Learning Forum

[AndyJP]

Why HFSS keeps its awful Smith plot without improvement for years?

1) It plots not what it takes as an input. It plots normalized impedance and S parameters, but takes reflection coefficient as an input!

That is insane! And there is no option to change the input, or the X-axis to match it.

2)It plots STRAIGHT lines between points! Madness! You either linearly interpolate the impedance, or the S-parameters, and the lines should bend smoothly.

3)There is no way to calculate and plot the input impedance. You have to build a crazy sandwitch of 2-screens long formulas for plotting input impedance of a multiport device. And you need to convert it to reflection coeff ? by the way. It can be so long that numetic error arizes and PC freezes for a few seconds when calculating this at user level.

4)Antenna design software CAN NOT plot VSWR cycles.

5) direction arrows are missed on short plots. Curve sweep points are not annotated - you can not figure out where 1 GHz, where 5GHz, and waht frequency the matching point is.

6) No tools for turning the plot, or tuning the impedance

5) No way to draw commenting impedance lines

6) After R15 vector export was removed, so publishing reports became problematic.

7)common to other data plots: the plot page can simply become blank, until the project is closed and reloaded.

 

- Most of that is implemented in Schematics, in the same bundle. But not in HFSS.

It continues for decades.

[Charlotte Blair]

Hi,

For decades HFSS shows its strength in providing you a fullwave 3D solution to Maxwell's equations as based on your model geometry. With this said we also assist our colleagues by providing tools that can assist them in other non-electrimagnetic solver domains such as circuit. Our Smith Tool is included in the HFSS Circuit solver. The solution it provides is highly dependent on how you setup you analysis. Directly to point 2 just add more points if you want to "see" a more curved trace. Of course we can add an interpolation to smooth the resulting curve. Also available as a post processing effort you can inport other data whether measured  or from some data table or from Field Calculator. You have an option not to normalize impedance.  Depending on your definition of input impedance (self or multual or port only etc etc ) we have Port Impedance as an already created variable in Results plot or again you can define a variable using Named Expressions. Again the definitions will be highly dependent on your model configuration hence there is not a "canned" definition. You can annotate curves with "Add Note" option in 3D Modeler window. To show arrows on curves just go ahead and check the box on the curve properties for "Show Arrows".. Tuning can be done using variables and optimetrics.. Many of your observations have been included in our latest release of AEDT please go ahead and be sure you are using the most current version and loot at our Usability enhancements. 

Please keep comments coming in so we know hat our users need.

 

Regards

 

 

 

 

 

 

[AndyJP]

The forum fails posting, when there are some specific sentences, or a number of letters reached:

Access Denied

You don't have permission to access "http://studentcommunity.ansys.com/topic/editposttopic/24e9b2ef-fb04-4f76-b5b2-abe6001d8659/" on this server.

Reference #18.7d2fc017.1593136061.11b9af0f

[AndyJP]

>For decades HFSS shows its strength in providing you a fullwave 3D solution
Exactly! For decades! Almost 3 decades ago, if I am not mistaken, solvers were designed, and around 2 decades ago the smith chart tool was designed.
And have almost not changed from that time.

>Our Smith Tool is included in the HFSS Circuit solver.
Exactly! And it is a superior tool! But in HFSS it is quite depreciated. It was still very popular when HFSS was exporting vector EMF plots. But now EMF is banned, and bitmap plots look... lets say cautiously, "not so good".

>The solution it provides is highly dependent on how you setup you analysis.
Exactly. And no matter how you setup it, you need irrational conversions for plotting it in the Smith chart.
It is just noniteligible. Like plotting a frequency response vs. time axis... Then, a conversion takes time(when there are a many thousands points), and introduce some numeric error (like ?=1.0000001, which clicks next scale in the chart, which is very annoying).
Really, HFSS is just freezing for a few minutes when preparing such plots.

 

[AndyJP]

>Directly to point 2 just add more points

Let's take a case with a point simulating 6 days. Really, just 2-3 more points?
At the moment I am optimizing multiport input impedances with 6 parameters per port. It takes a day plotting series with only 6 points per each curve.
Why should I spend 3 days for plotting 18 points per curve, when I need just a correct linear interpolation of impedance?

>Of course we can add an interpolation to smooth the resulting curve.

It would be appreciated by thousands of engineers and researchers. Really. The interpolation is that simple: "a*X+b".
You already have it implemented. But applied it to wrong values.
It should be applied to a dataset, or to polar coordinates at least. Applying it to X-Y coordinates of a polar chart is wrong.

>Also available as a post processing effort you can inport other data whether measured  or from some data table or from Field Calculator.
It is really a cool feature. But it does not solve the problem of "sawtooth" Smith plot. Interpolating the data in external software is not a productive way.

>You have an option not to normalize impedance.
Are we speaking about Smith chart?

>Depending on your definition of input impedance
Smith chart is not limited to "Input" impedance. But regardless of definition, you can not "just plot" an impedance in HFSS.

Speaking about input impedance... at least could you try implementing just something? HFSS have already several definitions for port impedance, and it is just great!
Implementing U/I impedance in modal report, isn't it logical? It is the same as Z11-Z12*Z21/(Z22+Zo), where Zii may be high rank matrices, depending on number of ports. Any other definition is fine: more tools we get, better HFSS becomes.

[AndyJP]

> have Port Impedance as an already created variable in Results plot
I got confused. Are you really familiar with HFSS part of EDT? The port impedance is a characteristic impedance of the input infinite transmission line, which represents the port. It is not meant to be plot, but is should be a normalizing impedance for Smith chart and other modal-report parameters.
There are really several definitions, which is good. But it does not change the situation.

> you can define a variable using Named Expressions
Named expressions do not allow dependent expressions, like Zin(i), where "i" is a port id.  You can create only Zin1, then Zin2, then Zin3, then Zi4....

That is what I am talking about. It overloads the engine with useless high-level calculations, results in freezing. It is a source of errors due to typos and numeric rounding. It can not be ported easily to other projects, or shared between users.

>. Again the definitions will be highly dependent on your model configuration
No. It depends only on physics. If there are several definitions, all of them should be implemented. There are not so many. For port, it was already implemented, and everyone are happy with that!

>hence there is not a "canned" definition.
OK. You can remove the definition of port impedance, following this logic. Just remove it and let people calculate the impedance themselves on paper.

 

>You can annotate curves with "Add Note" option in 3D Modeler window.
No, I can't. Did you try yourself? It jumps unpredictably whenever you change the scale. It does not follow the plot.
Why did you pulled that out? It is another problem which I did not mention.

I was talking about curve POINTS. Even when annotating manually, you can not see which is which in HFSS. It just does not show you!

>to show arrows on curves just go ahead and check the box on the curve properties for "Show Arrows".

AHA! IT DOES NOT WORK!
When a curve is short, it does not work. You need at leas a half-turn around the chart for the arrow to appear.

>Many of your observations have been included in our latest release of AEDT
I have 2020R1. Isn't it the latest?

really, HFSS did not change much from ver.13. I prefer ver 13, because UI was faster there, and plots were exported in vector EMF.