case if also all the boundary conditions are symmetric you can use inly one quarter (or even 1/8 if you allow an antisymmetry, but that means applying +/- V voltage and not +V to GND)
Only thing place your resistors in <1,1,0> direction and apply symmetry conditions on <1,0,0> and <0,1,0>
--
Good luck
Ivar
Yes in your case if also all the boundary conditions are symmetric you can use inly one quarter (or even 1/8 if you allow an antisymmetry, but that means applying +/- V voltage and not +V to GND)
Only thing place your resistors in <1,1,0> direction and apply symmetry conditions on <1,0,0> and <0,1,0>
--
Good luck
Ivar