Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

S11 issue in a terminated microstrip line in CST simulation

Status
Not open for further replies.

nikosnte

Junior Member level 3
Joined
Oct 4, 2019
Messages
29
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Activity points
369
Hello everyone, recently i simulated a microstrip line in CST Microwave Studio with width 1.53564[mm] and the characteristic impedance value of this line with a substrate with epsr=3 at 30[GHz] is 12.2 ohms as it can be seen in the correspoding picture attached. Because i wanted to test the S11 behavior when the line is terminated in a load with the same impedance and when it's not, first i placed a load wich i selected a lumped element(first i picked the two necessary edge centers) RLC serial with R=12.2 ohms(C=0 & L=0) and later on, i placed the same load but with R=50 ohms and ran the simulation for both loads. As for the source, i placed edge discrete port at both simulations with reference impedance 12.2 ohms. The S11 parameter at the second simulation appears to have smallest S11 than the first simulation at many frequencies with a much bigger drop although the line is supposed to be mismatched because of the difference of port-load impedance. Why is this happening?

PS: the first S11 pic corresponds to the 12.2 ohms load and the second to the 50 ohms load and there are pics of the mlines too.
 

Attachments

  • Screenshot_5.jpg
    Screenshot_5.jpg
    55.5 KB · Views: 213
  • load12.2 (3).jpg
    load12.2 (3).jpg
    60.6 KB · Views: 215
  • load12.2 (2).jpg
    load12.2 (2).jpg
    62.8 KB · Views: 234
  • load12.2.jpg
    load12.2.jpg
    51.8 KB · Views: 173
  • load12.2 (4).jpg
    load12.2 (4).jpg
    78.7 KB · Views: 246
  • load50 (4).jpg
    load50 (4).jpg
    68.6 KB · Views: 184
  • load50 (3).jpg
    load50 (3).jpg
    68.1 KB · Views: 198
  • load50.jpg
    load50.jpg
    57.9 KB · Views: 189
  • load50 (2).jpg
    load50 (2).jpg
    64.1 KB · Views: 251

Hi,

that's not really a field of my expertise and it is long time ago I used CST. Funny thing is that the 50 Ω load matched simulation results in a lower reflection compared with the designed 12.2 Ω result @ 30 GHz.
If I understood you correct, for both simulations the load impedance is 12.2 Ω?!

According to the first attached figure you are including dispersion effects. How does the impedance, seen from the load looks like? Maybe you can show us an ordinary frequency response for both cases as well as a smith chart based result.

Greets

- - - Updated - - -

I just realized that you are using "," as well as "." as decimal indicator in your first attached figure. I have no clue if that matters in CST, but I already had troubles with that in other microwave programs.
 

According to the first attached figure you are including dispersion effects. How does the impedance, seen from the load looks like?

..seen from the source.

If I understood you correct, for both simulations the load impedance is 12.2 Ω?!

I meant the source impedance :-D.
 

The dispersion effects option at the impedance calculator only affects only a little the calculated Z0 (without dispersion effects Z0 is 12.19 ohms instead of 12.2). The impedance as seen from the source is a real load wich stays the same in the whole frequency band. For both simulations, the source impedance is 12.2 ohms. the smith chart plots are also attached(the first smith chart is for the 12.2 ohm load and the second for the 50ohm load). The "," decimal indicator is for the Z0 wich is produced by CST because it is the calculated variable so it doesn't affect anything. Because i've never used smith chart in cst before the values on the right part of the smith plot inside the parentheses what are they stand for?Screenshot_3.jpgScreenshot_4.jpg
 

Hi,

the smith chart shows you the impedance (reflection-coefficient/S11) frequency behaviour. You start at 27 GHz (white filled marker) and than turning counter clockwise (towards the generator/source) for increasing frequencies (decreasing wave length). For a matched 50 Ω system your should stay close to the center of the smith chart. For a matched 12.2 Ω system you also should reside close to certain point. This point is ~2/3 left from the center of the smith chart at the horizontal line at ~0.2. So obviously something went wrong.

The decimal indicator is also mixed for your gemetrical data. Which values have been entered by you? I think they should all use the same decimal indicator.

Greets
 
Last edited:

Is it possible that this happened because W/h>15?? Because if i want to use waveguide port(although i'm using discrete port) and calculate the port extension coefficient it produces an error wich says that w/h>15
 
Last edited:

I have no idea how CST implements this, but at these extreme widths you need to use a port and load that is connected across the entire line width. In your screenshot it looks like the source(?) load(?) is only connected at the center of the line, which would mean extra inductance from the discontinuity. In circuit simulation that is modelled as a zero length "step in width" element. So port width modelling details matter here!
 

