AnsweredAssumed Answered

S-Parameter Simulation Simulation using T-Line

Question asked by aguntuk on Dec 12, 2017
Latest reply on Dec 15, 2017 by aguntuk

Like • Show 0 Likes0
Comment • 8

Hi experts,

I am having problem with S-parameter simulation of 4 port for tranmission line. I have used "LTLINE2" component as my channel and I want to use this channel later for high-speed SERDES communication with differential TX/RX. But before that I need to know channel characteristics. May be after building a good channel here I can export the s-parameter file and use for other transceiver already built in Cadence. This is just for my basic simulation understanding. I have used controlled impedance line designer to put the line multilayer channel design inside the "LTLINE2" component. Now if I simulate the simulation result is not convincing, the S(2,1) or S(1,2) curves are linear not bumpy. What is wrong with my simulation setup or circuit or any other thing?

I have attached the screenshot of schematic and simulation in the attached file.

You help will be highly appreciated.

Attachments

Simulation.JPG55.6 KB
Controlled Impedance Line Designer.JPG123.7 KB
Multi-layer TL.JPG115.1 KB
Schematic.JPG82.6 KB

No one else has this question

Outcomes

Tim Wang Lee

Dec 12, 2017 2:40 PM

Hello,

This is Tim, author of Tim's blackboard.

For S21 to have ripples, the peak of S11 ripple needs to exceed -10 dB.

Because you optimized the coupled line to have 100 Ohm Zdiff, there is no mismatch and I expect S11 to be lower than -40 dB.

If you want to introduce ripples to S21, you need a pair of lines that's not well-matched.

Feel free to send me an email at tim.wang-lee@keysight.com

Tim

1 person found this helpful

Actions

volker_muehlhaus @ Tim Wang Lee on Dec 12, 2017 11:01 PM

Hi all

(amended): I was thinking that something is wrong because in schematic simulation, the line has more than 300dB loss. However, I didn't notice the extreme line length of 15m in schematic.

To see the small ripple from mismatch, this 0-400dB is the wrong scale. A shorter line (some wavelength) might be more useful for that test.

Best regards

Volker

1 person found this helpful

Actions

Tim Wang Lee

@ volker_muehlhaus on Dec 13, 2017 10:18 AM

I agree. Using the rule of thumb for S21, -0.1 dB/in/GHz, at 10 GHz, the S21 of a 15-in line should roughly be -15 dB.

However, the data display shows about -200 dB of loss at 5 GHz.

In other words, the loss is -40 dB/GHz, with -0.1 dB/in/GHz, the line length is around 400 in.

In the schematic, if you correct the 15000 mm to 15000 mil, you should be able to see S21 around -15dB.

1 person found this helpful

Actions

aguntuk @ Tim Wang Lee on Dec 14, 2017 9:35 AM

Hi @ Tim Wang Lee and @volker_muehlhaus,

Thanks a lot for your help. It worked after I changed the length (reduced to 15000 mil and less also). Now I plotted the s-parameter using graphs.

Now I wanted to save the s-parameter file as .s4p. But it says an error message which I attached as screenshot here. It says:

Port Z not the same for all ports. Please normalize data to a single impedance using the measurement equation 'write_snp()' before touchstone export

Here Z should not be same as expected. Why need to normalize to save .s4p file to save 4-port touchstone file and how to do it as they said using 'write_snp()'?

Best Regards,

Piash

Actions

volker_muehlhaus @ aguntuk on Dec 14, 2017 9:47 AM

Double check that all ports in schematic have the same value (50 Ohm). For using the S-params in the circuit, it doesn't matter what port impedance value you use for export, but it must the same value for all ports.

Using the data file tool for export is fine. Using write_snp() in the schematic is just another way to export the data.

1 person found this helpful

Actions

Tim Wang Lee

@ volker_muehlhaus on Dec 14, 2017 10:07 AM

Another good way is to use SPOutput in the schematic.

