Tip:
Highlight text to annotate it
X
I’ll turn that back on. So, my first job is to define the ports here. So let me do
that. I’ll add a new waveguide port. The face I want is this one. Select that… I
can leave the defaults on the Properties Tab. I’ll give it a name – say wp1. For the
Waveguide Port Definition I want a 1 W Modal Power Feed. This picks up this component definition
here. The 1 W Model Power Feed is a pre-defined component. I’m going to define my impedance
using the Power/Voltage method. And as you know, voltage is path dependant so I need
to specify an impedance line that’s going to be used to measure the voltage along. Two
end points… With microstrip you want to go for the maximum voltage center-to-center.
And that defines my impedance line for the impedance calculation. I’ll do the same
for port 2. We’ll skip ahead, and I’ll add port 2.
So, adding port 2 is identical. I skipped ahead for you. You’ll notice that the ports
still have a caution on them. It’s saying ‘The waveguide is not on the simulation
bounding box.’ You have to have ports on the outer surfaces of the simulation because
you can’t simulate fields beyond the port obviously. So I’m going to fix that by using
my FEM Padding. I’ll just get rid of the free space. Now you’ll see the caution symbols
have gone away and the boundary of the simulation is right on the port planes.
So now we’re ready to simulate. I’m going to set up a simulation here. I’m going to
use FEM. Now FEM is slightly different from the finite difference time domain we talked
about before the Yee cells. This one uses tetrahedral meshing. It’s an adaptive mesh.
It’s automatic. But it’s kind of similar. Instead of using cubes, we’re using tetrahedrons.
So I’m going to use the iterative solver. The default is the direct solver. This machine
is a three-year old laptop. It’s only got 2 gigabytes of memory and the iterative solver
uses a lot less memory than the direct solver. And this particular structure doesn’t have
any convergence problems, so there’s no disadvantage of using the iterative one. If
you do get very complicated structures that don’t converge with iterative, then you
can use the direct solver. But you do then need a pretty powerful machine: a desktop
machine with enough memory for the size of structure you have. But iterative is fine
for this case. And I’ve also altered the delta error, which is basically the change
threshold in the s-parameters which kicks off a further iteration. This is like 2% magnitude
change in the s-parameters. This will be the criterion for the iterations to end. Once
I get smaller than 2% changes in my s-parameters, the iterations will stop because I’ve converged
to within that error on the Smith chart, the 2% error on the Smith chart. So, I think I’m
all set up here. I’ve got my frequency plan: start and stop frequency, number of steps
I want, adaptive sweeping,… I’m going to create and queue the simulation. Now this
takes about 30 minutes to run so after the first few seconds I’m going fast forward
and we’ll look at the end results because it’s not very interesting watching a simulator
run for thirty minutes! So I’ve fast forwarded and my simulation
is done. It took about 30 minutes, 38 minutes in fact. This is a three-year old laptop 2
gigahertz, two cores, 2 gigabytes of memory. Your mileage may vary, as they say! Let’s
look at the results. I pick my project name. There’s only one project open now. I want
the frequency results, the s-parameter results. I can plot these on a linear graph. See the
forward and return loss of the s-parameters there. We can do a Smith chart also. Once
you’re satisfied with the results of course you can export them. To a CITI file format.
I think I said Touchstone before? I meant CITI. And you can import them into any tool
including the Infiniisim software. Now, the s-parameters give you a good black
box model of the structure but sometimes you want to see inside the structure so you’ll
click on this “Advanced Visualization” button here. It will load the appropriate
project. So, we can orbit around the structure in 3D. We can select a given plane and then
check for connectivity. Very useful for finding little gaps that might be accidentally there.
It highlights the bond wires. I’ll do that again. The bond wires light up and so on.
You can also select a given plane. Look at the gridding there and animate the fields.
Let’s choose the lowest frequency, port one, mode one. Look at the E field. Select
the plane we want to visualize and we need to enable that plane. And we see the field’s
pretty much on sync across the whole structure here. I’ll just zoom out here. If we go
to highest frequency, however, you can see how the field pattern breaks up. Let me just
center this on the screen. You can see the field patterns are showing reflections and
attenuated behavior. So that’s the Advanced visualization. Now let’s go back to the
slides.