Is it possible that this happened because W/h>15?? Because if i want to use waveguide port(although i'm using discrete port) and calculate the port extension coefficient it produces an error wich says that w/h>15

I dont know, but it seems there are some restrictions in CST, and I have no clue which one. It sounds to me your ratio is physically some kind unrealistic.
You have not told as If you have tried to unify the decimal indicator in your calculations, and if it results in different results.

Because i've never used smith chart in cst before the values on the right part of the smith plot inside the parentheses what are they stand for?View attachment 158912View attachment 158913

That may also be the referece impedance for the smith chart, if so, your result for the compolete matched 12.2 Ω system should reside in the center of the smith chart.

Greets
 

We get the same 12 Ohm for w=1.5mm h=0.1mm from other micorstrip calculculators, so that part is ok.

But see my comment in #8 on port width ... There is a strong discontinuity when such a very wide line is connected only in the line center (by lumped port). You need something like a wide resistive sheet (same width as line) for termination.
 

I dont know, but it seems there are some restrictions in CST, and I have no clue which one. It sounds to me your ratio is physically some kind unrealistic.
You have not told as If you have tried to unify the decimal indicator in your calculations, and if it results in different results.
The "," decimal indicator is located at the values of Z0 and epsilon effective wich are produced by the impedance calculator of CST. As for the input values of line width, substrate height and metal thickness i used "." decimal indicator so the produced values are fine.

That may also be the referece impedance for the smith chart, if so, your result for the compolete matched 12.2 Ω system should reside in the center of the smith chart.
Sorry at the beginning i meant to write the values at the left side of the smith chart not the right(made a typo)

- - - Updated - - -

We get the same 12 Ohm for w=1.5mm h=0.1mm from other micorstrip calculculators, so that part is ok.

But see my comment in #8 on port width ... There is a strong discontinuity when such a very wide line is connected only in the line center (by lumped port). You need something like a wide resistive sheet (same width as line) for termination.

Yes you are right i changed the discrete edge ports and placed discrete face ports at the line to see what happened but again i'm not sure at all if i get correct results.
I also tried the same line in Momentum ADS and first i simulated the line in the schematic wich produced resonable results because i placed two terminals and put Z0 impedance(12.2 ohms). And then i tried at the Momentum and placed edge pin ports at the line but the problem is that i get a warning from Momentum wich says that the port becomes electricaly large above 10.25 GHz and the S parameters may become unphysical so i cannot trust the results i'm getting(wich are different from the results i got from schematic). I cannot use a point port either. What should i do?
 

the values at the left side of the smith chart

Isn't that complex Zin at these two frequencies?

- - - Updated - - -

a warning from Momentum wich says that the port becomes electricaly large above 10.25 GHz and the S parameters may become unphysical so i cannot trust the results i'm getting(wich are different from the results i got from schematic).

Yes, the line must NOT be wide compared to the wavelength, otherwise you get funny resonances and multi-mode issues in both simultation and reality. So have a look at current density that you get from Momentum, if that looks physically correct.

Placing the pin on the edge of the line is fine, that is what you want to have here - if it matches your physical reality. See my appnote on that topic:
https://muehlhaus.com/support/ads-application-notes/edge-area-pins
 
Sorry at the beginning i meant to write the values at the left side of the smith chart not the right(made a typo)

The values in the parenthesis on the left are your complex impedances Z = R + j • X. The fist one (left) is the real valued part (resistance) and the second one (right) the complex one (reactance).
According to this values, the reference impedance of your smith chart should be 12.2 Ω (center corresponds to 12.2 Ω).
 
- - - Updated - - -



Yes, the line must NOT be wide compared to the wavelength, otherwise you get funny resonances and multi-mode issues in both simultation and reality. So have a look at current density that you get from Momentum, if that looks physically correct.

Placing the pin on the edge of the line is fine, that is what you want to have here - if it matches your physical reality. See my appnote on that topic:
https://muehlhaus.com/support/ads-application-notes/edge-area-pins
The screenshot of the current density of the line is attached, but it doesn't seem physicaly correct because from what i'm seeing there, is that there are some extremes at the edges and at the rest of the line the density is much lower, looks like a standing wave at the two edges. Also very weak signal appears to travel to port 2. It looks like more of a cavity than a transmission line. Also, besides of the warning about the electrically large port, i get two more warnings about the substrate and layout(screenshot is attached)


The values in the parenthesis on the left are your complex impedances Z = R + j • X. The fist one (left) is the real valued part (resistance) and the second one (right) the complex one (reactance).
According to this values, the reference impedance of your smith chart should be 12.2 Ω (center corresponds to 12.2 Ω).