1 person found this helpful

Actions

aguntuk @ Tim Wang Lee on Dec 15, 2017 3:53 AM

Thank you for all the help

Actions

aguntuk @ volker_muehlhaus on Dec 15, 2017 3:53 AM

Thanks it worked...

Actions

8 Replies

Tim Wang Lee Dec 12, 2017 2:40 PM
Mark Correct
Correct Answer
Hello,
This is Tim, author of Tim's blackboard.
For S21 to have ripples, the peak of S11 ripple needs to exceed -10 dB.
Because you optimized the coupled line to have 100 Ohm Zdiff, there is no mismatch and I expect S11 to be lower than -40 dB.

If you want to introduce ripples to S21, you need a pair of lines that's not well-matched.

Feel free to send me an email at tim.wang-lee@keysight.com

Tim
1 person found this helpful
Like • Show 1 Like1
Actions
- volker_muehlhaus @ Tim Wang Lee on Dec 12, 2017 11:01 PM
  Mark Correct
  Correct Answer
  Hi all
  
  (amended): I was thinking that something is wrong because in schematic simulation, the line has more than 300dB loss. However, I didn't notice the extreme line length of 15m in schematic.
  
  To see the small ripple from mismatch, this 0-400dB is the wrong scale. A shorter line (some wavelength) might be more useful for that test.
  
  Best regards
  Volker
  1 person found this helpful
  Like • Show 1 Like1
  Actions
  - Tim Wang Lee @ volker_muehlhaus on Dec 13, 2017 10:18 AM
    Mark Correct
    Correct Answer
    I agree. Using the rule of thumb for S21, -0.1 dB/in/GHz, at 10 GHz, the S21 of a 15-in line should roughly be -15 dB.
    
    However, the data display shows about -200 dB of loss at 5 GHz.
    In other words, the loss is -40 dB/GHz, with -0.1 dB/in/GHz, the line length is around 400 in.
    
    In the schematic, if you correct the 15000 mm to 15000 mil, you should be able to see S21 around -15dB.
    1 person found this helpful
    Like • Show 1 Like1
    Actions
    - aguntuk @ Tim Wang Lee on Dec 14, 2017 9:35 AM
      Mark Correct
      Correct Answer
      Hi @ Tim Wang Lee and @volker_muehlhaus,
      Thanks a lot for your help. It worked after I changed the length (reduced to 15000 mil and less also). Now I plotted the s-parameter using graphs.
      
      Now I wanted to save the s-parameter file as .s4p. But it says an error message which I attached as screenshot here. It says:
      Port Z not the same for all ports. Please normalize data to a single impedance using the measurement equation 'write_snp()' before touchstone export
      Here Z should not be same as expected. Why need to normalize to save .s4p file to save 4-port touchstone file and how to do it as they said using 'write_snp()'?
      
      Best Regards,
      Piash
      Like • Show 0 Likes0
      Actions
      - volker_muehlhaus @ aguntuk on Dec 14, 2017 9:47 AM
        Mark Correct
        Correct Answer
        Double check that all ports in schematic have the same value (50 Ohm). For using the S-params in the circuit, it doesn't matter what port impedance value you use for export, but it must the same value for all ports.
        
        Using the data file tool for export is fine. Using write_snp() in the schematic is just another way to export the data.
        1 person found this helpful
        Like • Show 1 Like1
        Actions
        Tim Wang Lee @ volker_muehlhaus on Dec 14, 2017 10:07 AM
        Mark Correct
        Correct Answer
        Another good way is to use SPOutput in the schematic.
        1 person found this helpful
        Like • Show 1 Like1
        Actions
        aguntuk @ Tim Wang Lee on Dec 15, 2017 3:53 AM
        Mark Correct
        Correct Answer
        Thank you for all the help
        Like • Show 0 Likes0
        Actions
        aguntuk @ volker_muehlhaus on Dec 15, 2017 3:53 AM
        Mark Correct
        Correct Answer
        Thanks it worked...
        Like • Show 0 Likes0
        Actions

S-Parameter Simulation Simulation using T-Line

Attachments

Outcomes

Related Content

Recommended Content