So i simulated a 75 ohm line placing waveguide ports and from the results the left values are the Zin indeed at the these two frequencies.Momentum warnings.jpgcurrent density Momentum.jpg
 

that there are some extremes at the edges and at the rest of the line the density is much lower, looks like a standing wave at the two edges. Also very weak signal appears to travel to port 2. It looks like more of a cavity than a transmission line.

High edge current is exepcted and physically correct.
But I agree - it looks like no current goes to port 2. Does port 2 have a correct ground reference, so that currentcan flow?

Also, besides of the warning about the electrically large port, i get two more warnings about the substrate and layout(screenshot is attached)

That is information only, not a warning.
 

High edge current is exepcted and physically correct.
But I agree - it looks like no current goes to port 2. Does port 2 have a correct ground reference, so that currentcan flow?
Yes, the minus pin is assigned to GND just as port 1
Screenshot_5.jpg
 

The screenshot of the current density of the line is attached, but it doesn't seem physicaly correct because from what i'm seeing there, is that there are some extremes at the edges and at the rest of the line the density is much lower, looks like a standing wave at the two edges. Also very weak signal appears to travel to port 2. It looks like more of a cavity than a transmission line. Also, besides of the warning about the electrically large port, i get two more warnings about the substrate and layout(screenshot is attached)
View attachment 159008View attachment 159009

Hi, for which frequency was the simulation performed? 30 GHz for the 12.2 Ω line with a 12.2 Ω load and source?

According to your attached smith charts, the impedance seen from the source is higher than expected (@ 30 GHz), thus a missmatch and consequently multiple reflections and standing waves occuring along your microstrip.

Would it be possible to provide two 2D line plots, one showing the current along the microstrip line as well as one showing the voltage? As your transformed load appears to be larger than the source impedance (@ 30 GHz), I would expect a "high" voltage in combination with a low current at the load (smiliar to a open circuit condition).

greets

- - - Updated - - -

Yes, the minus pin is assigned to GND just as port 1
View attachment 159015

Does this port represent a physical connector like a SMA, BNC, N, etc. ? If so, is there an additional "lumped" load attached? I'm confused by "Ref Impedance", which seems to be the characteristic wave impedance of the connector, which in my opinion does not represent a physical connection between the "inner" conductor and ground by means of a 12.2 Ω load.
 

@stenzer
The port impedance settings are correct, this is how this is shown in ADS.
Voltage plot is not possible because Method of Moment solves for currents only, and voltage isn't available.

@nikosnte
I would like to see the S-parameters down to 0Hz, so that we can better check for simple mistakes in the setup.
 

@nikosnte
I would like to see the S-parameters down to 0Hz, so that we can better check for simple mistakes in the setup.
Screenshots of S11 and S22 from 0-33GHz and Smith charts of S11 and S22 at these frequencies are also attached (first two screenshots)

Hi, for which frequency was the simulation performed? 30 GHz for the 12.2 Ω line with a 12.2 Ω load and source?
The simulation was performed for the band 27-33 GHz. The central frequency is 30GHz. The line has characteristic impedance 12.2 Ohms and load impedance also 12.2ohms

According to your attached smith charts, the impedance seen from the source is higher than expected (@ 30 GHz), thus a missmatch and consequently multiple reflections and standing waves occuring along your microstrip.
Yes, weird thing that the input impedance at 33 GHz is greater than charecteristic impedance wich for a line with ZL=Z0 it is mathematically impossible.

Would it be possible to provide two 2D line plots, one showing the current along the microstrip line as well as one showing the voltage? As your transformed load appears to be larger than the source impedance (@ 30 GHz), I would expect a "high" voltage in combination with a low current at the load (smiliar to a open circuit condition).
I attached the current along the microstrip for the case at wich i used a discrete face port. As volker said, it makes sense that because of the wide line it is incorrect to place a lumped port as an excitation and i also placed a lumped port as the load so maybe there are some reflections also there. Next to the currents i attached the S11 parameter of the lumped port case. Something else interesting, is that i simulated the same line 1)with a substrate and ground plane with bigger width and 2) with a substrate and ground plane with the same dimension of the line and the results were much different. I also attached the screenshots for these two cases.First screenshot shows the current for the lumped port case and the rest of them is for the two other cases in that orders params 0-33GHz.jpgsmith charts 0-33GHz.jpgScreenshot_7.jpgScreenshot_8.jpgScreenshot_1.jpgScreenshot_2.jpgScreenshot_3.jpgScreenshot_4.jpgScreenshot_5.jpgScreenshot_6.jpg
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top